CHANGE IN ATTITUDE AND CHANGE IN RISK PERCEPTION OF GENETICALLY MODIFIED FOOD By TAYLOR KATHRYNE RUTH A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2015
Â© 2015 Taylor K . Ruth
To the memory of my Mother, Jean Taylor Ruth
4 ACKNOWLEDGMENTS I am grateful for the love and support I have re ceived from my friends, family , and the Department of Agricultural Education and Communication (AEC) faculty during my time in the m have achieved over the past year and half without the guidance of my mentor and advisor, Dr. Joy Rumble. Dr. Rumble provide d me with numerous opportunit ies, which have helped me develop as both a researcher and a professional. I learned more from her than I ever expected, and I realize how lucky I am to have been paired with such a caring advisor. I would also like to extend a thank you to my committee me mber, Dr. Alexa Lamm. Dr. Lamm provided me with a tremendous understanding of the importance of survey design and left me with knowledge I will use throughout my career. She pushed me to think critically, and with the help of Dr. Rumble, helped me develop a thesis I am proud to have written. I also need to thank Dr. Ed Osborne, not only for his insightful feedback as I wrote my proposal and for his part in influencing my writing style, but also for his co ntinuous support throughout my m appreciative of the financial support he gave me along with the many learning opportunities. Finally, I would like to thank Mrs. Becky Raulerson for introducing me to the field of agricultural communications during my senior year at the University of Flor ida. Her class was where I lear and where I first became interested in public opinion research. I am grateful that Dr. Rumble, Dr. Lamm, Dr. Osborne, Mrs. Raulerson, and the AEC graduate committee gave me , a student with a degree in microbiology, the chance to succeed in a completely different field . I could not have completed this thesis without the support of my wonderful friends. Ashley Leonard and Katie Shepherd were always there to give me a break fro m
5 my work; they provided me with endless laughs and so much love. Caroline Roper served as my mentor and friend through this process, and I am thankful she pushed me to apply for this degree. I would also like to thank my friends within the AEC department, Rachel Manning, Tahlia Pollitt, Courtney Owens, Chris Mott, Makenna Lange, Shuyang Qu, Pei Wen Huang, Milton Newberry, Seth Heinert, Keegan Gay, and Tre Easterly , for helping me in classes, grabbing coffee when energy was low, and always making me smile. I also have to thank Arthur Leal. He has been such a wonderful source of suppo rt and joy during my time as a m I have many fond memories of the two of us, and look forward to more in the future. Finally, I would like to thank my family. My father, David Ruth, has always been there to support me regardle ss of what my goals were. He has no agricultural background, but did not question my choice for a moment when I said I wanted to pursue a degree in agricultural communications. I am grateful f or him and my family, Carla, Jacqueline, and Jenni fer, for being there to support me through out this endeavor. In addition to my family, my dressage trainer, Maggie Selbert, has been a tremendous source of support. She cared for my horse, Millie , while I was in graduate school , and made sure I had one less thing to worry about . My time at the barn was a welcomed break from school , and I am appreciative that Millie was treated so well while I was not there. My thesis is now done, and I will be pursuing Ph.D . in the fall of 2015. Knowing I have this support system of family, friends, and mentors relieves any anxiety I have toward the doctoral program; I am lucky to have these people in my life.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURE S ................................ ................................ ................................ ........ 10 LIST OF DEFINITIONS ................................ ................................ ................................ . 11 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 Histor y of Genetically Engineered Food ................................ ................................ .. 14 Regulation of Genetically Engineered Food ................................ ..................... 15 Advantages of Genetically Engineered Food ................................ ................... 1 6 Production advantages ................................ ................................ .............. 16 Human health advantages ................................ ................................ ......... 18 Disadvantages of Genetically Engineered Food ................................ ............... 19 Environmental risks ................................ ................................ .................... 20 Human health risks ................................ ................................ .................... 21 Consumer A ttitudes toward Genetically Engineered Food ................................ ...... 21 Opinions toward Agricultural Biotech Research Companies ............................ 23 Genetically Engineered Food in the Media ................................ ....................... 25 Agricultural Communications ................................ ................................ .................. 27 Communicating about Genetically Engineered Food ................................ .............. 28 Research Problem ................................ ................................ ................................ .. 29 Purpose & Objectives ................................ ................................ ............................. 29 Significance ................................ ................................ ................................ ............ 30 Summary ................................ ................................ ................................ ................ 31 2 RELEVANT LITERATUR E ................................ ................................ ...................... 33 Shannon and Weaver Communication Model ................................ ......................... 33 Attitudes ................................ ................................ ................................ .................. 34 Persuasive Communication ................................ ................................ .................... 36 Elab oration Likelihood Model ................................ ................................ .................. 39 Elaboration ................................ ................................ ................................ ....... 40 Central Processing Route ................................ ................................ ................. 41 Peripheral Processing Route ................................ ................................ ............ 43 Prior Knowledge ................................ ................................ ............................... 45 Source Cues ................................ ................................ ................................ ..... 45 Attitudes toward Genetically Engineered Food ................................ ....................... 46 Demographics and Attitude ................................ ................................ .............. 48
7 Knowledge and Attitudes ................................ ................................ .................. 50 Source Credibility Research in the ELM ................................ ................................ . 51 ELM in Agricultural Research ................................ ................................ ................. 52 ELM Research with Genetically Engineered Food ................................ .................. 55 Risk Perception and Genetically Engineered Food ................................ .......... 56 Source Credibility and Genetically Engineered Food ................................ ....... 57 Conceptual Model ................................ ................................ ................................ ... 59 Summary ................................ ................................ ................................ ................ 60 3 METHODOLOGY ................................ ................................ ................................ ... 66 Experimental Design ................................ ................................ ............................... 66 Populati on and Sample Size ................................ ................................ ................... 69 Data Collection ................................ ................................ ................................ ....... 71 Validity and Reliability of the Instrument ................................ ................................ . 72 Threats to Validity in an Experimental Design ................................ .................. 73 Internal validity ................................ ................................ ........................... 73 Construct validity ................................ ................................ ........................ 74 External validity ................................ ................................ .......................... 75 Statistical conclusion validity ................................ ................................ ...... 76 Survey Error ................................ ................................ ................................ ..... 76 Instrumentation ................................ ................................ ................................ ....... 78 Demographics ................................ ................................ ................................ .. 79 Prior Knowledge of Genetically Modified food ................................ .................. 79 Source Credibility ................................ ................................ ............................. 80 Attitudes toward Genetically Modified Food ................................ ..................... 81 Risk Perceptions of Genetically Modified Food ................................ ................ 81 Analysis ................................ ................................ ................................ .................. 82 Limitations ................................ ................................ ................................ ............... 86 Assumptions ................................ ................................ ................................ ........... 87 Summary ................................ ................................ ................................ ................ 88 4 RESULTS ................................ ................................ ................................ ............. 104 Analysis of Demographics ................................ ................................ .................... 104 Analysis of Variables of Interest ................................ ................................ ............ 105 Prior Knowledge ................................ ................................ ............................. 105 Source Credibility ................................ ................................ ........................... 106 Change in Attitude ................................ ................................ .......................... 106 Change in Risk Perception ................................ ................................ ............. 107 An alysis of Objectives ................................ ................................ ........................... 108 Post Hoc Analysis ................................ ................................ ................................ . 111 Change in Risk Perception ................................ ................................ ............. 111 Final Attitude ................................ ................................ ................................ .. 112 5 CONCLUSIONS ................................ ................................ ................................ ... 126
8 Overview ................................ ................................ ................................ ............... 126 Key Findings ................................ ................................ ................................ ......... 126 Implications ................................ ................................ ................................ ........... 129 Theoretical Implications ................................ ................................ .................. 129 Practical Implications ................................ ................................ ...................... 132 Limitations ................................ ................................ ................................ ............. 135 Recommendations ................................ ................................ ................................ 137 Future Research ................................ ................................ ............................. 137 Industry and Practitioners ................................ ................................ ............... 141 Post Hoc Key Findings ................................ ................................ ......................... 143 Post Hoc Implications ................................ ................................ ........................... 144 Theoretical Implications ................................ ................................ .................. 144 Practical Implications ................................ ................................ ...................... 145 Post Hoc Recommendations ................................ ................................ ................ 146 Future Research ................................ ................................ ............................. 146 Industry and Practitioners ................................ ................................ ............... 147 Summary ................................ ................................ ................................ .............. 148 APPENDIX A IRB APPROVAL ................................ ................................ ................................ ... 150 B INSTRUMENT USED FOR STUDY ................................ ................................ ...... 154 C EXPERIMENTAL TREATMENT ................................ ................................ ........... 158 D COMPLETE SURVEY INSTRUMENT ................................ ................................ .. 160 REFERENCES ................................ ................................ ................................ ............ 188 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 204
9 LIST OF TABLES Table page 3 1 Experimental design for the study. ................................ ................................ ..... 90 3 2 Proportional weights of demographics in the sample. ................................ ........ 91 3 3 Normality assumptions of variables. ................................ ................................ ... 92 3 4 VIF and Tolerance for variables in objective three and four. .............................. 93 4 1 Demographic characteristics of the respondents . ................................ ............. 114 4 2 Description of source credibility. ................................ ................................ ....... 116 4 3 ANOVA for source credibility between groups. ................................ ................. 116 4 4 Description of attitudes toward genetically modified food. ................................ 116 4 5 Paired sample t test between prior a nd final attitude. ................................ ....... 116 4 6 Description of risk perception of genetically modified food. .............................. 117 4 7 Paired sample t test between prior and final risk perception. ........................... 117 4 8 ANOVA for change in attitude. ................................ ................................ .......... 117 4 9 Follow up test for change in attitude. ................................ ................................ 117 4 10 ANOVA for change in risk perception. ................................ .............................. 118 4 11 Multiple linear regression analysis for variables predicting change in attitude (Model 1 and 2). ................................ ................................ ............................... 11 9 4 12 Multiple linear regression analysis for variables predicting change in attitude (Model 3 and 4). ................................ ................................ ............................... 121 4 13 Multiple linear regression analysis for variables predicting change in risk perception. ................................ ................................ ................................ ........ 122 4 14 Post hoc analysis for change in risk perception. ................................ ............... 123 4 15 Post hoc analysis for final attitude. ................................ ................................ ... 124
10 LIST OF FIGURES Figure page 2 1 ................................ .. 63 2 3 Conceptual model of the affect of pe change in attitude and risk perception of genetically modified food.. .................. 65 3 1 Normality curve for prior knowledge prior to removal of outliers. ........................ 96 3 2 Normality curve for prior knowledge after removal of outliers. ............................ 97 3 3 Normality curve for source credibility prior to removal of outliers. ....................... 98 3 4 Normality curve for source credibility after removal of outliers. .......................... 99 3 5 Normality curve for change in attitude prior to removal of outliers. ................... 100 3 6 Normality curve for change in attitude after removal of outliers. ....................... 101 3 7 Normality curve for change in risk perception prior to removal of outliers. ....... 102 3 8 Normality curve for change in risk perception after removal of outliers. ........... 103
11 LIST OF DEFINITIONS Attitudes Attitudes are a learned and implicit process which can vary in intensity, as well as direction, and mediate evaluative behavior ( Osgood , Suci, & Tannenbaum, 1971). For this study, attitude was measured using six items on a bipolar semantic differential scale. Elaboration Likelihood Model (ELM) ELM describes the process through which people interpret persuasive communication (Petty & Caci oppo, 1986). There are two processing routes: central and peripheral. The central processing route requires the individual to use careful consideration to analyze the message , and attitude change is usually predictive of behaviors. The peripheral processin g route uses less consideration of the message and relies on peripheral cues, like sources, to form opinions (Petty & Cacioppo, 1986). Genetically Engineered Plants which have had their genes altered to produce favorable characteristics, such as growth and nutritional characteristics are called genetically engineered. The Food and Drug Administration (FDA) considers this a more precise term than genetic modification (FDA, 2014). Genetically Modified Food Genetic ally modified food was defined in this study as the desired trait. Genetically Modified Organism (GMO) According to the World Health Organization (WHO, 2009), genetically modified organisms are d genetic material (DNA) has been altered in a way that does not occur naturally. It allows selected individual genes to be transferred from one organism into another, also between non (p.104). Persuasion symbolic process in which communicators try to convince other people to change their attitudes or behaviors regarding an issue through the transmission of a message in an atmosphere of free Shannon and Weaver Model of Commun ication This model explains the linear communication that occurs between an information source and recipient. However, unwanted signals called noise can distort the intended message before the recipient interprets it (Lee & Baldwin, 2004; Shannon & Weaver, 1949).
12 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science CHANGE IN ATTITUDE AND CHANGE IN RISK PERCEPTION OF GENETICALLY MODIFIED FOOD By Taylor Kathryne Ruth August 2015 Chair: Joy N. Rumble Major: Agricultural Education and Communication Science has shown the safety and benefits of using genetically modifi ed food. Consumers have been historically misinformed and uniformed about genetically modified food , which has led to skepticism related to both the products and the producers. A greater understanding for how consumers form perceptions of genetically modified food is essential since public acceptance is related to the success of a product. The purpose of this study was to examine the influence of persuasive communication on consumers change in attitude and change in risk perception of genetically modi fied food. Using a conceptual model based on the ELM and Shann on ommunication model, an experimental design test ed how different message sources (Industry and Government) influenced changes in attitude and risk perception. An online survey wa s administered to Florida residents using non probability sampling and post s ( n = 515 ). Results from this study suggest ed that the message source was associated with change in attitude, but not chan ge in risk perception. Additionally, source credibility influenced attitude change while prior knowledge did not. Risk perception did not appear
13 to operate within the EL M or proposed conceptual model. The findings suggest ed that changes in attitude and ris k perceptions are processed differently from one another, and further research is needed to explore these differences. Recommendations for practitioners include using m ore credible sources when communicating about genetically modified food and using value driven messages because the y may elicit more favorable attitudes than simply stating facts about the technology.
14 CHAPTER 1 INTRODUCTION The purpose of this research was to analyze the influence of persuasive change in attitude and change risk perception of genetically modified food . The study was specifically interested in the effect different change in attitude and change in risk perception of genetically modified food . Chapter 1 described the history and science opinions of the technology, and the importance of proper communication. History of Genetically Engineered Food Humans have been altering the genes in plants for centuries using the science of Charles Darwin and Gregor Mendel, along with laws of hereditability, to select for specific traits ( Henig , 2000). These alterations have given the world modern strawberries, wheat, and co rn, unrecognizable today compared to their ancestors ( Chassy , 2007). Through selective breeding, the productivity of corn has grown from 10 bushels per acre to 200 bushels per acres in only 125 years ( University of Illinois Extension, 2001 ). Watson and Cri ck discovered the double helix model of DNA in 1954, which led to the realization that each gene encoded for a unique protein and was related to the phenotypes that were expressed. These breakthroughs led to further genome research and the discovery that r ibosomal DNA (rDNA) could be inserted into another living organism ( Chassy , 2007). This process became known as genetic engineering, and in 1988 genes were successfully inserted into soybean plants, which allowed 70% of the soybeans grown around the world to have been genetically engineered ( James , 2007).
15 Genetically engineered crops have been planted on over one billion acres worldwide (James, 2007). Over 10 million farmers, including eight million in developing countries, have chosen to purchase the more expensive genetically engineered seeds to produce a higher yield, while using less chemicals, labor, and resources (Chassy, 2007; James, 2007). In the United States (U.S.), around 80% of processed food has been estimated to contain genetically engineered f ood ingredients (Hallman, Hebden, Aquino, Cutie, & Lang, 2003). Eight transgenic crops have been sold commercially as a whole product: corn, soybeans, cotton, canola, alfalfa, sugar beets , papaya, and squash (James, 2007). These crops have been engineered to reduce yield loss due to pests, drought, and disease (GMO answers, 2014). Genetically engineered plants have been grown in over 27 countries , and insecticide use in corn has decreased from 0 .21 pounds per acre in 1995 to 0 .02 pounds per acre in 2010 ( Fe rnandez Cornejo, Wechsler, Livingston, & Mitchell, 2014; GMO answers, 2014). Regulation of Genetically Engineered Food The United States National Academy of Science (NAS) was asked by the White House to investigate potential threats related to genetically engineered crops (NAS, 1987). NAS concluded that the genetically engineered plants posed little environmental, agricultural, or consumer risk (NAS, 1987). The National Research Council (NRC) followed by stating that engineered plants were just as safe, if not safer, than traditional crops (NRC, 1989). Despite agreement in the scientific community that genetically engineered crops were safe, government regulators were asked to develop separate regulations for engineered plants, due to consumer concerns (Chas sy, 2007). The Food and Drug Administration (FDA) created protocol to regulate the product itself and not the process (FDA, 2014). Using this protocol, the developer s of the crop
16 identified distinguishing attributes of the product and presented allergens and nutrients hnology Evaluation Team then ass essed the products to ensure they were within limits of the law (FDA, 2014). The FDA had conducted 95 consultations with seed developers by 2012, 30 of which were with corn (FDA, 2014). As part of the regulations for genet ically engineered plants, t he U.S. Environmental Protection Agency (EPA) evaluated the environmental impacts associated with genetically engineered, pest resistant crops. Additionally, the United States Departme nt of Agriculture (USDA) division of Animal and Plant Health Inspection Service (APHIS) has ensured all field testing of genetically engineered crops were done in a controlled manner to lesson environmental consequences (Lemaux, 2008). Advantages of Geneti cally Engineered Food Approximately 800 million people do not have enough food to lead healthy lives, and poor nutrition leads to the death of three million children each year ( Black et al., 2013; Food and Agricultural Organization, 2015 lation has been projected to increase from 7.3 billion people to 12.3 billion by 2100, and the available land for agriculture has been predicted to decrease (Chassy, 2003; Gerland et al., 2014). The discovery of genetically engineered plants , and their abi lity to produce a higher yield using fewer resources , could be a promising solution to this dilemma (Chassy, 2007; Phillips, 2008). Production advantages Increased crop yield, reduced cost for food and drug production, reduced pesticide use, and enriched nutrient content have been typical advantages to growing genetically engineered plants (Phillips, 2008). The previously mentioned glyphosate resistant soybeans have allowed farmers to spray the ir fields with Roundup, a type of
17 herbicide, killing the weeds yet leaving the plants healthy (Phillips, 2008). A similar and common product is Bt corn. The insecticidal gene protein from the bacterium Bacillus thuringiensis was inserted into the maize genome, allowing the plant to be resistant to the European corn bo rer (Phillips, 2008). Pesticide quantity has been reduced by 37% and cost has decreased by 39%. Even though the genetically engineered seeds are more expensive, the average farmer still gains approximately 69% profit due to a reduction in pest management ( Klumper & Qaim, 2014). Benefits of genetically engineered crops have not been limited to just pest resistance. The papaya industry of Hawaii was saved through genetic research. Papaya ring spot virus (PRSV) was discovered in the 1940s on Oahu and eliminate d most of the crop in just 10 years (Gonsalves, Ferriera, Manshardt, Fithc, & Slightom, 2000). The industry relocated to the island of Hawaii in the 1960s, which allowed scientists to find a solution before the virus was able to cross over the water. Tradi tional treatments, like exposing the fruit to a milder strain of the virus, have proven ineffective when met with an aggressive form of PRSV (Gonsalves et al., 2000). However, a solution was found with the introduction of genetically engineered papaya cont aining a viral coat protein gene (Gonsalves, 1998). By 2006, over half of the papayas grown on Hawaii were genetically engineered (Lemaux, 2008). Similarly, the citrus disease huanglong b ing, or citrus greening, has had devastating effects on the citrus ind ustry globally and in Florida (Satran, 2014). The government and citrus industry have already invested more than $220 million dollars into finding a cure (Putnam, 2012), but genetically engineering the fruit has appeared to be the most promising solution t o save the industry (Bove, 2012).
18 The advantages of genetically engineered crops expand beyond physical tens of thousands genes involved in the alterations (Chetelat , Deverna, & Bennet, 1995). Since selective breeding has not been specific for a gene, scientists have had no way of knowing what secondary effects may occur. Additionally, selective breeding can only be done between closely related species or genera (Chet elat et al., 1995). Genetic engineering has allowed for precise gene control, meaning that only the specific gene a scientist is interested in will be altered (Cho, Kim, Choi, Buchanan, & Lemaux, 2000). These genes can be linked to specific regulatory sign als, making expression only occur in certain parts of the seed (Cho et al., 2000). Genetically engineered plants have not been limited to similar species, and genes can be inserted from other plants, animals, and bacteria (Lemaux, 2008). This has led to DN A combinations never seen before. Human health advantages Golden rice was another genetic engineering breakthrough for the agricultural industry and has offered a separate set of benefits. In developing countries, approximately 500,000 children will go bli nd each year as a result of Vitamin A deficiency, and up to half of those children will die within a year ( World Health Organization , 2015 ). Despite efforts , such as providing Vitamin A pills, fortifying sugar with Vitamin A, and various gardening projects , this issue has still prevailed (Lemaux, 2008). These proposed solutions have been costly and have required continuous public education, making it difficult for developing countries to use these practices (Lemaux, 2008). Golden rice was developed as a gen etically engineered variety of rice with increased beta carotene levels from both daffodil and maize genes ( Paine et al., 2005). Beta carotene is the molecular precursor to Vitamin A, which when eaten will likely be
19 converted to the vitamin after digestion (Paine et al., 2005). There are currently Golden Rice breeding programs in India, China, Bangladesh, Philippines, and Vietnam (Paine et al., 2005). This genetically engineered rice may not be the only solution to Vitamin A deficiency, but it has been seen as a step toward combating the crisis (Lemaux, 2008). A number of genetically engineered plants have been developed but not yet released to the public. Genetically engineered, heart healthy oils will likely enter the market soon. These plant derived oils have been thought to offer low trans fat, high mono saturated fat , and omega 3 fatty acids to help increase heart health (Takeda & Matsuka, 2008). Scientists have also engineered a variety of maize, which can express an immune response in the kernels equiv alent to a vaccine, which could eliminate the need for injections and increase mass immunization (Takeda & Matsuka, 2008). The production of blight resistant potatoes has been yet another innovation which could have eliminated the great potato famine in Ir eland (Takeda & Matsuka, 2008). Disadvantages of Genetically Engineered Food Even though genetically engineered food has been consumed and produced worldwide, and a recent meta analysis found no issues related to its safety (Nicolia, Manzo, Veronesi, & Ros ellini, 2014), the technology has been surrounded by debate and skepticism ( Senauer , 2013). An anti genetically engineered food website has merges DNA to create unstable combi nations of plant, animal, bacterial, and viral the lack of long term studies examining the possible effects of genetically engineered food ( Kantor , 2013). Some people suggest genetically engineered crops could cause unknown effects on the environment and people ( Nelson , 2001 ).
20 Environmental risks Potential risks to the environment have been a major concern related to the use of genetically engineered crops (Nelson, 2001 ). Hor izontal gene transfer could occur between organisms and promote pesticide and herbicide resistance in plants (Phillips, 2008). If this genomic exchange were to occur between genetically engineered plants and surrounding weeds, the weeds would become herbic ide resistant and begin to grow uncontrollably (Ma, Drake, & Christou, 2003). The development of these superweeds has been a major concern and could lead to ecological imbalances due to their resistance to herbicide applications (Ma et al., 2003; Philips, 2008). Superweeds were essentially unknown before the introduction of genetically engineered crops (Benbrook, 2012), but by 2015, weeds had become resistant to 22 out of 25 herbicide action cites as identified by the Weed Sci ence Society of America (WSSA, 2015). Even though pesticide use decreased in the U.S. during the first six years of commercialized genetically engineered crop use, by 2012, pesticide use had actually increased by seven percent (404 million pounds) as a result of the emergence of superwe eds (Benbrook, 2012). Genetically engineered plants may also be harmful to beneficial insects. One study showed that when exposed to Bt corn pollen, the mortality rate of monarch butterfly larva e significantly increased (Losey, Raynor, & Carter, 1999). Wh ile this study showed the pollen was harmful, the actual threat level was debated among scientists (Phillips, 2008). The concentration of pollen used in the original study was extremely high, and the migratory patterns of the butterflies did not put them i n the area during the transgenic pollen shed (Sears et al., 2001). The threat against the monarch butterfly was later determined to be relatively low (Sears et al., 2001).
21 Human health risks Members of the public and some in the scientific community have concluded that genetically engineered crops have been related to allergies, irritable bowels, organ damage, and cancer (Phillips, 2008). Until recently, v ery little literature has been availab le regarding the safety of genetically engineered food, and most of the health safety research has been conducted by the private companies who developed the seeds (Dona & Arvanitoyannis, 2009). Many people have been concerned that genetically engineered cr ops could be connected to the rise in allergies over the past decade (Philips, 2008). In 2000 a strain of Bt corn, Starlink, was recalled, and the Center for Disease Control (CDC) was asked to investigate the 51 people who fell ill after consuming food wit h Starlink as an ingredient ( CDC , 2001). Over half the people expressed symptoms consistent with an allergic reaction, but none of the patients had Starlink specific antibodies in their serum (CDC, 2001). This indicated that the corn might not have caused the allergy, although some allergic reactions can occur without the presence of the specific antibody (CDC, 2001). Another concern amongst the public has been the possibility of antibiotic resistant genes being transferred to humans ( Dona & Arvanitoyannis, 2009 ). These concerns have not been solidified in research, and an overview of literature pertaining to the safety of genetically engineered food from the past 10 years did not identify any significant health issues related to the products (Nicolia et al. , 2014). Consumer A ttitudes toward Genetically Engineered Food Consumers have tended to believe that genetically engineered foods are not regulated by the government and are not as nutritious as organic options (Chassy, 2007). While numerous peer reviewed studies have shown that genetically engineered
22 food have no significant differences in nutritional value, when compared to conventional crops, the public has still felt that genetically engineered food are not as healthy (Lemaux, 2008). Even with scientif ically supported advantages , a large sense of risk has been associated with these new technologies from consumers. Over half of Americans have believed that genetically engineered food are unsafe to eat compared to only 11% of scientists ( Funk et al., 2015 has been a result of having limited information about the products (Carlson, Frykblom, & Lagerkvist, 2007; Lidskog, 1996). This risk perception has led people to purchase produce free of genetic engineering at a premium price, up to 43% higher than genetically engineered food (Lusk, Jamal, Kurlander, Roucan, & Taulman, 2005). Studies have shown that a little over half of the American population believes genetically engineered foods are not safe to eat (Langer, 2013). An overwhelming majority has agreed that the federal government should require foods containing genetically engineered food to be labeled, and more than half of the public has said they would not purchase food that was labeled as genetically engineered ( Lange r , 2013; Pounds, 2014). engineered food (Langer , 2013). Women were significantly less likely to purchase food that had been genetically engineered, compared to men ( Napier , Tucker, Hen ry, & Whaley, 2003; Pounds, 2014 ). Langer (2013) reported that people under the age of 45 were more likely to call genetically engineered food safe than those over 45 years old, and only a small portion of young adults thought genetically engineered fo od were
23 unsafe. People may also not purchase genetically engineered food because of their a. 6). Some religions have chosen to abstain from introducing new material into their food (Phillips, 2008). The higher the level of perceived risks consumers associate d with genetically engineered food, the less likely they were to purchase genetically engineered food (Napier et al., 2003). Common risks consumers associated with genetically engineered food included human health complications, harm to wildlife, and loss o f agricultural productivity (Napier et al., 2003). The top reasons consumers felt genetically engineered food posed a risk were the creation of pesticide resistant weeds or insects and the threat posed to beneficial insects (Napier et al., 2003). A study i n Florida showed that consumers believed that genetically engineered food presented a greater risk of food allergies or poisoning and were unsure of the possible advantages of genetically engineered food (Rumble & Leal, 2013 ). The majority of the public di d not believe it had consumed genetically engineered food and agreed that the quality of the products had decreased over recent years (Rumble & Leal, 2013). Opinions toward Agricultural Biotech Research Companies Consumer s skepticism surrounding genetica lly engineered food has not been confined to just the product. The main sources of information regarding genetically engineered food have been institutions directly i nvolved with the products, and the public has not always viewed the information as unbiase d ( Huffman , Roussu, Shogren, & Tegene , 2004 ). The companies selling and researching the technology have been under scrutiny from the public and by the media ( Caffrey , 2014; Chaussee , 2014;
24 Nichols , 2014). Dr. Kevin Folta (2012), a plant molecular biology p rofessor at the University of Florida, said the public has trouble separating its feeling s toward these companies with its feelings toward the science. Two major businesses that have been in the spotlight include Green Giant and AgLabs (these are pseudonym s for real companies which will be used throughout this thesis). Green Giant is a large agricultural biotechnology company that is known for its development of herbicide resistant corn and other herbicide resistant crops ( Green Giant, 2014 ). AgLabs is a si milar business involved in genetic engineering research and recently developed a variety of drought resistant corn ( Karole, 2014 ). These are also the two leading companies in field releases for testing genetically engineered crops. As of September 2013, Gr een Giant had 6,782 authorized field releases and AgLabs had 1,405 (Fernandez Cornejo et al. , 2014). Green Giant has fallen under a lot of criticism over the past decade and has often March agai nst Green Giant, 2014) . While there have been a number of companies researching and selling genetically engineered seeds, Green Giant has become the most active. From 2000 to 201 5 , around 1000 articles were published in T he New York Times focusing on t h e c ompany ( The New York Times, 2015c ). Most media coverage remarked on the seed protected seeds on their family farms (Caffrey, 2014). Green Giant has also been instrumental in lobbying efforts against the labeling of genetically engineered food, saying the labels would increase production costs, making food more expensive (Russia Today, 2014). The media and the public have interpreted these movements as Green
25 Giant trying to ke ep consumers in the dark about what they are buying so the company May 24, 2014 across 50 countries asking the public to sign a petition calling for a five year ban on gene tically engineered food in order to conduct more comprehensive tests (March against Green Giant, 2014). AgLabs has not had the same media coverage as Green Giant, with only around 121 articles written in The New York Times f rom 2000 to 2015 about t he compa ny ( The New York Times, 2015b ). The biotechnology company has produced relatively similar seed products and has the same stance on labeling laws as Green Giant but has not been as publicly ridiculed. When AgLabs did appear in popular media, it was almost always in an article involving Green Giant. Typically, the media has focused on the agritech giants and their lobbying against the labeling of genetically engineered food (Dubois, 2014). Genetically Engineered Food in the Media Consumers have often sought agricultural information from news media, due to their lack of knowledge about the industry (Zimbelman, Wilson, Bennett, & Curtis, 2005). The popular media has often reported misinformation, due to the complexity surrounding genetic engineering technology and the lack of credible sourc es ( Whitaker & Dyer, 2000). This theme of misinformation has not been confined to just the agricultural industry, but has been seen as an issue with science research in general (Weigold, 2001). The general consumer has shown limited science knowledge, and scientists have not been trained in communication, making it difficult for people to gather accurate information. Due to the media reporting on information, which has often
26 been vague or biased, consumers have often made conclusions with limited informatio n (Goodwin, 2013). As the prevalence of genetically engineered crops in the U.S. has risen, so has the coverage by the media. The New York Times, one of the most circulated newspapers in the world, published 660 articles about genetically engineered food f rom 2 000 to 2015 ( The New York Times, 2015a ). Despite the numerous scientific advancements made in genetically engineered crop production, global coverage of genetically engineered food from 1997 to 2001 has been approximately 90% negative, focusing on hea lth risks (Abbott, Lucht, Jensen, & Jordan Conde, 2001). Genetically engineered crops wer letter written to The New York Times editor by Paul Lewis in 1991 . This term gained popularity in 1998 when non government organizations (N toward the technology (Lemaux, 2008). Content analysis of newspaper coverage of agricultural biotechnology has shown that the media typically cover the danger of the technology rather than the safety ( Hoban , 1995). Additionally, Marks , Kalaitzandonakes, Allison, and Zakharova (2002, 2003) identified newspapers in the U.S. and U.K. as covering the environmental risks of genetically engineered plants over the benefits. Companies have also st arted expressing their views on the debate of genetically engineered food. Chipotle is one of the most vocal businesses, having created two separate televised commercials supporting locally grown organic food and even creating a satire which aired on HULU injected with various drugs before being released by a benevolent farmer (Chipotle,
27 2014). The burrito restaurant has claimed that genetically engineered food production has supported only big agricultural businesses i nstead of the farmer, and little objective research has existed to support any of the reported benefits of genetically engineered foods. Chipotle said that plants have evolved alongside people naturally for centuries and that tampering with the food is wro ng (Ells, 2014). In 2015, Chipotle said they would remove any and all genetically engineered food from their menu, which made them the first national restraint chain to use all non genetically engineered ingredients (Chipotle, 2015). During the past severa l years, the media has focused its attention on genetically engineered food and the lack of labeling of food products containing genetically engineered material. A number of states, including California and Vermont, have voted on whether labeling of geneti cally engineered food should take place (Chaussee, 2014). Newspapers have also claimed that labeling should occur, because people have the right to know what they are eating and that current label laws hide the ingredients, making it dangerous for consumer s to eat ( Rabin , 2014). Agricultural Communications The knowledge gap between agriculture and consumers has become increasingly apparent as rural and urban areas have begun to intersect ( Wachen h eim & Rathge, 2000). Knowledge gap, referring the difference b etween information known and understood in agriculture, has been contributed to by consumer skepticism about agricultural practices and technology. In order to develop effective agricultural
28 communication practices, understanding consumer perceptions relat ed to agriculture and the developments of these perceptions is essential ( Verbeke , 2005 ). Communicating about Genetically Engineered Food Innovations in the food industry are facilitated through the use of new technologies, like genetic engineering ( Siegr est , 2008). As the agricultural industry has made improvements in practices and technology, consumers have become increasingly skeptical toward the advancements ( Sparks, Shepherd, & Frewer, 1994) . Consumer acceptance of new technologies is essential for th eir success in the food industry ( MacFie production, such as genetic engineering, have failed to be addressed by the agricultural industry (Goodman & Dupuis, 2002). Consumers typically have limited knowledge of new technologies, including genetic engineering ( Durant , Bauer, & Gaskell, 1998). Experts within the biotechnology engineered food (Frewer, Scholde r, & Bredahl, 2000), but their lack of knowledge may be a result of the agricultural industry not properly communicating with the public (McCullum Gomez & Palmer, 2010). Not understanding genetic engineering has made it difficult for consumers to decide ab out possible risks associated with the technology (S iegrest, 2008). In order to make up for their lack of knowledge, consumers have had to rely on the trust of communication to lessen the complexity of their attitude formation ( Earle & Cvetkivich, 1995). H owever, there has been a lack of communication with the public about genetically engineered food, which has led to debates about the safety of the product and caused distrust toward food producers among consumers (McCullum Gomez & Palmer, 2010).
29 Research P roblem Genetically engineered food s have been proven safe and beneficial, but consumers have remained suspicious of the technology and have called for tighter regulations. Modernization of the agricultural industry has developed a disconnect between the fa rmer and the consumer (Zimbelman et al. , 1995). The public has too often been misinformed about the facts surrounding genetically engineered food, causing skepticism and distrust toward the product (Durant et al., 1998; Siegrest, 2008). These views have pe rhaps been fueled in combination by the negative portrayal of genetically engineered food by the media and lack of communication between the public genetically engineered f ood has forced them to rely on the trust of the communication for information (Earle & Cvetkivich, 1995). This has created a need to develop better communication practices, mainly with a focus on the consumer (Telg & Irani, 2012). Purpose & Objectives The purpose of this study was to analyze how persuasive communication influenced genetically modified food . The following objectives guided this study: 1. Compare in attitude toward genetically modified food after receiving persuasive communication from Green Giant, AgLabs, FDA, or USDA. 2. Compare food after receiving persuasive communication from G reen Giant, AgLabs, FDA, or USDA. 3. change in attitude toward genetically modified food.
30 4. Determine how change in risk perception of genetically modified food. Significance This research will give greater insight into changes in attitude and risk perceptions made by consumers toward genetically engineered crops after receiving persuasive communication . Since consumers possess limited knowledge of new technology, such as genetic engineering, they have to trust communica tion about the products to be accurate (Durant et al., 1998; Earle & Cvetkivich, 1995). The agricultural industry has lacked exhibiting communication with the public, making it difficult for consumers to trust the technology and food producers ( McCullum Go mez & Palmer, 2010). In order for a new food technology to be successful, consumers must first accept it ( MacFie , 2007). Consumer acceptance is largely guided by their perceptions, which must be further analyzed to develop proper communication for genetica lly engineered food ( Verbeke , 2005 ). The results of this research can be applied in the industry to help consumers better understand the technology and to make more informed decisions about genetically engineered products. Industry companies, retailers, ad vertisement agencies, agricultural communicators, extension agents, and government organizations can use this information to develop messages that will be better received by the consumer. Agricultural companies, agricultural communicators, and food retaile rs can use the results of this study to create effective messages and communication to promote the sale of genetically engineered food. Advertising agencies can use this at titude and as a result, develop better advertising campaigns. Extension agents will find this research useful when presenting information about genetically engineered food
31 to the general public or teaching farmers how to present information about genetical ly engineered products. Finally, government agencies can use this research to guide their future communication strategies and policies. Summary Humans have been modifying the genetic traits of plants for centuries. It was not until the late 1980s that scie ntists were able to manipulate specific genes in organisms, even inserting DNA from one genome into another (Henig, 2000; Hinchee et al. , 1988). Genetic engineering of plants has given the world crops which will grow faster, cheaper , and use less land, helping to solve the issue of the expanding population (James, 2007; Chassy, 2007). The technology has also created crops resistant to disease and higher in vitamin content (GMO answers, 2014). Even though genetic engineering has been p roven safe and the science detailing the possible dangers is limited, the media has still portrayed genetically engineered food negatively (Abbott et al., 2001). Additionally, the agricultural industry has not properly communicated with the public on the t opic, which has led to skepticism and concern (Goodman & Dupuis, 2002). Consumers have been demanding tighter government regulation, as well as labeling laws, and have grown more and more concerned about the environmental and health impacts of genetically engineered food (Phillips, 2008). Consumers have also held strong feelings toward the companies producing the seeds. Organized marches against agri tech companies have even petitioned for a five year ban of genetically engineered food (March against Green G iant, 2014). Consumers have been misinformed concerning genetically engineered food, making it more difficult for the industry to market these scientifically advantageous products (Weigold, 2001). Agricultural communicato rs need to determine more effective
32 messaging strategies to allow the public to make informed decisions. Looking at message sources and the effect they have on attitudes and risk perceptions of genetically engineered food has been one approach to the probl em (Durant et al. , 1998) . This study will be valuable to the industry, communicators, extension agents, and government organizations to create new communication strategies to aid consume rs in making informed decisions .
33 CHAPTER 2 RELEVANT LITERATURE Chapter 1 discussed the advantages and disadvantages of using genetically importance of properly communicating about the technology. Chapter 2 focused on the theoretical foundation for th formation of attitudes toward genetically engineered food. Shannon and Weaver Communication Model Claude Shannon developed a classical model for transmitting information in 1949 at Bell Telephone Lab oratories (Lee & Baldwin, 2004). Warren Weaver discovered that the model could be applied to more than just the engineering project for which Shannon had intended (Shannon & Weaver, 1949). The model (Figure 2 1) has been used to examine a number of communi cation processes between people, from interpersonal to mass communication (Lee & Baldwin, 2004). The Shannon and Weaver communication model explains the linear process of how a message moves from the information source to the final destination (Lee & Baldw in, 2004). The model starts by using a source to develop a thought or idea . A transmitter then transforms the thought into a signal, which moves through a channel . Finally, a receiver accepts the signal, creating a new mental image of the thought for the f inal destination. In simplest form , the model demonstrates how Person A is the information source with a thought, using his/her mouth to transmit the message . Air can is the receiver. As the destination, Person B then develops a mental image of the message (Lee & Baldwin, 2004). The transmitter is not always the only sound in the room (Lee & Baldwin, 2004). Unwanted
34 signals, called noise, can add clutter and distort a message (Lee & Baldwin, 2004; Shannon & Weav er , 1949). Even though this model was intended for the transmission of voices over radio waves, communicators have applied the theory to the encoding and decoding of messages (Lee & Baldwin, 2004). In the example described above, Person A had to transform a thought into a code the receiver could understand. The code in that case was language. Person B then had to decode the message to create his or her own receiver during the decoding process and alter the intended message (Lee & Baldwin, 2004). Attitudes (Alport, 1935 , p. 6 ). Attitude has also been defined as: evaluation . 155) manner with respect to a given object . 6) which predisposes an individual to react in a characteristic way to any object or situation with which it is related . , p. 804 ) These various d efinitions have demonstrated that many components are involved in attitude formation. The reoccurring themes have been that attitudes are a learned tendency and can steer behavior (Perloff, 2008). No one is born with attitudes, and much of initi al opinion formation occurs during adolescence. Attitudes are developed over time through various socializations and interactions (Perloff, 2008). Tesser (1993) argued that attitudes are not genetic, except for inherited traits like taste
35 and smell, which can impact attitudes toward things like food or perfume. Attitudes are essential in the development of thought and action (Perloff, 2008). They help to Behavior is also desire to stay consistent with their opinions and actions (Perloff, 2008). Characteristics of attitudes, like structure and strength, c an vary from person to person, as well as situation to situation (Perloff, 2008). Structure, such as general versus highly specific, is one way to differentiate attitudes (Ajzen & Fishbein, 1977). A gene ral attitude is considered the global evaluation, can be used in many different situations, and is typically directed toward an object (Perloff, 200 8). A specific attitude differs because it is evaluative of a single incident and is typically directed toward a behavior (Perloff, 2008). For most cases, genera l attitude cannot predict behavior but can be generalized to the public. In contrast, a specific attitude cannot be applied to a general public but is predictive of behavior. Attitude titudes make more predictable behavioral decisions (Perloff, 2008). This can be best demonstrated with political issues. If individuals feel strongly about immigration, they are more likely to lobby for legislation or stand in a picket line. Characteristic s of the person can be just as influential toward behavior as characteristics of the attitude. Two major factors influence attitudes : Self monitoring habits and direct experiences of the individual (Perloff, 2008). Snyder (1974) described self monitoring p eople as those who pick up social cues to attempt to behave in a self monitors differ by looking at
36 their internal feelings rather than the situation to decide how to behave. Experience has also been concluded to affect attitude (Perloff, 2008). Direct experience can lead to 185). The theory o behavior, along with the subjective norm in society, and their perception of how much control they have over a behavior. Fishbein and Ajzen (1975) developed a separate model to predict behavior called the theory of reasoned action. This concept relies on four different factors: attitude toward the behavior, subjective norm, behavioral intention, and behavior (action in a situation). Both the theory of planned behavior and the theory of reasoned action suggest that attitudes do not predict behavior when subjective norms apply. That is, when there is pressure from peers or society, behavior will not always refl ect attitude. In most cases though, these theories support the idea that people will behave according to attitude. Persuasive Communication Since attitudes are developed over time from learned experiences, attempts to change an existing attitude can be met with resistance. Persuasion can be described as their attitudes or behaviors regarding an issue through the transmission of a message in ff, 2008, p. 17). Persuasion is often a symbolic process (Perloff, 2008). The symbols can range flag or the Starbucks logo. Communicators can use symbols to alter the opin ions of the
37 pub lic (Perloff, 2008). Persuasion , the study of attitudes and how to change them, has also attempted to influence individuals (Perloff, 2008). However, persuasion has not always been seen as acceptable. In order for persuasion to succeed, comm unicators the recipient must also persuade himself or herself for an attitude to ch ange. Whalen (1996) said, You [cannot] force people to be persuaded you can only activate their desire and show them the logic behind your ideas. You [cannot] move a string by pushing it, you have to pull it. People are the same. Their devotion to and tot al commitment to an idea come only when they fully understand and buy in with their total being. (p. 5) Another important aspect of persuasion is that it involves the transmission of a message (Perloff, 2008). This message can be verbal or non verbal, rati onal or irrational, via mass media or person to person (Perloff, 2008). People must also have free choice for persuasion to be effective. Since self persuasion is essential, individuals must be able to freely process the information (Perloff, 2008). Persua sion can be described as having an impact on attitude by shaping, reinforcing, or changing the attitude (Miller, 1980). Perceived association can attract and mold attitudes; this type of communication is called s haping persuasion (Perloff, 2008 ). An exampl e of this type of messaging has been when celebrities endorse a makeup brand to convince consumers that the products will make them beautiful. Reinforcing persuasion serves to strengthen current attitudes (Perloff, 2008). If a consumer is already a fan of a certain football team, exposure to a commercial will only strengthen their attitude. Finally, changing persuasion refers to communication trying to actively alter attitudes (Perloff, 2008). Changing persuasion has often been seen in politics,
38 racial segr egation , , and requires repeated exposure to the communication (Perloff, 2008). The Yale attitude change approach was the first empirical research that was conducted on the effect of persuasive communication (Perloff, 2008) . Hovland, Janis, requires that receivers learn the message arguments, and change in attitude occurs in a series of steps. While studies have indicated that as people become more knowledgeable about an argument, they are more likely to accept the position, researchers cannot assume that people will be able to just passively rece ive information and understand it (Chaiken, Wood, & Eagly, 1996; Perloff, 2008). The cognitive response approach to persuasion has suggested that an approach has helped t o fill a missing gap in the Yale attitude change approach (Brock, 1967). Cognitive responses can include both favorable responses, as well as criticisms elicited by the message. Persuasion can only occur if people view the message as more favorable than un favorable. However, some researchers have argued that the cognitive approach still has issues. First, it assumes that people always think highly about messages, and secondly, the cognitive approach does not examine how messages influence people ( Perloff , 2 008 ) . To address issues with the cognitive model approach, researchers have developed process based models for persuasion. The Heuristic Systematic Model (Chaiken, Liberman, & Eagly, 1989) and the Elab oration Likelihood Model (ELM; Petty
39 & Cacioppo, 1986) have been the most common process based models used to understand persuasion. These are both dual process models that have examined how ways of thinking and processing information can affect persuasion. The Heuristic Model uses two routes to describe how individuals process information (Chen & Chaiken, 1999). The first route is the systematic route, which requires the receiver to use a high level of thought when considering an argument. The second route is heuristic processing, which allows the person to apply memories and cognitive structures to an argument rather than intense analysis (Chen & Chaiken, 1999). While the Heuristic model has been proven reliable in persuasion research, the ELM has been examined more often in research (Perloff, 2008 ), as it provides a more comprehensive understanding of persuasive communication effects (Perloff, 2008). Elaboration Likelihood Model The Elaboration Likelihood Model (Figure 2 2) of persuasion was originally developed to account for both active and passive processors of information (Petty, Brinol, & Priester, 2009). ELM demonstrates that the consequences of persuasion can be different, depending on whether the thought process is high or low (Petty & Cacioppo , 1986; Petty & Wegner, 1999). The model ex plains how an individual processes persuasive communication, but it can also be used to explain attitude shifts, which are not always associated with persuasion (Petty & Cacioppo, 1986). There are two routes in wh ich attitude change can occur: t he central processing route and the peripheral processing route. The central processing route occurs when an individual uses careful consideration, along with past experiences, to develop opinions (Petty et al., 2009). The peripheral processing route uses a less exte nsive thought process;
40 instead, the route relies on peripheral cues, like message source or number of arguments (Petty & Cacioppo, 1986). Seven postulates have provided the basis of the ELM (Petty & Cacioppo, 1986) : 1. People want t o hold what they believe t o be correct attitudes. 2. Elaboration is on a continuum and is dependent on the situation of the argument, as well as the amount and nature of issue relevant material in the message. 3. Variables can play multiple roles within an argument depending on the conte xt of the communication. The role can be a persuasive argument, serve as peripheral cue, or impact the extent of elaboration. 4. Objective processing occurs when variables influencing ability and motivation to process are unbiased and either enhance or reduce scrutiny toward an argument. 5. Biased variables can lead to either positive or negative ability/motivation to process information and create bias in the issue relevant thoughts of the receiver. 6. As issue relevant thinking increases, the use of peripheral cue s decreases. However, when argument scrutiny is low, the impact of peripheral cues increase. 7. Attitudes changed by the central processing route are more persistent and are resistant to persuasion. Elaboration Petty and Cacioppo (1986) defined elaboration in which a person thinks about the issue 129). The likelihood of elaboration is high when the motivation and ability to engage in thinking about an issue are also high (Petty & Caciop po, 1986). High elaboration leads people to draw upon past experiences, scrutinize and elaborate on present messages, and access relative associations and images to develop overall attitudes toward the message (Petty & Cacioppo, 1986). The model suggests t hat once elaboration has structure, or schema, and a new attitude toward the object will be formed ( Cacioppo & Petty, 1986 ). Elaboration can be conducted in a fairly object ive manner when
41 dependent on the strength of the issue relevant arguments in the message (Petty & Cacioppo, 1986). Other times, the elaboration is actually biased and governed by the The amount of ela boration people use in response to persuasive communication is considered to be on a continuum (Petty & Cacioppo, 1986). Elaboration ranges from no thought of considering the issue relevant material of a message to scrutiny of every argument presented, res (Petty & Caci oppo, 1986). The central route of persuasion is used when the motivation and ability to scrutinize arguments are high, leading to a higher likelihood of elaboration (Petty & Caci oppo, 1986). The peripheral processing route is used when motivation and ability are low, resulting in a low likelihood of elaboration (Petty & Cacioppo, 1986). Central Processing Route As mentioned earlier, the central processing route is used when the mo tivation and ability to process a message are high, consequently leading to a greater likelihood of elaboration and attitude change (Petty & Cacioppo, 1986). The process involves actively generating positive or negative thoughts in response to a persuasive message (Petty et al., 2009). Not every argument is interesting to a consumer, and not every situation allows time for proper evaluation of the message (Petty et al., 2009). However, when the content is relevant, and individuals are able to carefully cons ider the message, people can evaluate the extent to which they agree or disagree with the arguments, ultimately affecting attitudes. As mentioned previously, the more knowledgeable individuals are toward the subject, the more likely they are to use greater elaboration and thoughtfully process the information. The purpose of this route is to determine if the arguments presented have any merit (Petty et al., 2009).
42 The information that is considered central to an issue can differ, depending on the person vie wing the message and the situation itself (Petty et al., 2009). For example, when considering social issues (such as capital punishment), some people may feel religious considerations hold more weight, while others focus more on the legalistic and logistic al arguments (Cacioppo, Petty, & Sidera, 1982). The most important dimensions of the argument, as evaluated by the person, receive the most scrutiny and lead to elaboration (Petty et al., 2009 ; Petty & Wegener, 1998 ). If the individual has the ability and motivation to process the information, either toward the topic remain the same, the original attitude will be retained (Petty et al., 2009). Once individuals have developed their thoughts toward the message, the new thoughts need to be integrated into their cognitive structure for attitude change to occur (Petty et al., 2009). If there is no change in the cognitive structure, the individual will shift from use of th e central processing route to the peripheral processing route for attitude formation (Petty et al., 2009). Cognitive structural change is more likely to occur if the attitudes are rehearsed and held with confidence (Petty et al., 2009). Even if the new th oughts become fully integrated, one cannot always cla im the thoughts to be accurate (Petty et al., 2009). Regardless of how extensive the information processing may be, bias can still occur (Petty et al., 2009). Bias can be a direct result of an time of message exposure (Petty et al., 2009). Essentially, attitudes are changed as the result of an extremely thoughtful process when people use their past experiences and knowledge, along with the dimensions they deem as central to the issue, to integrate
43 thoughts into their schema (Petty et al., 2009). After a change in cognitive structure occurs, there will either be a central positive or central negative attitude change (Petty et al., 2009). Attitude change through the central processing route has distinct characteristics (Petty et al. , 2009). These attitudes are predictive of behavior, persistent, held with confidence, and resistant to change until ch allenged by opposing arguments (Petty, Haugtvedt, & Smith, 1995). These persistent attitudes are cognitive structure (Petty et al., 1995). Peripheral Processing Route bility to process information is low, and persuasion occurs through the peripheral processing route (Petty et al., 2009). This route acknowledges that people are not always actively thinking about the communication they receive and sometimes must rely on p eripheral cues to form an attitude (Petty et al., simpler means of evaluations when exposed to persuasive co mmunication (Bem, 1972). After individuals are exposed to persuasive communication, they must first be motivated to process the information (dependent on persona l relevance, need for cognition , etc.) using the central processing route (Petty et al., 2009). If motivation is not present, the peripheral processing route will be used. This process will be effective in changing attitude if the peripheral cues are operating effectively. If the cue is not effective, the initial attitude will be retained. Just beca use the individual is motivated to process does not guarantee use of the central route. A person must also possess the ability to process the communication. The ability to process information is often
44 associated with knowledge of the topic (Petty & Caciopp o, 1986). When knowledge of the topic is limited, elaboration is reduced, and the peripheral route will be used (Petty et al., 2009). Individuals either retain their initial attitude or experience a peripheral attitude change, depending on the effectivenes s of the peripheral cue (Petty et al., 2009) . Peripheral cues can serve to elicit positive associations with a product, like the pleasant scenery in a commercial, similar to classical conditioning (Staats & Staats, 1958). Message sources are one type of pe ripheral cue and can be viewed as experts by the recipients, eliciting more favorable responses to an argument than just the message alone (Chaiken, 1987). Additionally, when a large group supports a message, others will follow simply because it appears co rrect (Axsom, Yates, & Chaikem, 1987). This bandwagon effect has been used by a number of persuasive speakers in the past (Lee & Lee, 1939). The peripheral approach has proven to be effective in the short term (Petty et al., 2009). The issue lies in the fa cues associated with messages can dissipate (Petty et al., 2009). Therefore, attitude accessible, enduring, and resistan (Petty et al., 2009, p. 135) than attitudes formed by the central processing route . Essentially, this type of attitude formation is the result of passive evaluation of simple cues with a weaker foundation (Petty et al., 2009). As t ime passes, these cues lose meaning, and attitudes shift back to the original thoughts of the individual (Petty et al., 2009).
45 Prior Knowledge process information (Petty & Cacioppo, 1986). When people are well informed concerning an issue, they are much more likely to thoughtfully process a persuasive message. Having more knowledge about the subject allows people to evaluate the legitimacy of the information more carefully and identify shortcomings of the communication (Wood, Rhodes, & Biek, 1995). Since people who are knowledgeable about a topic process information with a higher amount of elaboration, they typically use the central processing route (Wood et al., 1995). Thos e who are not as informed rely on the peripheral route. These peripheral processors are not as confident in their opinions and are more susceptible to contradicting persuasion to their new attitude (Perloff, 2008) . Source Cues Petty and Cacioppo (1986) pro 134). A common example of a peripheral cue is the message source (Petty et al., 2009). The way a source is perceived has been linked to the likeliness of elaboration and changes in attitude (Priester & Petty, 1995). McCroskey , 1997, p. 87). Source credibility is part of a two way interaction between a communicator and receiver (Perloff, 2008). Thus, just because a speaker may be famous does not mean he or she is viewed as credible by the listener. A number of components have been found to comprise a credible source. Perloff (2008) listed exper tise, trustworthiness, and goodwill as the most researched and most important aspects of credibility.
46 Expertise refers to the perceived knowledge of the communicator (Perloff, 2008). udes, and experts have often been perceived as credible (Perloff, 2008). However, communicators should be careful how they use experts. For example, if a group is trying to communicate with inner city drug abusers, using doctors would not be the best choic e. Even though doctors are knowledgeable about public heal would not view them as expert. A better choice would be a former drug user with whom the intended receivers could connect (Perloff, 2008). Trustworthiness refers to th (Perloff, 2008). The trustworthiness of a source can sometimes be a crucial component of credibility. Communicators can lack expertise, but if they are viewed as trustworthy, persuasio n can still occur . Goodwill communicators make receivers feel as though the speaker has their best interest at heart. Sources that are considered credible typically have at least one of the previously mentioned aspects (Perloff, 2008). Attitudes toward Genetically Engineered Food The ELM explains changes in attitudes (Petty & Cacioppo, 1986), which is why it is important to understand attitudes toward genetically engineered food before further discussing ELM related research. The Center for Public Issues Education in Agricult ural and Natural Resources at the University of Florida conducted a public opinion study regarding food in Florida with a segment focused on genetically modified organisms (GMOs), which is another common term for genetically engineered food ( Rumble & Leal, 2013). The survey was distributed online to 500 respondents in Florida, and demographics were weighted to reflect the 2010 Florida census. Slightly over 40% of respondents were unsure if GMOs had improved their quality of life, and the majority
47 agreed or strongly agreed that their food quality used to be better. Once again, around 40% of the sample was unsure if scientists should genetically modify crops to make them resistant to disease. The same proportion of respondents was unsure if they h ad ever consu med or purchased GMOs . Fewer than 40% felt that GMOs were a possible solution to world hunger, as well as pest and disease problems. However, around the same percent were unsure about these benefits. Almost half of the respondents agreed that GMOs presente d a greater risk for food allergies and food poisoning and were unsure if they threatened the environment. GM O food was also identified as being artificial and unhealthy. The survey also looked at the purchasing intentions of the Florida consumers. Almost 40% disagreed that they would purchase food labeled as GMO (Rumble & Leal, 2013). When asked about specific products, around 40% of respondents reported they would not purchase meat from animals that were was fed GMO feed, and slightly less said they would not purchase GMO produce. When asked about GMOs being used to combat citrus greening (a disease in Florida threatening the citrus industry), 52% responded that genetic engineering should be used , and 42% said they would purchase GMO citrus. The researcher s concluded that most respondents are likely answering The only time a positive response was seen was for the citrus greening questions, and this was likely because there was greater personal relevance with that topic (Rumble & Leal, 2013). attitudes toward genetically modified food and their intent to purchase the products. The
48 study used the term genetic modification to describe food that had been genetically altered in some way (Bredahl, 2001), which is similar to the term genetic engineering. Over 2000 consumers were interviewed in Germany, Denmark, the United Kingdom (UK) , and Italy about their attitudes toward genetically modified yogurt and beer. The study found that attitudes toward genetically modified products were similar among the Denmark, Germany, and UK consumers, but the Italian consumers had typically less negative associations with ge netically modified food (Bredahl, 2001). The attitudes were influenced by the perceived risks and benefits of the food. However, consumers did not distinguish between the risks and benefits of the product compared to the technology (genetic modification ). Bredahl (2001) concluded that the perceived risks and benefits attitudes toward food biote chnology and causes them to reject the technology all together (Bredahl, 2001). Demographics and Attitude Attitudes toward genetically engineered food have been strongly influenced by the demographics of consumers . Verdu r me and Viaene (2003) developed a m odel on Finnish consumers purchasing intent for genetically modified food, and began the model with the cultural and socio economic impact of the consumers . The model suggested that the demographic characteristics greatly influenced genetically modified food and overall attitudes and risk perceptions of genetically modified food. This model was developed from qualitative data and a review of existing literature. Through quantitative data collection, other research has concluded that socio demographic characteristics did not clearly predict attitudes, but political values did
49 when examining Greek consumers ( Antonopoulou, Papadas, & Targoutzidis, 2009 ). Antonopoulou et al. (2009) added that age did not have a huge impact on attitudes ; h owever, younger consumers typically held more favorable attitudes toward genetically modified food. Literature has also illustrated how education level can have an impact on the perceptions of risk associated with genetically modified food ( Hall & Moran, 2 006 ; Gaskell , 2003; Moon & Balasubramanian, 2001 ). Consumers with post graduate degrees have been identified as having lower perceptions of risk for genetically modified food. A study by Irani, Sinclair, and Malley (2001) described how various demographic characteristics impacted the perceptions of GMOs and GMO labels. Race, gender, and culture were examined to see if they had an effect on attitudes. A survey was distributed to approximately 400 college age students at three different universities in the U.S. An overwhelming 85% of respondents agreed that GMO food should be properly labeled, and demographics showed no impact on the responses. The majority of the respondents reported that even if the food were labeled GMO, they would still consider purchasing the product. The majority of white and Hispanic respondents said they would consider purchasing food labeled GMO, but only 33% of African American respondents said yes to this question. Additionally, m en were significantly more likely to consider purchasing the labeled food, as were the students located at a more rural campus. Respondents were also asked how much trust they placed in six different sources. The Food and Drug Administration (FDA) and the United States Department of Agriculture (USDA) were the first and second most trusted sources, respecti vely, and the companies producing the GMO products were the least trusted (Irani et al., 2001).
50 Pounds (2014) identified significant differences in the purchasing intent of GMOs between men and women in the state of Florida. Overall, females had a lower pu rchasing intent compared to men. Men appeared unsure if they would purchase GMOs, while women were less likely to engage in purchasing behaviors (Pounds, 2014). The study also concluded that both genders supported ballot initiatives to label GMO products, but women agreed more that they would support the initiative. Other studies have determined that women held more negative perceptions of genetically modified food compared to men ( Lockie , Lawrence, Lyons, & Grice, 2005), and were less likely to accept GMOs ( Hall & Moran, 2006 ). Knowledge and Attitudes The majority of people in developed nations have a good familiarity with the concepts of genetics ( Condit, 2010 ), and believe science to have a positive impact on society ( Pew Research Center, 2009 ). Scientist s have thought that greater understanding of science would lead to more support for research. A study by Evans and Durant (1995) concluded that as science knowledge increased, so did general attitudes toward science in general. However, respondents who rep orted a higher level of knowledge reported a lower level of acceptance for morally contentious areas of research. Literature has also shown that when respondents gain ed new information about genetically modified food specifically, their negative attitudes were actually enhanced ( Grice & Lawrence, 2003 ). Similar studies have concluded that an increase in knowledge did not necessarily have a positive influence on attitudes toward genetically modified food (McFadden & Lusk, 2015; Verdurme & Viaene, 2003) .
51 Sour ce Credibility Research in the ELM Understanding the role of source credibility in the ELM was essential to this research, and a number of studies have examined its relationship with attitude change within the model . Hovland and Weiss (1951) first determin ed that high credibility sources produced greater attitudinal change than low credibility sources. The study also noted that credibility had greater effect on attitudes toward a topic when people had less prior knowledge on the subject and saw the message as less relevant (Hovland & Weiss, 1951). Petty and Cacioppo ( 1979 ) found that when message relevance and source expertise interacted, the source cue was more effective in determining attitudes toward low relevance messages. Another study by Petty et al. i n 1983 looked at message endorsers of advertisements using magazines. The researchers concluded that using a celebrity endorser of a product was important for low relevance messages, compared to high relevance messages (Petty et al., 1983). These studies d emonstrated the importance of source credibility when message relevance was low. Trustworthiness of a source has been identified as one factor that can impact motivate d to hold correct attitudes, perceived trust in a source may impact elaboration by validating an argument (Petty et al. , 2009). Therefore, when a source is perceived as trustworthy and knowledgeable, people assume the source presents accurate information ( Petty et al., 2009). This trustworthiness of sources allow people to be et al., 2009). A study by Priester and Petty (1995) manipulated trust of a source while keeping the expertise high. Th e manipulation occurred either by making a speaker look dishonest or advocating a self serving position. Regardless of how the trustworthiness
52 was altered, less trustworthy sources led to greater elaboration than the trusted sources. Other studies supporte d these results and showed that when trustworthiness was low, elaboration was high (Priester & Petty, 2003). This aligned with research showing that as people were less inclined to think about issues, they were forced to elaborate when presented with a dis trusted source (Petty et al., 2009). However, people who enjoyed thinking elaborated equally despite the level of trust associated with the source (Petty et al., 2009). The previously mentioned research presented the sources prior to the message (Petty et al., 2009). When the source was revealed after message exposure and thought processing had begun, c onfidence in thoughts increased if the source was considered an expert (Brinol, Petty, & Tormola, 2004). However, this effect was reversed when weak argument s were supported by a credible source, likely because weak argument (Petty et al., 2009). These studies demonstrated the importance of source cues when people were not li kely to use a great deal of cognitive effort when thinking about persuasive communication (Petty et al., 2009). Trustworthiness and expertise support most messages when motivation to process was low. However, distrusted sources sometimes caused higher elab oration for those who may not typically be inclined to thoughtful thinking (Petty et al., 2009). ELM in Agricultural Research ELM research in agriculture has been used to help researchers understand icultural products. Verbeke (2005) wrote a literature review that described communication about agriculture and the
53 food industry. Information processing was identified as a major component affecting as a model guiding this process. A study by Verbeke and Ward (2006) looked into attitudes associated with a beef traceability campaign in Belgium. The researchers us ed a pre and post campaign survey to measure the impact of information cues. The campaign u sed full page advertisements in over 20 newspapers and offered a phone number for participants to call to receive an information packet. Only around 300 consumers called for more information out of the estimated 15,000 people exposed to the advertisements. This lack of participation supported the assumption that the motivation or ability to process the message was low, and likelihood for elaboration was likely limited. A separate study by Verbeke and Vackier (2004) analyzed how issue involvement could alter attitude formation with a study involving perceptions of meat. Their sample was divided into four groups; meat lovers, meat consumers, cautious meat lovers, and concerned meat consumers. The research showed that only meat lovers (highly involved) were int erested in intangible qualities, in addition to the tangible qualities the other groups prioritized. This supported the view that involvement is connected with motivation to process persuasive communication and can lead to higher elaboration. ELM studies have not been confined to just agricultural products. A study conducted by Morgan and Gramann (1989) used the model to develop effective teaching strategies for wildlife education. An experimental design was used to manipulate the amount of information pre sented to children, along with their level of involvement of the subject (snakes). The information presented to the children was in the form of a slide show. Students who only saw the slide show and were not exposed
54 to the snakes saw no attitude change. Ho wever, students who viewed the slide show and were able to interact with live snakes in the classroom exhibited an attitude change. This was likely because the snakes were not of high importance to those who only viewed the slide show, and the children use d the peripheral route to form opinions. The process the information and use of the central processing route. Research involving message testing and message frames has als o used ELM to examine how communication impacts attitude. Meyers (2008) examined how persuasive communication influences media coverage of agricultural biotechnology. Positively framed messages were used to determine the impact of the frame on the communic information. The study also looked at issue involvement and prior attitudes. Meyers (2008) concluded that preexisting attitudes had more effect on attitudes toward agricultura l biotechnology than issue involvement, which indicated that high amount of elaboration was not likely. The study did not directly look at the routes of information processing , but it did suggest further research about the topic. A similar study conducted by Goodwin (2013) used ELM to explore how personal communication about the livestock industry. The study found that message transparency had an impact on attitude and trust, but pe rsonal relevance did not. Even though the ELM suggests that personal relevance is associated with the motivation to process information, transparent communication may have been more salient, allowing it to have a greater impact on attitudes. The lack of si gnificance was also supported by
55 assumptions that food is considered a low involvement good (Beharrell & Denison, 1995), and prior knowledge may be confounded by personal relevance (Petty & Cacioppo, 1986). These findings indicated that in the absence of t ransparent communication, consumers do not exhibit a great deal of elaboration concerning agricultural messages. In general, resear ch has supported that consumers use a low amount of elaboration when presented with information about agricultural products ( Goodwin, 2013; Meyers, 2008; Morgan & Gramann, 1989; Verbeke & Vackier, 2004; Verbeke & Ward, 2006). Frewer , Howard, Hedderley, and Shepherd (1997) concluded that the majority of food related decisions made by consumers are developed using the peripheral p rocessing route. ELM Research with Genetically Engineered Food ELM has also been applied to research specifically concerning the communication of genetically modified food. Frewer, Howard, Hedderley, and Shepherd (1999) examined how personal relevance and persuasiveness impact attitudes toward genetically modified food. An experimental design was used to present information of This form of data collection asks participa nts to write down any thoughts that cross their mind after exposure to communication (Petty, Cacioppo, & Schumann, 1983). The list was analyzed by experts on both the dimensions of the thoughts, as well as the number of thoughts (Petty et al., 1983). The r esults of this study were contradictory to predictions made by the ELM. Personal relevance, which is associated with motivation to process information, did not influence the elaboration process to the extent researchers expected. In fact, messages low in r elevance led to more elaborative
56 processing. The researchers suggested this might be because participants felt they did personal relevance was high and products were already available for sale. The study also showed that respondents became more negative when exposed to negative cues but were not more positive when exposed to positive cues. This finding may have been the result of thought listing measuring the strength of part mediating cognitive responses. Positive attitudes may not have been expressed by respondents who did not feel strongly enough about the issue, thus altering the results. Krause, Meyers, Irlbeck, and Chambers (2015) used the ELM to guide a content analysis of YouTube videos for and against Proposition 37 in Cal ifornia. If the bill were passed , genetically engineered food would have be en legally required to be labeled. The bill did not pass, and the study found that scientists were typically used as sources in the videos opposing the proposition. Krause et al. (2015) concluded scientists offered high credibility and worked effectively as a peripheral cue. In addition to sources used in the videos, message frames were also analy zed. Different from prior research, this research identified emotionally driven frames supporting the bill. The researchers concluded that agricultural communicators should shift from using fact based message s to more emotional appeals to target non agricu ltural consumers (Krause et al., 2015). Risk Perception and Genetically Engineered Food Risk communication research has been conducted on food products using the ELM to determine how different variables affect consumer attitudes (Frewer et al. , 1997). Risk perception has often driven consumer acceptance of products, as opposed to actual risk estimates made by professionals (Frewer, Howard , & Aaron, 1998). Food technology in particular can possess a number of risk factors , which are of great
57 concern to c onsumers (Ronteltap, van Trijp, Renes, & Frewer, 2007 ). Specifically with genetic engineering, the unknown consequences of the technology likely shape the risk perception (Sparks & engineered food supports the trend that consumers have typically viewed products as riskier when effects of the possible hazard are mostly unknown (Slovic, 1987). The effect is heightened when th e proposed hazard is viewed as hidden by the producers ( van Kleef et al., 2006 ). A study conducted by Frewer , Howard, and Shepherd (1998) examined how initial attitudes toward GMOs affect communication about food production. A survey was distributed to ass perceptions associated with GMOs. The researchers concluded that prior risk perception is an important indicator for attitudes after exposure to a message. When respondents viewed G MOs with a higher level of risk, they perceived the information source as being less knowledgeable and less trustworthy. The study also found that when a source admitted uncertainty of risk rather than denying it, consumers viewed the source as more credib le. Source Credibility and Genetically Engineered Food As more demand has been placed on regulations requiring the labeling of genetically modified food, communication research has looked into the effects of different labels, along with message sources. A number of studies have reviewed trust associated with regulatory agencies, but not specific companies in the agricultural industry ( Barnett, Cooper, & Senior, 2007; Poortinga & Pidgeon, 2005; Siegrest, 2000). Frewer et al. (1997) looked specifically at ho w source credibility impacts attitudes within the ELM. A distrusted source (government), trusted source (consumer
58 organization), and collaboration of both types of sources were tested using an experimental treatment. Results showed that the hypothesized di strusted source (government) was connected to information about GMOs, which resulted in greater acceptance (Frewer et al., 1997). The research also looked at prior attitude and determined that positive initial attitudes toward GMOs were not greatly affecte d by the credibility of a source. However, respondents who were initially found to have negative attitudes showed a greater impact on their attitude from the message, depending on the source. Individuals with negative prior attitudes viewed the sources as less trustworthy and less knowledgeable than those with positive attitudes. Credibility also appeared to be linked with attitude formation toward GMOs, but attitude was dependent on various rnment and consumer organization both endorsing a message), respondents did not improve perceptions of information or credibility, as the researchers expected. This was likely because the respondents did not expect the two sources to agree, leading to redu ced impact of the consensus (Frewer et al., 1997). Frewer et al. (1999) conducted a similar experiment to study the relationship of source characteristics, personal relevance, and persuasiveness in communication about GMOs. Once again, the government was used as a distrusted source, while a consumer organization was treated as a trusted source. Thought listing was used to determine the amount of elaboration used. This study showed that elaboration was high when persuasive information about GMOs was low and source credibility was high. The same held true when persuasive information was high and credibility was low. These results indicated that these conditions facilitated the central processing route of elaboration.
59 he information source is an important (Frewer et al., 1999, p. 45). This contrasted with previous research showing that the message source did not impact attitudes ab out microbiological risk. The study recommended that distrusted sources remain proactive in their communication with the public about controversial technologies. Conceptual Model The conceptual model (Figure 2 3) for this study was based partly on the Shan non and W eaver (1949) m odel for communication, as well as ELM (Petty & Cacioppo, 1986). The two models were used to describe how individuals show a change in attitude and change in risk perception of genetically engineered food products after being exposed to some type of persuasive communication. The conceptual model shows how a message is encoded and decoded before reaching the final destination. The studies in the literature review used a variety of terms for genetically engineered food, but this concept ual model used genetically modified to describe the food since the public has been familiar with the term ( Miller, Annou, & Wailes, 2003). The encoding section of the model was the persuasive communication about genetically modified food that the responden ts received. This communication signal would be transmitted, and the decoding process would begin. The peripheral processing route from the ELM was used to model the decoding process. Research in agriculture using the ELM has shown that respondents almost exclusively use the peripheral route (Frewer et al . , 1997). As described by ELM, peripheral cues, such as sources , can be used to influence attitude formation. Since consumers have been skeptical of organizations involved in the developme nt of genetically modified food (McCullum -
60 Gomez & Palmer, 2010), the message source served as the noise in the model (Shannon & Weaver, 1949). The noise has the potential to distort the message before the recipient can begin decoding. The decoding process of the communicat demographics (age, race, sex , education, income, and whom food was purchased for), prior knowledge of genetically modified food, and source credibility. Prior research knowledge of genetically modified food (Verdurme & Viaene, 2003). Additionally, l iterature has suggested that people have limited knowledge concerning genetically modified food (Durant et al., 1998; Rumble & Leal, 2013), which would decrease their ability to process information and use a higher level of elaboration (Wood et al., 1995). Prior k nowledge has been perception of source credibility (Frewer et al., 1999; Frewer e t al, 1997; Petty et al., 2009), and s ou rce credibility has been identified as having influence on final attitudes after receiving communication (Petty et al., 2009). R esearch has also suggested that there is an influence on risk perception of genetically modified food in relation to source cred ibility and prior knowledge ( Frewer et al., 1998; Sparks & Shepherd, 1994). Since the ELM was used to guide this model, the change in attitude and change in risk perception were measured as the dependent variables rather than just the final attitude or fin al risk perception of genetically modified food . Summary communication process (Shannon & Weaver, 1949). Part of the communication model displayed how noise could interfere with the messa ge signal and distort the purpose of
61 a topic (Perloff, 2008). Elaboration Likelihood Model is one way to explain attitude change as a result of persuasive communication (Petty & Capaccio, 1986). ELM suggests that two different information processing routes can be used when people are exposed to communication (Petty & Capaccio, 1986). The r oute used by a person depends on his or her motivation and ability to process the message. Prior knowledge et al. , 1995). These individuals with prior knowledge use higher elaboration and the central processing route. Those who do not have the knowledge to scrutinize a message use lower elaboration through the peripheral processing route. This route relies on peripheral cues, like sources, to inform opinion. In order for a source to be effe ctive, it needs to be viewed as credible (Perloff, 2008). Research focused on attitudes toward genetically modified food has shown that people are typically negative and unsure about the technology (Rumble & Leal, 2013). ELM research in agriculture has de monstrated that people form attitudes using the peripheral route because they simply do not possess the motivation to thoughtfully analyze agricultural messages (Frewer et al., 1997). Trust in the information source has appeared to be the most important fu message (Frewer et al., 1999). Research related to risk perceptions of genetically modified food has shown that when initial associated risks are high, sources are typically viewed as more distrusted and les s credible, leading to the peripheral processing route ( Frewer, Howard, & Shepherd, 1998).
62 theory to demonstrate the process of how a message is encoded by an information source befo re being decoded by the target destination. Noise, like a message source, can distort the message before reaching the recipient. The peripheral processing route demographics , prior know interpret the message . The decoding process will result in a change in attitude and risk perception of genetically modified food.
63 Figure 2 1 . ommunication (1949) .
64 Figure 2 2. The Elaboration Likelihood Model of persuasion (Petty et al., 2009) .
65 Figure 2 3. Conceptual mo del of the a ffect of persuasive communication change in attitude and risk perception of genetically modified food . Adapted from the ELM ommunication .
66 CHAPTER 3 METHOD O LOGY Chapter 1 described a growing problem the agricultural community has been facing over the past several decades. Consumers have become more skeptical about genetically modified food, even though the technology has been scientifically proven to be safe and advantageous. Chapter 2 described the theoretical framework guiding this study u sing the ELM and Shannon and ommunication. Literature source credibility were discussed pertaining to their influences on attitudes and risk perception related to genetically modified food. The pur pose of this thesis was to attitude and change in risk perception of genetically modified food. The followin g objectives guided this study: 1. titude toward genetically modified food after receiving persuasive communication from Green Giant, AgLabs, FDA, or USDA. 2. food after receiving persuasive communication from Green G iant, AgLabs, FDA, or USDA. 3. Determine how the message source, prior knowledge of genetically modified food, and source credibility change in attitude toward genetically modified food. 4. Determine how the m of genetically modified food, and source credibility change in risk perception of genetically modified food. Experimental Design This research was a quantitative study that utilized a pretest posttest experimental design within a survey to answer the research objectives. One intervention
67 was used in this study with four different variations of the treatment. The intervention was the source attributed to a message which desc ribed genetically modified food. Four groups were used, each one presented the same message about genetically modified food, but each group used only one of the four sources. Even though genetic engineering is the technically correct term (FDA, 2014), the questionnaire designed for this study used the term genetically modified because consumers have been more familiar with the term (Miller et al. , 2003). Additionally, genetic engineering has less positive associations than genetic modifications (Miller et a l., 2003) and could have engineering to provide the following definition of genetic modification to the DNA in order to promote a desired Two government agencies and two agricultural biotechnology companies were selected as the message sources. The FDA , USDA , Green Giant, and AgLabs were chosen as the information sources based on con flicting literature and lack of research for the credibility associated with these organizations/companies ( Barnett, et al., 2007; Frewer et al., 1997; Irani et al., 2001 ; Poortinga & Pidgeon, 2005; Siegrest, 2000 ) . The names Green Giant and AgLabs were se lected as pseudonyms for the companies used in the study; however, respondents were exposed to the actual names. Literature has suggested that consumers have had little trust in government organizations regulating a product or the companies which have deve loped them ( Rothenberg & Becker , 2004 ). However, the research has been inconsistent, and other studies have suggested that agencies like the FDA, along with the USDA, have been more trusted than consumer
68 organizations when concerning genetically modified food (Frewer et al., 1997; Irani et al. , 2001). The FDA and USDA were selected as the two government sources since research indicated these were trusted sources (Irani et al., 2001) . Little research had been conducted examining how the specific agricultura l companies delivering a message coverage to choose the indu stry sources. Green Giant had been frequently reported on in the news, and while AgLabs was a similar company, it had only rec eived a fraction of the publicity ( The New York Times, 2015 a, 2015b ) . Additionally, these are the two leading companies in the development of g enetically modified seeds (Fern andez Cornejo et al., 2014). A pretest posttest design was used to measure attitu de and risk perception change resulting from the source treatment. The independent variable, X , was the message source and was manipulated by using FDA, USDA, Green Giant, or AgLabs in each group ( Table 3 1) . As stated previously, the message stayed consta nt, and only the source was manipulated between groups (Appendix C) . The survey instrument also asked questions about knowledge of genetically modified food . After completion of the posttest questions, respondents were asked questions to measure . The message source served as the independent variable in this study, and the instrument measured change in attitude toward genetically modified food and change in risk perception of genetically modifi ed food as dependent variables. Moderating and mediating variables were also measured in this experiment. A moderating variable affects the strength and/or direction of the relationship between the independent and
69 dependent variable (Baron & Kenny, 1986). The moderating variables in this study were the prior knowled ge of genetically modified food and consumer demographics . Mediating variables are different than moderators because they intervene between the independent and dependent variable (Baron & Kenny, 1986). Also, variations in the independent variable impact the mediator, and variations in the mediator impact the dependent variable. The source credibility measure was a mediating variable in this study . Population and Sample Size This study look ed at Florida change in attitude and change in risk perception of genetically modified food. related issues and genetically modified food has been vital for the success and sustainability of the agricult ural industry. Florida agriculture has contributed $7.8 billion 2011 ) ; one billion dollars alone came from the citrus industry in 2012 ( NASS , 2012). The Florida citrus i ndustry has recently been affected by the devastating disease citrus greening, and with limited solutions, genetically modified citrus may be the only solution ( Bove, 2012 ). Additionally, the Florida House of Representatives (2015) has denied two bills whi ch would have required the labeling of genetically modified food if passed. The , combined with the threat of citrus greening and increased proposals for regulation , has made it important to study Florida att itudes and risk perception of genetically modified food. The target population was all Florida residents ( N = 15,321,354 ) who were 18 years and older ( United States Census Bureau, 2014). A required sample size of 385 was calculated using a margin of error of +/ 5%, a 95% confidence interval , and a standard deviation of 0.5 (Ary, Jacobs, &
70 Sorensen, 2010). The survey was distrib uted to 7 70 respondents, and 523 ( n = 523) of the surveys were usable, due to incomplete questionnaires and respondent errors . The survey also included quality check questions, which required respondents to select a specific answer (e.g. select strongly agree). If the requested choice was not selected, the survey was terminated and responses were excluded from analysis. The quality check questions were used to reduce straight line responses. Outliers were also removed, which made the sample size 514 ( n = 514) respondents. This l arger sample size resulted in a smaller margin of error (+/ 4.3 % , Ary et al., 2010). Non probability sampling with an opt in panel was used for the sample in this study (Baker et al., 2013). In non probability sampling, not every person in the population has the same chance of being chosen for the research. This study was limited to only people who had Internet access and had opted to take the survey. Opt in panels consist of respondents who have typically been recruited in advance and have agreed to compl ete surveys. Opt in panels have evolved over time, and previous research on these types of panels held little relevance to the current methods used (Baker et al., 2013). The public survey software company, Qualtrics, hired to administer the survey, used mo netary incentives to recruit the opt in panel for this study. Limitations associated with non probability sampling, like selection, exclusion, and non participation bias, can be addressed by using post stratification sampling. Post stratification sampling has been used to weight the sample after data collection, based on demographic characteristics of the population (Baker et al., 2013). T his study weighted the sample based on the 2010 Florida census fo r sex , race, ethnicity, age , and rural/ urban continuum . The weights and population percentages for the individual
71 demographics can be seen in Table 3 2. The ages were later grouped into the following generations for analysis by birth year: Millennials and younger (1977 1996), Generation X (1965 1976), Young Baby Boomers (1955 1964), Old Baby Boomers (1946 1954), and the Silent Generati on and older (1945 and earlier; Zickuhr, 2010 ). Some demographic groups had to be condensed for analysis, and demographics were fully described in Chapter 4. Rounding error fr om the w eighting of the respondents changed the n from 514 to 515 ( Maletta , 2007) . Post survey adjustment of the non probability sample has been shown to mirror the effects of probability sampling , but selection bias can still occur (Baker et al., 2013). While post stratification weighting does increase generalizability of the sample to the population (Baker et al., 2013), random assignment of the sample to the four treatments groups was more important. Randomization provides the best way to achieve the co ntrol necessary for an experiment to evaluate the independent variable (Ary et al., 2010). Qualtrics was programmed to randomly assign respondents to each group, and the weighting of the demographics allowed the four groups to be equal in regards to the ch aracteristics of the respondents. Data Collection Before data collection began , the survey instrument was approved by the research ( IRB#2013 U 0494 , Appendix A ). An infor med consent form, along with the purpose of the study and the survey, were submitted to the IRB prior to release of the survey. Data collection occurred after the IRB approved the instrument and procedures. The survey was created in an online public opinio . An online survey was deemed appropriate for the study since a larger sample could be
72 collected when compared to mail or telephone surveys and respondents could easily be randomly assigned to one of the four treatment groups (Dil lman , Smyth, & Christian , 2009). The company, Qualtrics, was employed to col lect the data for this study. The survey was released in September 2014 and was open for 10 days before closing. All respondents were given an anonymous survey link to protect the privacy of the individuals. Post stratification weighting procedures were completed after the survey was closed. Validity and Reliability of the Instrument ( Ary et al. , 2010, p . 225) and is a vital consideration in the development of a survey. The focus of validity has shifted over recent years from the Validity of this r esearch was supported by the adoption of previous instruments , which had operationalized the conceptual constructs (Ary et al., 2010; Hallman & Metcalf, 1993; Frewer et al. , 1997; Frewer, Howard, & Shepherd, 1998 ; Osgood, Suci, & Tannenbaum, 1971; Roe & Te isl , 2007 ; Rumble & Leal, 2013 ). Validity was also ensured through the use of a panel of experts. The panel included a University of Florida professor in the Plant Molecular and Cellular Biology program, two faculty members associated with the UF Center fo r Public Issues Education in Agriculture and Natural Resources (PIE Center) , and three industry leaders known for their expertise in agricultural policy and specialty crops. A soft launch of the survey (similar to a pilot study) was used to ensure the inst rument was working properly and free of error (Dillman et al. , 2009 ). The soft launch
73 also allows researchers to make appropriate adjustments to the instrument before the survey has been released on a larger scale (Ary et al., 2010). Threats to Validity i n an Experimental Design Researchers must determine if the conclusions made about the relationship between variables demonstrated in an experiment are valid or not (Ary et al., 2010). Cook and Campbell (1979) identified four different types of validity: in ternal validity, external validity, construct validity, and statistical conclusion validity. Internal validity Internal validity is necessary for correct conclusions to be made from an experiment (Campbell & Stanley, 1963). A number of different threats to internal validity were present in this study: History effect. History effect can occur when extraneous events happen outside the experimental treatment at the same time as the study and could alter the outc ome (Ary et al., 2010). The longer the time is between the pretest and posttest, the greater the history threat becomes. An example of history threat for this study would be if research were released supporting the dangers of genetically modified food half way through the data collection. Respondents completing the survey after the event would likely have different attitudes toward genetically modified food than those who took it before the event. To lessen the history threat, the survey was only active for 10 days. Media was also tracked during this time, and no major news stories pertaining to GMOs, genetically engineered, of genetically modified food was covered in the national papers (The New York Times, 2015 a ) . Pretest sensitization. Pretest sensitizatio n can occur when the pretest causes respondents to think more carefully about the questions and give different responses in
74 the posttest (Ary et al., 2010). This would mean the pretest caused the change in attitude rather than the intervention (Ary et al., 2010). In this study this threat was present and could be used to explain unanticipated changes in attitude and risk perception. Instrumentation threat. Instrumentation threat to validity can occur when the instrument is altered during the study (Ary et a l., 2010). Changes can include the type of instrument, the difficulty level, and the way tests are administered (Ary et al., 2010). The instrumentation threat was limited by not changing the study once the survey was activated online. Selection bias. Selec tion bias can occur when there is a significant difference in the sample between the control and experimental groups before the study begins (Ary et al., 2010). To avoid selection bias, the survey computer software randomly assigned the respondents to one of the four treatment groups. Experimental attrition. Experimental attrition threat is present when there is a differential loss of participants from the treatment groups (Ary et al., 2010). This can alter the measurement of the dependent variable for the experiment. A survey panel was used to avoid this threat (Ary et al., 2010). The survey software company guaranteed complete surveys by each respondent through the use of incentives and randomly assigned each person to a treatment group. Construct validity Construct validity construct based on the measures, treatment, subject, and setting used in an experimen . 291). To account for construct validity, clear operational measurements of the construct were based on previous literature (Frewer et
75 al., 1997; Frewer, Howard, & Shepherd, 1998 ; Hallman & Metcalf, 1993; Osgood et al. , 1971; Roe & Teisl , 2007) and outlined in the conceptual model (Figure 2 3) . The following are t hreats to construct validity (Ary et al., 2010): Measures of the construct used were not appropriate leading to inaccurate results. Manipulation of the construct was not properly done leading to incorrect inferences. External validity External validity is concerned with the generalizability of the findings from a study (Ary et al., 2010). The following are threats to external validity: Selection treatment interaction. Selection treatment interaction can occur when results for certain subjects are not true f or a different kind of subject (Ary et al., 2010). This is typically the result of the sample not being representative of a larger population (Ary et al., 2010). Using volunteers is another threat to external validity since the sample may have different ch aracteristics than non volunteers (Ary et al., 2010). Post stratification weighting an d random assignment of the sample was used to limit the threat of selection treatment interaction (Kalton & Flores Cervantes, 2003) . Pretest treatment interaction. Pretest treatment interaction can cause respondents to be more or less sensitive to the experimental treatment (Ary et al., 2010). The only way to account for this threat would be to eliminate the pretest (Ary et al., 2010). Since the experiment looked at how persuasive communication changed attitude and risk perception , data collection would have been difficult without a pretest. However, small changes in attitude and risk perception in the results could be attributed to a pretest treatment interaction.
76 St atistical conclusion validity Statistical conclusion validity refers to the correct use of statistics to infer that a relationship between variables is true and not a result of chance (Ary et al., 2010). This was a threat to validity in this study because incorrect statistical procedures can lead to inaccurate interpretations of the results. To account for this threat all assumptions had to be met for the statistical procedures used. Each item measuring attitude toward genetically modified food, risk percep tion risk perception of genetically modified food, knowledge of genetically modified food, source credibility, and attitudes toward the source used an interval scale, which allowed for an index to be created for each variable. Survey Error Even though thi s study used an experimental design, it was administered in the form of a survey, so survey errors must be acknowledged and addressed. The five types of survey error are as follows: coverage error, sampling error, rounding error, nonresponse error, and mea surement error (Dillman et al., 2009 ; Maletta, 2007 ). Coverage errors. Coverage errors occur when members of the population are not given the same probability of being chosen, often due to the survey method (Dillman et al., 2009). For this study, coverage error could have occurred because the survey was administered online and an opt in panel was used. However, post stratification weighting was used to lesson this error (K alton & Flores Cervantes, 2003), and due to the experimental design of the study, it w as more important for random assignment of the sample to the treatment groups. Sampling Error . Sampling Error can occur when a sample is gathered from a larger target population (Dillman et al., 2009). For results to be considered generalizable
77 to the popu lation, the entire population would ideally be studied (Dillman et al., 2009). Since this is not always practical, a sample is used, and the larger the sample the smaller the margin of error. An online survey can lead to larger sample sizes, but non probab ility sampling methods can cause the sample to not necessarily be representative of the population. This study attempted to decrease the sampling error by using post stratification weighting methods, so the sample reflected the target hics (Baker et al., 2013) and by collecting a large sample of respondents . Rounding errors. Rounding errors can occur when using post stratification weighting methods ( Maletta , 2007). When respondents are weighted on more than one category, underrepresented cases will be weighted higher and over represented cases will be weighted lower. Data was analyzed in SPSS 21.0, which round ed the frequency of the demographic categroies to nearest integer. This rounding is not done on individual cases, but rather on the total weighted frequency (Maletta, 2007). This can cause inconsistency in the data, such as the sample reported as 515 cases rather than 514 in this study . Nonresponse err ors. Nonresponse errors are common in surveys and occur when not all the respondents in the sample complete the survey (Dillman et al., 2009). Nonresponse can impact results if a group or demographic elects not to complete the survey (Dillman et al., 2009) . For example, people with stronger negative feelings toward genetically modified food could complete the survey, but those who were more passive toward the subject may not feel the need to respond, thus skewing the results. Nonresponse error was accounted for by using post stratification weighting methods to
78 weight the sample to reflect the demographics of the population (Baker et al., 2013). This also ensured that respondents would be equally represented in each of the four treatment groups. Additionally, quality check questions were used which asked respondents to answer a question a certain way to ensure they were reading the survey and not straight lining their answers. Incorrect responses led to termination of the survey. Measurement error. M easurement error is the final type of error and can occur be the result of a complicated survey design or unclear questions (Dillman et al., 2009). The panel of experts used to ensure validity of the instrument served to lower measurement error. The panel reviewed the questionnaire content to make sure the questions were clear and concise. The study also removed any respondents who did not complete the survey or were outliers , c hanging the total number of actual respondents from 770 to 514. Instrumentation . The questions developed for this research were a part of a larger survey done for the Center for Public Issues Education in Agricultural and Natural Resources at the University of Florida as part of the 2014 Florida Food Panel (Anderson, Ruth, & Rumble, 2014). The complete questionnaire consisted of 62 questions (Appendix D) , and six of those questions were analyzed for this study (see Appendix B) . The intervention in the survey was the message source. Four groups were used, each containing only one source (FDA, USDA, Green Giant, or AgLabs). The constant in this study was the message , which was use d in all four groups and said,
79 Before [genetically modified foods] reach the market, crops from [genetically modified seeds ] are studied extensively to make sure they are genetically modi fied products are the most researched and tested agricultural products in history . (GMO answers, 2014, para. 16) Respondents saw the same message but were exposed to only on e of the four sources. The survey software randomly assigned an equal number of respondents to each group to account for selection bias (Ary et al., 2010). Real limits were created for each variable to aid interpretation of the results. The limits were ass igned to the scales to standardize the numerical data and allow for easier discussion of the descriptive data ( Sheskin, 2004) . A description for how each variable was measured was described in the following sections . Demographics Demographics were measured using a multiple choice or check all that apply question style. The following demographics were analyzed for this study: generation, sex, race, level of education, income level, rural/urban continuum, and whom consumers typically purchased food for. The d emographic data was measured through descriptive statistics. Prior Knowledge of G enetically Modified food through a seven item, five point Likert type scale adapted from an instrument used in general knowledge of science and technology, their knowledge of food science and food technology, and how much they had heard and read about genetically modified food. The scale was labeled as strongly disagree = 1 , disagree = 2 , neither agree nor disagree = 3 , agree = 4 , and strongly agree = 5 . A 5 indicated a higher level of prior knowledge
80 while a 1 indicated a lower level. For a scale to be considered reliable, it has to have a Cronbach alpha value of .7 or higher ( Field, 2013 ).The prior knowledge scale was this scale by adding the value for each item and dividing by the total number of items ( seven ). The real limits used to interpret the results for the respondents agreement with their knowledge of genetically modified food were 1.00 1.49 = strongly disagree, 1.50 2.49 = disagree, 2.50 3.49 = neither agree nor disagree, 3.50 4.49 = agre e, 4.50 5.00 = strongly agree. Source Credibility Source credibility was measured with a six item, five point Likert type scale that was shown after persuasive communication about genetically modified food. The scale used was as follows: 1 = strongly disagree , 2 = disagree , 3 = neither agree nor disagree , 4 = agree , and 5 = strongly agree . Higher source credibility was assigned a 5 and lower source credibility was assigned a 1. The reliability for the source credibility scale in each of the gr oups ranged between = .75 and = .85. The scale had six items to measure trustworthiness, knowledge, and goodwill of the source, which are the three areas defined as part of credibility by Perloff (2008). Items in the scale were adapted from an instrume nt used by Frewer et al. (1997). Indexed means were created for source credibility in each of the four groups by summating the items in the scale and were as follows: 1 .00 1.49 = strongly disagree, 1.50 2.49 = disagree, 2.50 3.49 = neither agree nor disagree, 3.50 4.49 = agree, 4.50 5.00 = strongly agree.
81 Attitudes toward Genetically Modified F ood Attitude was measured using an eight item, five point semantic differential scale. The scale was adapted from definitions of attitudes described by Osgood et al. (1971) Frewer, Howard, and Shepherd (1998). Eight different pairs of adjectives were used on a scale of 1 to 5. For analy sis, the negative adjective s were assigned a 1, and the positive adjective s (e.g. were assigned a 5. This variable was considered reliable; the pretest had an .95 in each of the f our groups. Indexed means were calculated for the overall pretest, and the posttest for each of the four groups by adding the value for each item in the scale and dividing by eight (separa te indexes were created for pre test and posttest). Real limits were used to interpret genetically modified food: 1.00 1.49 = negative, 1.50 2.49 = slightly negative, 2.50 3.49 = neutral, 3.50 4.49 = slightly positive, 4.50 5.00 = positive . Risk Perceptions of Genetically Modified F ood Risk perceptions of genetically modified food was measured using a six item, five point Likert type scale: strongly disagree = 1 , disagree = 2 , neither agree nor disagree = 3 , agree = 4 , and strongly agree = 5 . Lower perceptions of risk were the instrument were adapted fro m similar studies by Roe and Teisl (2007 ), Frewer, Howard, and Shepherd (1998 ), and Rumble and Leal (2013). Researcher developed items were included to assess risks perceived by consumers as described in Chapter 1. Risk perception was measured in the prete st and posttest , and separate indexes were
82 created by calculating the overall average of six items in the scale. Responses were categorized into real limits of 1.00 1.49 = strongly disagree , 1.50 2.49 = disagr ee , 2.50 3.49 = neither agree nor disagree , 3.50 4.49 = agree , 4.50 5.00 = strongly agree . Analysis Data for this study were analyzed using SPSS 21.0. Below is a description of the data analysis for each objective. Objective 1. Compare Florida cons modified food after receiving persuasive communication from Green Giant, AgLabs, FDA, or USDA. The dependent variable, change in attitude , was created by subtracting the index created for the prior attitude from the index of final attitude toward genetically modified food. An independent analysis of variance (ANOVA) was used to compare the change in attitudes between groups to determine if any significant differences existed between the groups using different mes sage sources. This type of analysis was used since the independent variable was categorical and the dependent variable was continuous. Initially, the assumptions for normality were not met for the change in attitude variable (Table 3 3 , Figure 3 5 ) . An acc eptable skewness and kurtosis would be +/ 2 (George & Mallery, 2010). The skewness for change in attitude was 1.14, but the kurtosis was 3.07. Nine outliers were removed from the data, and the adjusted skewness (1.04) and kurtosis (1.57) met the criteria for normality . After the dependent variable was adjusted for normality, the assumption s for normality were met for an ANOVA to be used (Field, 2012). The histogram for the adjusted change in attitude can be seen in Figure 3 6.
83 Homogeneit y of variance, or the assumption that variance in the change in attitude was similar in all four groups, was another assumption that had to be met. A , which test s the null hypothesis that variance between the groups was the same , can be used to test for homogeneity. This test was performed and was not significant ( p > .05), meaning there were no differences in variance (Field, 2013). Additionally, the large sample size of this study would generally lower issues with homogeneity of variance . Al l assumptions for the ANOVA were met, and a Bonferonni test was performed as post hoc analysis to identify which groups were significantly different for change in attitude. Objective 2. genetically mo dified food after receiving persuasive communication from Green Giant , AgLabs , FDA, or USDA. Similar to objective two , the prior risk perception index was subtracted from final risk perception index to create a dependent variable for change in risk percept ion . An ANOVA was conducted to identify significant differences between the message source groups for change in risk perceptions . Objective two also used a categorical independent variable with a continuous dependent variable, which is why an ANOVA was sel ected. Assumptions for normality were not initially met (skewness = 1.74, kurtosis = 8.79 , see Table 3 3 , Figure 3 7 ). After the removal of the nine outliers described in objective one, the adjusted skewness was .79 and the adjusted kurtosis was 1.64 (Table 3 3 , Figure 3 8 ) . The adjust ed data met the assumption s of normality. p > .05), and no issues with the homogeneity of
84 variance were identified. All assumptions the ANOVA was met an d analysis was performed. Objective 3. Determine how the prior knowledge of genetically modified food, and source credibility predict Florida change in attitude toward genetically modified food. A multip le regression analysis was performed to identify how well variables in the conceptual model (Figure 2 genetically modified food. Dummy variables were used for the different sources in order to compare one control ( FDA ) to the other three groups since literature has already identified the FDA as more trusted than the USDA or industry sources when communicating about agricultural biotechnology (Irani et al . , 2001) . Source credibility, prior knowledge, an d demographics were also included in the model. Assumptions for normality were met for both source credibility (Figure 3 3) and prior knowledge (Figure 3 1) before removal of the previously described nine outliers . The outliers for change in attitude and c hange in risk perception were completely excluded from analysis, but t he adjusted skewness and kurtosis for source credibility (Figure 3 4) and prior knowledge (Figure 3 2 ) still fell within +/ 2 and the exact values be seen in Table 3 3 . The demographic variables had to be dummy coded, and the category with the highest percent of respondents was used as the constant in the model (sex females ; education completing a four year degree; generation Millennial Generation or younger ; race whit e; income $25,000 to $49,999; Field, 2013). Since respondents were able to select multiple answers for the question asking who m they purchased groceries for, each predictor was treated as a dichotomous variable .
85 Hierarchical order of entry of the predictor variabl e s was used since the variables were likely related to each other as identified in previous research (Field, 2013 ; Irani et al., 2001; Wood et al., 1995 ) . The first model used only the grouping variable ( message source ), and models two and three included k nown predictors (Field, 2013): demographics (model two) and prior knowledge (model three; Irani et al., 2001; Wood et al., 1995 ) . Source credibility was not included until the final model since its importance was still unclear (Hovland & Weiss, 1951). Mult iple regression analysis assumptions were met because the outcome variable was continuous and an index was created, and more than one continuous or categorical predictor variable was used (Field, 2012). Some demographic groups had to be grouped since their n was relatively small. Ten cases per predictor is typically sufficient (Field, 2010), but due to the large sample size of the study, demographics were grouped to ensure approximately 30 cases were in each predictor (see T able 4 1 for demographics) . Addit ionally, the assumptions for normality were met once all outliers were removed (Table 3 3 ) . An additional concern for multiple linear regression was multicollinearity, which can occur when a strong correlation is present between two predictor variables. Th e variance inflation factor (VIF) indicates how strong the relationship is between two variables, while the toleranc e measures its inverse (1/VIF, Field, 2012). Multicollinearity is not an issue when the values for VIF are not substantially higher than one and when tolerance does not fall below 0.1. Table 3 4 indicated that there was little concern for multicollinearity since all values fell within the previously described parameters (Bowerman & O
86 Objective 4 . Determine how the prior knowledge of genetically modified food, and source credibility predict Florida change in risk perception of genetically modified food. Consistent with objective three , a multiple linear regression model using hierarchical order of entry of predictor variables was created using the following predictors : message source (dummy coded), demographics (dummy coded) , prior knowledge of genetical ly modified food , and source credibility . Predictor variables were entered in the same manner as objective three to predict the outcome variable, change in risk perception . All assumptions discussed in objective three were met in objective four (Table 3 3 , Table 3 4 ) . Limitations Since this study was administered online, the sample was limited to people who had a computer. In addition, the sample was limited to people recruited by the online public survey company use d to administer the instrument. Another l imitation for the online survey was information was only collected for the specific questions asked. This can sometimes lead to researchers missing information that may be important to the research problem that could have been found in qualitative re search . Limitations associated with non probability sampling include that not every person in the population has an equal chance of being selected (Avery et al., 2010). This limitation was reduced through post stratification weighting for the Florida population (Baker et al., 2013). Another limitation for this research was the feasibility for collecting information about the processing route used by the respondents when presented with persuasive information toward genetically modified food. A common way to gather this information wou ld be by using thought listing procedures, which would ask respondents to write
87 down every thought they have about a given subject. Since this research was a small section of a larger study, and was adminis tered online, the cognitive load for the respondents was assumed to be too high to use this procedure. It was not likely that the respondents would type out their thoughts without a researcher sitting alongside them. Additionally, since the survey already covered a number of topics with more than 60 questions, the number of questions this study could feasibly ask was limited. Another limitation to consider was that the other questions in the survey could have influenced s analyzed in this study. Also, using a pretest may have interacted with the treatment, and cause respondents to be less sensitive to the message prompt, which would result in small changes in attitude and risk perception. Assumptions An assumption for thi s research was that consumers, at the very least, had been exposed to genetically modified food and had already formed an opinion toward them. A similar assumption was that respondents were aware of the industry companies and government agencies used as so urces for the treatments and had some prior opinions toward those sources. Prior literature indicated that people rely on the peripheral processing route when forming attitudes about agriculture, including genetically modified food. This study assumed that people would follow this trend and take notice of the source presenting the information. Another assumption made was that the sample accurately reflected the population of Florida. The sample was weighted to reflect the 2010 Florida census in order to be more generalizable to the public. Since the researcher could not see who had completed the survey online, another assumption was that each respondent was a
88 different person. This was also assured by the public opinion company by using specific links for th e survey associated with each individual. Summary This research was conducted using a pretest posttest experimental design implemented through an online survey. The independent variable in the study was the message source: FDA, USDA, Green Giant, and AgLab s. The online pretests and posttests measured change in attitude toward genetically modified and change in risk perception of genetically modified food. Additional questions gathered information about g with their opini ons of the . The target population was all Florida consumers 18 years and older . The sample size was 515 ( n = 5 15 ) and provided a margin of error of +/ 4.3% (Ary et al., 2010). The survey software company, Qualtrics, used non probability methods to recruit an opt in panel with a monetary incentive. To lower various bias es associated with non probability methods, post stratification weighting was used so the s ample demographics better reflected to demographics of the population (Baker et al., 2013). Validity was accounted for by adapting previously used instruments and having a panel of experts review the questionnaire (Ary et al., 2010). Reliability was review ed by using a soft launch to ensure the instrument was working correctly (Ary et al., 2010). internal consistency errors included coverage, sampling, rounding, nonresponse, and measurement error ( Dillman et al., 2009; Maletta , 2007). Survey errors were accounted for by using post -
89 stratification sampling methods, discarding incomplete surveys, and having a panel of exp erts review the instrument before distribution. The instrument was adapted from six similar studies (Frewer et al. , 1997; Frewer, Howard, & Shepherd, 1998 ; Roe & Teisl , 2007; Rumble & Leal, 2013; Osgood et al. , 1971; Hallman & Metcalf, 1993). Likert typ e scales, along with bipolar semantic differential scales, were used to measure the previously described variables. Analysis of the instrument included descriptive statistics, ANOVAs, and multiple linear regression models using statistical software SPSS 21.
90 Table 3 1. Experimental design for the study. Note . X 1 = FDA, X 2 = USDA, X 3 = Green Giant, X 4 = AgLabs Pretest Independent Variable Pos t test O 1 X 1 O 2 O 1 X 2 O 2 O 1 X 3 O 2 O 1 X 4 O 2
91 Table 3 2. Proportional weights of demographics in the sample. Population % Proportional Weight Age 19 and under 1.3 .013 20 29 12.8 .128 30 39 12.2 .122 40 49 14.2 .142 50 59 13.5 .135 60 69 11.1 .111 70 79 7.4 .074 80 and older 4.9 .049 Sex Male 41.1 .489 Female 51.1 .511 Hispanic 22 .220 Race White 75 .750 American Indian/Alaska Native 0.4 .004 African American 16 .0 .16 0 Asian or Pacific Islander 2.5 .025 Multiracial 2.5 .025 Other 3.6 .036
92 Table 3 2. Continued. Population % Proportional Weight Rural/Urban Continuum Metro Counties in metro areas 1 million population or more 63.1 .631 Metro Counties in metro areas of 250,000 to 1 million population 25.7 .257 Metro Counties in metro areas of fewer than 250,000 population 4.8 .048 Non metro Urban population of 20,000 or more, adjacent to a metro area 3.5 .035 Non metro Urban population of 2,500 to 19,999, adjacent to a metro area 2.6 .026 Non metro Completely rural or less than 2,500 urban population, adjacent to a metro area 0.3 .003 Table 3 3 . Normality a ssumptions of variables. With Outliers Without Outliers Skewness Kurtosis Skewness Kurtosis Change in Attitude 1.14 3.07 1.04 1.57 Change in Risk Perception 1.74 8.79 .79 1.64 Prior Knowledge .372 .410 .328 .271 Source Credibility .334 .270 .190 .179 Note . Acceptable skewness and kurtosis level is +/ 2 (George & Mallery, 2010).
93 Table 3 4 . VIF and Tolerance for variables in objective three and four. Variable Model 1 Model 2 Model 3 Model 4 Tol. VIF Tol. VIF Tol. VIF Tol. VIF Green Giant .695 1.439 .663 1.508 .655 1.526 .655 1.527 AgLabs .688 1.454 .639 1.564 .637 1.569 .637 1.571 USDA .685 1.460 .613 1.633 .609 1.642 .608 1.646 Generation Generation X .690 1.449 .690 1.449 .686 1.458 Young Baby Boomers .688 1.454 .658 1.520 .652 1.534 Old Baby Boomers .734 1.363 .725 1.380 .720 1.388 Silent Generation or older .661 1.513 .626 1.596 .622 1.608 Men .886 1.129 .874 1.144 .845 1.183 Education High School Degree or less .669 1.495 .607 1.647 .607 1.648 Some college, no degree .704 1.421 .697 1.434 .694 1.442 Note. Acceptable tolerance does not fall below 0.1 and acceptable VIF is not substantially higher than 1.0.
94 Table 3 4 . Continued. Variable Model 1 Model 2 Model 3 Model 4 Tol. VIF Tol. VIF Tol. VIF Tol. VIF Education Two year college degree .714 1.401 .709 1.410 .709 1.411 Graduate or Professional Degree .792 1.262 .791 1.265 .789 1.267 Race African American .818 1.223 .817 1.223 .817 1.224 Other .845 1.183 .845 1.183 .842 1.187 Annual Income $24,000 or less .766 1.305 .765 1.306 .765 1.307 $50,000 $74,000 .727 1.376 .726 1.377 .725 1.378 $75,000 or more .720 1.388 .700 1.429 .696 1.437
95 Table 3 4 . Continued. Variable Model 1 Model 2 Model 3 Model 4 Tol. VIF Tol. VIF Tol. VIF Tol. VIF Purchase Self .803 1.245 .803 1.245 .797 1.255 Spouse .724 1.382 .713 1.402 .709 1.411 Children .754 1.326 .754 1.326 .753 1.328 Others .857 1.167 .857 1.168 .855 1.170 Prior Knowledge .754 1.326 .741 1.350 Source Credibility .874 1.144 Note. Acceptable tolerance does not fall below 0.1 and acceptable VIF is not substantially higher than 1.0.
96 Figure 3 1. Normality curve for prior knowledge prior to removal of outliers .
97 Figure 3 2. Normality curve for prior knowledge after removal of outliers .
98 Figure 3 3. Normality curve for source credibility prior to removal of outliers .
99 Figure 3 4. Normality curve for source credibility after removal of outliers .
100 Figure 3 5. Normality curve for change in attitude prior to removal of outliers .
101 Figure 3 6. Normality curve for change in attitude after removal of outliers .
102 Figure 3 7. Normality curve for change in risk perception prior to removal of outliers .
103 Figure 3 8. Normality curve for change in risk perception after removal of outliers .
104 CHAPTER 4 RESULTS Chapter 3 described how a n online survey was developed to fit the model for this research based o model . The population for the study was Florida consumers over the age of 18. Non probability sampling was used to collect panel data, and the sample demographics were weighted to match the 2010 Florida Cens us. The total sample for the study was 515 respondents, but rounding error can change the n for each objective due to weighting (Maletta, 2007) . The purpose of the study was to analyze the influence of persuasive in attitude and change in risk perception of genetically modified food. The experimental design for the study included one treatment with four different sources: Green Giant , AgLabs , FDA, and USDA. These sources served as the indepen den t variables and the message they presented was the constant. Change in risk perception and change in attitude toward genetically modified food were the dependent variables. Moderating and mediating variables included prior knowledge of genetically modified food, source credib ility, and demographics. Chapter 4 presented an analysis of the variables of demographics, variables of interest, and objectives. Analysis of Demographics Demographics for the respondents are reported in Table 4 1. The instrument was distributed to 770 Flo rida consumers, and 514 ( n = 514) completed the questionnaire. Demographics were weighted to match the Florida 2010 census for age, sex, rural/ urban continuum , race, and, ethnicity (Table 4 1). Due to rounding errors from post stratification weighting, the adjusted n was equal to 515 (Maletta, 2007) . The
105 weighted demographics will be described since all analysis used the weights; unweighted demographics can als o be seen in Table 4 1. For analysis, the weighted ages of respondents were grouped into appropriate generations, and t he majority of respondents n = 156). T here were more women (51.1%, n = 263) th an men , and the majority of respondents were white (75.1%, n = 387) ; 22.9% ( n = 118) of the sample indicated they were Hispanic. Education was also included in demographics, and the largest percent of respondents had graduated from college with a four yea r degree (31.4%, n = 162). Most respondents reported earning less than $50 ,000 for their annual income (54 .8%, n = 282) and resided in metro areas with a population over 1 million people (62.0%, n = 325) . The census for Florida also reported residents livi ng in urban and rural areas not adjacent to metro areas. None of the respondents in the survey indicated they lived in these area s , which is why the wei ghted percentages for the rural/ urban continuum do not equal 100%. Respondents were also asked who m they purchased food for as part of the demographic questions. The majority of respondents purchased food for themselves (97.5%, n = 502), their spouses (57.8%, n = 298) , and their children (29.2%, n = 151). Analysis of Variables of Interest Prior Knowled ge Prior knowledge was measured using a seven item, five point Likert type scale ranging from strongly disagree = 1 , disagree = 2 , neither agree nor disagree = 3 , agree = 4 , to strongly agree = 5 . This scale was adapted from an instrument designed by Hallma n and Metcalf (1993) with the overall reliability for the index The grand mean for respondents prior knowledge was 3.71 ( SD = .70 ) , which indicated respondents agreed that they were knowledgeable about genetically modified food .
106 Source Credibility An index for source credibility was measured using a six item , five point Likert type scale adapted from previous research (Frewer et al., 1997; Perloff, 2009) . The scale used the same label s as the prior knowledge index. The same credibility i ndex was completed by each group after exposure to the treatment and had reliabil ities ranging from = .75 and = .85 in each of the four groups. The average for each source credibility was close between AgLabs ( M = 2.92 , SD = .70 ) , FDA ( M = 2.91 SD = .76 ) , and USDA ( M = 2.93 , SD = .76 ); Green Giant received the lowest credibility average ( M = 2.87 , SD = .78 ). The overall mean for the s ource credibility index was 2.91 ( SD = .75 , Table 4 2 ) . The individual and overall credibility scores indicated th at respondents neither agreed nor disagreed that the source(s) was credible. An ANOVA was run to determine if there was a significant difference between the source credibility associated with the four different source s (Table 4 3). The p value was .93 , and no differences in credibility were identified ( F (3, 511) = .15, p = .93) . Change in Attitude Change in attitude toward genetically modified food was calculated using a pre test post test design. The prior attitude index was subtracted from the final attitude index . Attitude was measured using six bipolar semantic differentia l scales based off of attitude measurements suggested by Osgood et al. (1971) and Frewer, Howard, and Shepherd (1998 ) . Measurements included natural/artificial, unhealthy/healthy, dangerous/safe, beneficial/not beneficial, wholesome/not wholesome, and unnecessary /necessary. Negative adjectives were assigned a 1 and positive adjectives were assigned a 5 . The pre test index for attitude and posttest ing from .94 to .95 . T able 4 4 shows t he overall prior attitude index mean was
107 slightly negative ( M = 2.33, SD = 1.07 ) and final attitude mean was neutral ( M = 2.61, SD = 1.10) . Change in attitude was calculated, and FDA produced the greatest average change in attitude ( M = .40 , SD = .53 ) and Green Giant produced the smallest change ( M = .17 , SD = .54 ). Even though the FDA did produce the greatest change in attitude, it had the most negative attitude score both before and after receiving the message. T he grand mean for change in attitude was index . 28 ( SD = .54) . A paired t test was performed to see if there was a significant change in attitude for each group and for the overall change in attitude (Table 4 5) . All p values were less than .01, which mean t that there was a significant difference between the prior and final attitude. Change in Risk Perception Risk perception was measured through questions adapted from Frewer, Howard, and Shepherd (1998), Roe and Teisl (2007), and Rumble and Leal (2013) . A f ive point Likert type scale with six items made up the index. Lower perceptions of risk were assigned a 5 and higher perceptions of risk were assigned a 1 . The overall reliability for the prior risk index was .87 and the posttest reliability score fell between .83 and .88 in each group . The average for the prior risk perception index was 2.83 ( SD = .89 ) and final risk perception was 2.97 ( SD = .89). Both the prior and final risk perception score s indicated that re spondents neither agreed nor disagreed about the risks of genetically modified food. The results can be seen in T able 4 4 . Change in risk perception of genetically modified food was calculated by subtracting the prior risk perception from the final risk pe rception (Table 4 6 ) . FDA produced the greatest change in risk perception ( M = .16 , SD = .33 ), but Green Giant and AgLabs yielded similar changes in risk perception ( M = .15 , SD = . 35 and M = .15 , SD = .36 respectively ). The smallest change in risk perception came from the USDA with only a .09 increase in the mean
108 ( SD = .29). The overall average for the change in risk perception index was M = .14 ( SD = .33 ). A paired t test was performed to see if there was a difference b etween the prior and final risk perception of genetically modified food (Table 4 7) . There was a significant difference in all four groups and the overall change in risk perception (all p < .01). Analysis of Objectives Objective 1 . modified food after receiving persuasive communication from Green Giant, AgLabs, FDA, or USDA. Objective one examined if there were any differences in the change in attitude between the four differe nt groups. An ANOVA in Table 4 8 showed that there were significant differences between the four groups ( F (3, 511) = 4.24, p = .01). A Bonferro ni test was performed as a post hoc analysis to determine which groups showed sig nificant differences. T able 4 9 showed the results of the test. The only significant differences in groups were between Green Giant and FDA ( p = .00). The mean difference showed that Green Giant mean change in attitude was .23 lower than the mean for change in attitude t oward genetically modified food. Objective 2. genetically modified food after receiving persuasive communication from Green Giant , AgLabs , FDA, or USDA. Table 4 10 showed the ANOVA for change in risk perception between the source groups. There were no significant differences ( F (3, 511) = 1.36, p = .25), therefor e, post hoc tests were not performed.
109 Objective 3: Determine how the prior knowledge of genetically modified food, and source credibility predict Florida change in attitude toward genetically modified food. A hierarchical regression was performed to satisfy objective thre e. The first model used only the message source to determine if the source used could predict the change in attitude toward genetically modified food (Table 4 11 ) . The model was significant ( F (3, 511) = 4.237 , p = .006), and Green Giant and USDA were iden tified as significant predictors ( p = .001 and .022 respectively). When compared to the FDA, there was a .229 smaller attitude change when respondents were exposed to Green Giant ( B = .229) and .149 smaller change in attitude when exposed to USDA ( B = .149) . An examination of the means in T able 4 4 shows that attitude change was still positive for all four sources though. The R 2 value was .024 , which indicated the model only accounted for 2.4% of variance in change in attitude . The second model was also significant ( F (3, 511) = 2.137 , p = .002 ) and included demographics ( generation , sex, education, race, annual income, and who m the respondent purchased food for when shopping ) and the message source . Green Giant ( B = .252 , p < .000 ) and USDA ( B = .145 , p = .032 ) were still predictors of change in attitude . The Silent Generation and older were identified as significant predictors of attitude change as well ( p = .008 ). When compared to respondents in the Millennial Generation , those in the Silent Generation and older showed a .203 larger change in attitude ( B = . 203 ). Men were also a significant predictor ( p = .002 ); males had a .155 smaller change in attitude than females ( B = .155 ). The R 2 value increa sed from the first model by .059 to . 083 ; th e second model could account for 8.3 % of the
110 variance in change in attitude accounted for by the predictors . This was a significant change in the R 2 ( p = .026 ). Models three and four can b e seen in Table 4 12 . Model three was significant ( F (3, 511) = 2. 334 , p = .001 ), and used all the same predictors as before with the addition of prior knowledge of genetically modified food . The use of Green Giant ( B = .234 , p = .001) and USDA ( B = .133, p = .050) as a source were still significant predictor s of attit ude change. Men were also still a significant predictor of a change in attitude ( B = .141 , p = .004 ) , as was the Silent Generation or older ( B = .159, p = .039 ) . The addition of prior knowledge to the model did make education level significant. Respondents with a high school degree or less were predictors of attitude change when compared to those with a four year degree ( p = .027), and had a .166 smaller change in attitude com paratively ( B = .166). Prior k nowledge was also a predictor ( p = .015 ), and a s prior knowledge incr eased one unit, there was a predicted . 092 decrease in the change in attitude ( B = .092 ) . The third model represented 9.4 % of the variance (R 2 = . 094 ) in t he change in attitude as explained by the set of predictors , which was .0 11 higher than the second model . This was a significant change ( p = .015 ). The fourth and final model included all previous predictors along with source credibility . This model was s ignificant ( F (3, 511) = 3. 905 , p < .000) and ha d the highest R 2 value , representing 15.5 % of variance in the outcome of change in attitude ( R 2 = . 155 ). Green Giant ( B = . 237 , p < .000 ) , USDA ( B = .151 , p = .022 ) , men ( B = .193 , p < .000), and having a high school diploma or less ( B = . 156 , p = . 032 ) were still significant predictors of attitude change. Additionally, respondents who reported they purchased groceries for a spouse were a significant predictor compared to those who
111 did not; these respondents were reported to have a larger change in attitude ( B = .104, p = .052). With the addition of source credibility to the model, the Silent Generation or older ( p = .106) and prior knowledge ( p = . 086 ) were no longer significant predictor s of chan ge in attitude. However, s ource credibility was identified as a significant predictor ( p < .000 ) , and for every one unit increase in sour ce credibility, there was a . 188 increase in attitude change ( B = . 188 ). The R 2 for this model was .06 0 higher than in model three and was significantly different ( p < .000), which made model four the best fit. Objective 4. Determine how the prior knowledge of genetically modified food, and source credibility predict Florida risk perception of genetically modified food. Similar to objective three, a hierarchical regression model was used to answer objective four. Table 4 13 showed regression model one the change in risk perception after receiving persuasi ve communication about genetically modified food . Model one ( F (3, 511) = 1.362 , p = .254) was not significant , so subsequent models were not tested. The message source alone was not predictive of a change in risk perception . Post Hoc Analysis Post hoc tes ts were performed to further explore the influence of persuasive communication on both risk perceptions and attitude s toward genetically modified food. Change in Risk Perception Since the regression model for objective four was not significant for predicti ng a change in risk perception using only the message source, a regression was run including the remaining moderating and mediating variables without the pr esence of the source. Table 4 14 showed the results from this regression. The model was significant ( F (3, 511) = 1.924, p = .010), but it only accounted for 7.2% of the variance in change
112 in risk perception (R 2 = .072) . When demographic characteristics were examined , men were significant predictors of a smaller change in risk perception when compared to women ( B = .065, p have a smaller change in risk perception than white respondents ( B = . 109 , p = .043 ) . Additionally, when compared to respondents earning between $25,000 and $49,999 annually, those who earned $50,000 to $74,999 were predicted to have a larger change in risk perception ( B = .093, p = .015). Source credibility was not a significant predictor of change in risk perception ( p = .217), but prior knowledge was ( p = .030). As prior knowledge increased one unit, change in risk perception was predicted to decrease by .052 ( B = .052). Final Attitude The relatively small R 2 seen in the regression models for both change in attitude and change in risk perception indicated that the conceptual model created for this research did not predict the de pendent variables as anticipated . Since risk perception did not appear to be operati ng within the proposed conceptual model or ELM , further research on this dependent variable was not conducted for this study. The ELM does discuss attitude changes and shifts, but the model more specifically examines the actual final attitude rather than i ndividual changes (Petty et al. 2009). For this reason, the final attitude was treated as a dependent variable in post hoc analysis. Additionally, literature has suggested that prior risk perception can be predictive of final attitudes toward genetically m odified food (Frewer, Howard, & Shepherd, 1998). Therefore, prior risk perception was added as a moderating variable to the regression model along with message source, demographics, prior knowledge, and source credibility. Results from the regr ession can b e seen in Table 4 15 . The model was significant ( F (3, 511) =
11 3 68.845, p < .000) and could account for 74.6% of the variance in the final attitude toward genetically modified food (R 2 = .746). A number of different demographic categories were predictive o f the final attitude toward genetically modified food. Generation X ( B = .160, p = .026), Young Baby Boomers ( B = .215, p = .008), and Old Baby Boomers ( B = . 237 , p = . 008 ) were all predicted to have a more positive final attitude compared to the Millennial Generation or younger. Respondents earning less than $25,000 a year were predicted to have a more positive final attitude than respondents earning between $25,000 and $49 ,999 annually ( B = .183, p = .017). Finally, respondents who purchased food for themselves were predicted to have more negative final attitudes toward genetically modified food compared to those who did not purchase food for themselves ( B = .485, p = .007 ). Prior knowledge was not a significant predictor of final attitude toward genetically modified food ( p = .343), but prior risk perception ( p < .000) and source credibility ( p < .000) were significant predictors. As prior risk perception increased one un it, the final attitude increased by .776 ( B = .776 ) ; prior risk perception of a higher score indicate d more positive perceptions of risk. Source credibility was the final significant predictor, and as it increased by one unit, the final attitude toward gen etically modified food increased by .423 ( B = .423).
114 Table 4 1. Demographic characteristics of the respondents. n % Weighted n Weighted % Generation Millennials or younger 136 26.5 156 30.3 Generation X 102 19.8 115 22.4 Young Baby Boomers 120 23.3 91 17.7 Old Baby Boomers 93 18.1 65 12.7 Silent Generation or older 63 12.3 87 16.9 Sex Male 188 36.6 252 48.9 Female 326 63.4 263 51.1 Education High School Degree or less 93 18.1 98 19 Some college, no degree 131 25.5 121 23.5 Two year college degree 76 14.8 82 15.9 Four Year College Degree 159 30.9 162 31.4 Graduate or Professional Degree 55 10.7 52 10.1 Hispanic 52 10.1 118 22.9 Race White 463 90.1 387 75.1 African American 25 4.9 82 15.8 Other 26 5.1 47 9.1 Annual Income $25, 999 or less 95 18.5 87 16.9 $25,000 $49,999 188 36.6 195 37.9 $50,000 $74,999 130 25.3 138 26.8 $75,000 or more 101 19.6 95 18.4
115 Table 4 1 . Continued . n % Weighted n Weighted % Rural/ Urban Continuum Metro Counties in metro areas 1 million population or more 309 60.1 325 62.0 Metro Counties in metro areas of 250,000 to 1 million population 133 25.9 132 25.2 Metro Counties in metro areas of fewer than 250,000 population 25 4.9 25 4.8 Non metro Urban population of 20,000 or more, adjacent to a metro area 30 5.8 18 3.4 Non metro Urban population of 2,500 to 19,999, adjacent to a metro area 16 3.1 14 2.7 Non metro Completely rural or less than 2,500 urban population, adjacent to a metro area 1 0.2 2 0.4 Self 507 98.6 502 97.5 Spouse 302 58.8 298 57.8 Children 158 30.7 151 29.2 Other 72 14.0 74 14.4 Total 514 100 515 100
116 Table 4 2. Description of source credibility. Green Giant ( n = 120) AgLabs ( n = 128) FDA ( n = 137) USDA ( n = 131) Total M(SD) M(SD) M(SD) M(SD) M(SD) Source Credibility 2.87(.78) 2.92(.70) 2.91 (0.76) 2.93 (0.76) 2.91(.75) Note . 1.00 1.49 = strongly disagree, 1.50 2.49 = disagree, 2.50 3.49 = neither agree nor disagree, 3.50 4.49 = agree, 4.50 5.00 = strongly agree ; 5 indicated high credibility, 1 indicated low credibility. Table 4 3. ANOVA for source credibility between groups . SS df MS F p Between Groups .25 3 .08 .147 . 932 Within Groups 286.86 511 .56 Total 287.11 514 Table 4 4 . Description of attitudes toward genetically modified food. Green Giant ( n = 120) AgLabs ( n = 128) FDA ( n = 137) USDA ( n = 131) Total M(SD) M(SD) M(SD) M(SD) M(SD) Prior Attitude 2.47(1.06) 2.25(1.04) 2.13(1.01) 2.49(1.13) 2.33(1.07) Final Attitude 2.63(1.17) 2.55(1.12) 2.52(1.00) 2.73(1.10) 2.61(1.10) Change in Attitude .17(.54) .30(.55) .40(.53) .25(.51) .28(.54) Note . 1.00 1.49 = negative , 1.50 2.49 = slightly negative , 2.50 3.49 = neutral , 3.50 4.49 = slightly positive , 4.50 5.00 = positive. Table 4 5 . Paired sample t test between prior and final attitude. Green Giant ( n = 120) AgLabs ( n = 128) FDA ( n = 137) USDA ( n = 131) Total p .000 .000 .000 .000 .000
117 Table 4 6 . Description of risk perception of genetically modified food. Green Giant ( n = 120) AgLabs ( n = 128) FDA ( n = 137) USDA ( n = 131) Total M(SD) M(SD) M(SD) M(SD) M(SD) Prior Risk Perception 2.87(1.03) 2.83(.87) 2.75(.86) 2.87(.78) 2.83(.89) Final Risk Perception 3.03(1.07) 2.97(.84) 2.92(.86) 2.96(.78) 2.97(.89) Change in Risk Perception .15(.35) .15(.36) .16(.33) .09(.29) .14(.33) Note. 1.00 1.49 = strongly disagree, 1.50 2.49 = disagree, 2.50 3.49 = neither agree nor disagree, 3.50 4.49 = agree, 4.50 5.00 = strongly agree ; Lower perceptions of risk assigned a 5 and higher perceptions of risk assigned a 1 . Table 4 7 . Paired sample t test between prior and final risk perception. Green Giant ( n = 120) AgLabs ( n = 128) FDA ( n = 137) USDA ( n = 131) Total p .000 .000 .000 .000 .000 Table 4 8 . ANOVA for change in attitude. SS df MS F p Between Groups 3.57 3 1.19 4.24 .01 Within Groups 143.40 511 .28 Total 149.28 514 Table 4 9 . Follow up test for change in attitude. Mean Difference p Green Giant AgLabs .14 .27 FDA .23 .00 USDA .08 1.00
118 Table 4 9. Continued. Mean Difference p AgLabs Green Giant .14 .27 FDA .09 .93 USDA .06 1.00 FDA Green Giant .23 .00 AgLabs .09 .93 USDA .15 .13 USDA Green Giant .08 1.00 AgLabs .06 1.00 FDA .15 .13 Table 4 10 . ANOVA for change in risk perception. SS df MS F p Between Groups .45 3 .15 1.36 .25 Within Groups 56.80 511 .11 Total 57.26 514
119 Table 4 11 . Multiple linear regression analysis for variable s predicting change in attitude (Model 1 and 2). Variable Model 1 Model 2 B t p B t p Constant .395 8.726 .000 .403 2.074 .039 Green Giant .229 3.449 .001 .252 3.769 .000 AgLabs .093 1.426 .154 .083 1.249 .212 USDA .149 2.299 .022 .145 2.149 .032 Generation Generation X .005 .069 .945 Young Baby Boomers .039 .533 .594 Old Baby Boomers .116 1.441 .150 Silent Generation or older .203 2.682 .008 Men .155 3.171 .002 Education High School Degree or less .110 1.533 .126 Some college, no degree .044 .678 .498 Two year college degree .022 .300 .765 Graduate or Professional Degree .026 .305 .761 Race African American .031 .438 .661 Other .145 1.665 .096
120 Table 4 11 . Continued. Variable Model 1 Model 2 B t p B t p Annual Income $24, 999 or less .015 .218 .828 $50,000 $74,999 .019 .311 .756 $75,000 or more .028 .397 .691 Purchase Groceries Self .043 .260 .795 Spouse .096 1.751 .081 Children .036 .613 .540 Other .007 .100 .920 R 2 .024 .083 F 4.237 .006 2.137 .002 R 2 .059 1.768 .026
121 Table 4 12 . Multiple linear regression analysis for variable s predicting change in attitude (Model 3 and 4). Variable Model 3 Model 4 B t p B t p Constant .769 3.148 .002 .206 .807 .420 Green Giant .234 3.494 .001 .237 3.655 .000 AgLabs .074 1.113 .266 .087 1.356 .176 USDA .133 1.966 .050 .151 2.306 .022 Generation Generation X .007 .102 .919 .023 .362 .718 Young Baby Boomers .001 .013 .990 .041 .576 .565 Old Baby Boomers .094 1.161 .246 .130 1.652 .099 Silent Generation or older .159 2.065 .039 .121 1.621 .106 Men .141 2.880 .004 .193 4.000 .000 Education High School Degree or less .166 2.211 . 027 .156 2.148 . 032 Some college, no degree .028 .440 .660 .056 .885 .377 Two year college degree .008 .104 .917 .017 .243 .808 Graduate or Professional Degree .036 .424 .672 .058 .696 .487
122 Table 4 12 . Continued. Variable Model 3 Model 4 B t p B t p Race African American .028 .398 .691 .019 .276 .783 Other .140 1.616 .107 .112 1.331 .184 Annual Income $24, 999 or less .021 .304 .762 .029 .432 .666 $50,000 $74,999 .023 .374 .709 .034 .571 .568 $75,000 or more .057 .805 .421 .026 .377 .706 Purchase Groceries Self .047 .286 .775 .130 .814 .416 Spouse .080 1.450 .148 .104 1.950 .052 Children .038 .654 .513 .050 .883 .378 Other .010 .140 .889 .007 .105 .917 Prior Knowledge .092 2.452 . 015 .063 1.720 .086 Source Credibility .188 5.910 . 000 R 2 .094 .155 F 2.334 .001 3.905 .000 R 2 .011 .060 6.010 .015 34.926 .000 Table 4 13 . Multiple linear regression analysis for variable s predicting change in risk perception . Variable B t p Constant .166 5.840 .000 Green Giant .012 .293 .770 AgLabs .014 .345 .730 USDA .076 1.853 .064 R 2 .008 F 1.362 .254
123 Table 4 14 . Post hoc analysis for ch ange in risk perception . Variable B t p Constant .256 1.551 .121 Generation Generation X .014 .327 .744 Young Baby Boomers .054 1.157 .248 Old Baby Boomers .042 .814 .416 Silent Generation or older .037 .757 .449 Men .065 2.072 .039 Education High School Degree or less .078 1.653 .099 Some college, no degree .028 .691 .490 Two year college degree .088 1.869 .062 Graduate or Professional Degree .061 1.138 .256 Race African American .013 .296 .767 Other .109 2.027 .043 Annual Income $24, 999 or less .048 1.093 .275 $50,000 $74,999 .093 2.434 .015 $75,000 or more .075 1.687 .092 Self .030 .292 .771 Spouse .025 .716 .474 Children .038 1.055 .292 Other .061 1.392 .164 Prior Knowledge .052 2.176 .030 Source Credibility .026 1.235 .217 R 2 .072 F 1.924 .010
124 Table 4 15 . Post hoc analysis for final attitude . Variable B t p Constant .477 1.678 .094 Generation Generation X .160 2.229 .026 Young Baby Boomers .215 2.684 .008 Old Baby Boomers .237 2.682 .008 Silent Generation or older .027 .325 .745 Men .085 1.545 .123 Education High School Degree or less .084 1.021 .308 Some college, no degree .083 1.178 .239 Two year college degree .117 1.453 .147 Graduate or Professional Degree .161 1.729 .084 Race African American .047 .631 .528 Other .054 .585 .559 Annual Income $24, 999 or less .183 2.396 .017 $50,000 $74,999 .089 1.351 .177 $75,000 or more .089 1.158 .248 Self .485 2.710 .007 Spouse .062 1.039 .299 Children .032 .514 .607 Other .100 1.311 .190
125 Table 4 15 . Continued. Variable B t p Prior Risk Perception .776 18.209 .000 Prior Knowledge .039 .949 .343 Source Credibility .423 8.506 .000 R 2 .746 F 68.845 .000
126 CHAPTER 5 CONCLUSIONS Overview This study explored how persuasive communication influenced change in attitude and change in risk perception of genetically modified food. The elaboration likelihood conceptual model explaining how message sources affected ch anges in attitude and changes in risk perception. An online survey was distributed to Florida residents. The same message was shown to each respondent in the survey, but different message sources were used. Two government sources (FDA and USDA) and two ind ustry companies (Green Giant and AgLabs) were used as the sources in the four groups. Respondents were randomly assigned to each of the four groups. Chapter 4 showed the majority of the respondents were in the Millennial Generation (30.3%, n = 156), women (51.1%, n = 263), and white (75.1%, n = 387 ). Post stratification weighting methods were used so the sample demographics matched the demographics from the 2010 Florida census, and respondents were randomly assigned to each group. An ANOVA was used to compa re the changes in attitude and changes in risk perception between the source groups, and a hierarchical regression was ran to see how well the variables identified in the conceptual model predicted change in attitude and change in risk perception. Key Find ings Descriptive data from this study showed that respondents agreed they were knowledgeable (prior knowledge) about science and genetically modified food ( M = 3.71, SD = .70 ). Respondents neither agreed nor disagreed that the sources were
127 credible, and th ere was no statistical difference between the four groups. The USDA was viewed as having the highest credibility ( M = 2.93, SD = .76), closely followed by AgLabs ( M = 2.92, SD = .70), and the FDA ( M = 2.91 SD = .76). Green Giant had the lowest credibility score out of the four groups with an overall mean of 2.87 ( SD = .78). The average credibility score between the four groups was 2.91( SD = .75). Change in attitude was also measured in this study, and the group exposed to the message using the FDA as the source showed the greatest change in attitude ( M = .40, SD = .53). Green Giant was associated with the smallest change in attitude ( M = .17, SD = .54), and the overall mean for change in attitude was .28 ( SD = . 54). Overall, the prior attitude was slightly negative ( M =2.33, SD = 1.07) and changed to a neutral final attitude ( M = 2.61, SD = 1.10) after exposure to the persuasive communication. Change in risk perception was the final dependent variable in this stu dy, in which lower scores indicated higher perceived risk, while higher scores indicated lower risk perception. Prior to receiving the persuasive communication, respondents neither agreed nor disagreed about the risks associated with genetically modified f ood ( M = 2.83, SD = .89). The final risk perception showed little change ( M = 2.97, SD = .89) after the treatment. Between the four groups, the FDA message source was associated with the greatest change in risk perception ( M = .16, SD = .33). The smallest change in risk perception came from the USDA, which only changed risk perceptions by .09 ( M = .09, SD = .29). Overall, the grand mean for risk perception changed by .14 ( M = .14, SD = .33). This study examined four objectives, the first of which determined if there were any differences in the change in attitude associated with message sources. An ANOVA
128 showed that there were significant differences, specifically between Green Giant and the FD A. The change in attitude for Green Giant was .23 lower than the change associated with the FDA. This meant the FDA was able to create a more positive change in attitude when compared to an industry source like Green Giant. The second objective in the stud y looked to see if there were any differences between message groups for the change in risk perception. The ANOVA was not significant, which meant the message sources were not associated with any differences in change in risk perception. Objective three in the study sought to further explore change in attitude after receiving persuasive communication about genetically modified food. A hierarchical regression was ran to see how well message source, demographics, prior knowledge, and source credibility could predict a change in attitude. The fourth model included all predictors and could account for 15.5% of the variance in change in attitude (R 2 = .155), which was the highest R 2 out of the four models. The use of Green Giant or USDA as sources were significan t predictors of change in attitude, as were men, respondents purchasing food for a spouse, and having a high school diploma. Green Giant and USDA were predicted to have smaller changes in attitude compared to the FDA, men had smaller changes in attitude co mpared to women, and respondents with a high school diploma were predicted to have smaller changes in attitude than those with a four year college degree. Also, respondents who purchased food for a spouse were predicted to show a larger change in attitude than those who did not. Even though prior knowledge was a significant predictor in model three, the addition of source credibility eliminated its significance. Source credibility was a significant predictor though, and for
129 every one unit increase in source credibility, there was a .188 increase in change in attitude ( B = .188). The fourth objective used a regression similar to objective three, but with change in risk perception as the dependent variable. The first model was not significant, concluding that the source alone was not a good predictor for the change in risk perception. The remaining models were not tested since the first was not significant. Implications There are various theoretical and practical implications that can be made from this study. Results from this research can provide further insight into the ELM and the Shannon and Weaver communication model. Practical implications can also be made from these results. Theoretical Implications This study offered greater insight into how attitudes a nd risk perceptions are influenced by persuasive communication. The ELM shows that when motivation or knowledge is low, people will use the peripheral pathway to assess a message (Petty et al., 2009). This route relies on peripheral cues, such as sources. Data from this research showed that there were differences in attitude change associated with different message sources. The source identified as the least credible (Green Giant) showed a significantly lower change in attitude when compared to a source wit h higher credibility not the peripheral cue was operating correctly. If the cue was not effective, then no attitude change would occur according the ELM (Figure 2 2). A dditionally, since Green Giant had a lower credibility score than the other sources, the respondents may have actually given more consideration to the message (Frewer et al., 1997). If more thought
130 was used to analyze the message from Green Giant, a change in attitude may not have been seen because When examining the regression model for change in attitude, the effect of the message source could still be seen. Compared to the FDA, Green Giant predicted a smaller change in attitude, which supported findings from the ANOVA. Additionally, the was higher than that of the FDA. This may be because the USDA was viewed generally as a credible source, but lost its effect when com municating about genetically modified food. The model did show that as source credibility increased, the change in attitude increased. This was consistent with previous research showing that high credibility sources were associated with larger changes in a ttitude compared to low credibility sources (Hovland & Weis, 1951). The fact that prior knowledge was no longer significant in the presence of source credibility indicated that respondents did not have either the motivation to process the information, or that another factor inhibiting the ability to process (e.g. distraction, lack of repetition, etc.) was present, and respondents used the peripheral processing route to assess the message (Petty et al., 2009). This finding supports literature which has show n that most consumers use the peripheral pathway when assessing food communication (Frewer et al., 1997; Goodwin, 2013; Meyers, 2008). The message source and source credibility would likely not be predictive of attitude change if the central processing rou te was used. In the third model, prior knowledge had an influence on the change in attitude in the absence of source credibility. As knowledge increased, the change in attitude
131 ability to process information (Petty & Cacioppo, 1986). In the presence of knowledge, peripheral cues will not be as effective, but if the message does not elicit more positive or negative thoughts or if there was no change in the cognitive structure of the lack of attitude change may explain why the change in attitude became smaller as prior knowledge increased. These results supported previous findings that an incre ase in knowledge does not necessarily mean an increase in positive perceptions toward genetically modified food (McFadden & Lusk, 2015; Verdurme & Viaene, 2003 ). When change in risk perception was analyzed, there were no differences between the message gr oups. Additionally, the descriptive statistics showed that respondents neither agreed nor disagreed about risks associated with genetically modified food for both prior and final risk perceptions, which indicated there was no practical change in risk perce ption. Additionally, the regression model for change in risk perception which used the message source alone as a predictor was not significant. Based on the ELM, it was apparent that prior perceptions of risk were retained (Petty et al., 2009); however, fu rther conclusions could not be made about how people processed the information regarding risks. Changes in risk perception did not seem to be operating within the ELM or the conceptual model developed for the study. This study also supported Shannon and We message source was used as noise in the conceptual model; it could distort the intended message (Lee & Baldwin, 2004) . The results from the ANOVA of change in attitude showed that there were differences in change in attitude between the sources
132 change in attitude, indicating the message was interpreted differently between the groups. However, the noise did not appear to matter when discussing a cha nge in risk perception. There may also be other distractions or noise that could have distorted the message when communicating about risks (Lee & Baldwin, 2004). The decoding process of the communication model was also explored in this study. Demographics were important predictors for attitude change. Source credibility was also significant in the decoding process for change in attitude, but prior knowledge was not. Since the first regression model for risk perception was not significant, and subsequent mod els were not tested, no conclusion could be made about the decoding process. Practical Implications Consumers have had limited knowledge concerning genetically modified food and have relied on various communications to provide trustworthy information (Dura nt et al., 1998; Earle & Cvetkivich, 1995). In the past, the agriculture industry has not been open when communicating to the public, which has led to distrust amongst consumers (McCullum Gomez & Palmer, 2010). Since the success of new technology is often dependent on consumer acceptance ( MacFie , 2007), further investigation into how consumers form opinions toward genetically modified food was necessary to develop strategic communication plans. This study looked at message sources specifically since the pub lic has had to rely on outside communication for information on genetically modified food (Durant et al., 1998; Earle & Cvetkivich, 1995). This research found that message source was associated with differences in change in attitude exhibited by the respo ndents; however, the difference in change in
133 attitude between the groups was relatively small and held little practical implications. Two government (FDA and USDA) and two industry companies (Green Giant and AgLabs) were used as the message sources in this study. Research has shown that the government has typically been more trusted than producers of genetically modified seeds (Irani et al., 2001), but an ANOVA showed no statistical differences between the message sources for source credibility. This findin g conflicts with previous literature concluding that the FDA was more trusted than the USDA, and both were more trusted than industry companies (Irani et al., 2001). Trust is only one component of credibility (Perloff, 2008), which may explain why the FDA was not viewed as being the most credible source since goodwill and expertise were also considered for credibility. Additionally, the topic of genetically modified food may have generated such strong attitudes by the respondents that all sources were perce ived similarly for delivering the message. The FDA proved to be more persuasive compared to Green Giant since there were significant differences between the change in attitude between shown in an ANOVA and Bonferroni test. Even though there was a statisti cal difference between the FDA and Green Giant, there were a few considerations that needed to be made when interpreting the data. The first was that the mean difference between the two sources was smaller than the standard deviation for change in attitude , which indicated that the difference may not have been that large in a practical discussion. Additionally, the group who was assigned to the FDA had the lowest prior attitude and final attitude toward large change in attitude may be a
134 result of their negative prior attitude rather than the effect of the FDA (Frewer et al., 1997). When examining prior knowledge in the regression models for change in attitude, it was significant until the addition of sou rce credibility. The credibility of the source was genetically modified food. Also, an increase in prior knowledge was associated with a decrease in attitude change. This m ay have occurred due to the moral implications of using genetic technology as seen in earlier studies (Evans & Durant, 1995). Demographics also played a role in the change in attitude exhibited by the ally modified food exhibited by demographic groups has been seen previously ( Antonopoulou et al., 2009; Hall & Moran, 2006; Irani et al., 2001; Pounds, 2014; Verdurme & Viaene, 2003), but this study further explored how well the demographics could predict changes in attitude or risk perception as a result of persuasive communication about genetically modified food. When discussing changes in attitude, men generally had smaller changes compared to women. Literature has shown that men typically hold more pos itive views toward genetically modified food (Irani, 2001; Pounds, 2014, Verdurme & Viaene, 2003), but this research demonstrated men were not as greatly influenced by persuasive communication as women. Women may have had the motivation or ability to proce ss the information more so than men, or the peripheral cue may have greater influence on Having a high school diploma or less was also a significant predictor of change in attitude compared to those with a four year college deg ree. Respondents with a high
135 school diploma or less had a smaller change in attitude comparatively. This may have occurred because consumers with a higher education typically have had more negative attitudes related to genetically modified food ( Hall & Mor an, 2006 ; Gaskell, 2003; Moon & Balasubramanian, 2001 ), which allowed them to demonstrate a greater change in attitude (Frewer et al., 1997). Respondents who purchased food for a spouse were the final demographic predictor for change in attitude. This char acteristic was not specifically discussed in Chapter 2, and this finding indicated that demographic change in attitude. Since the statistical analyses performed for chang e in risk perception were not significant, few practical implications could be made. The descriptive statistics showed that respondents neither agreed nor disagreed about the risk genetically modified food posed to consumers, the environment, and the world . Additionally, the persuasive communication did not increase or decrease the risk perceptions. Post hoc analyses were required for further conclusion to be made about changes in risk perception. Limitations This research provided further insight into how persuasive communication genetically modified food, but there are limitations to the research. One of the first limitations of the study was that the instrument did not inclu de a manipulation check. A manipulation check is used to ensure that respondents actually saw whatever treatment was used in the experiment. In the absence of this check, this study had to assume respondents saw the source used to deliver the message.
136 Ther e were also limitations associated with the method data was collected. Since prior knowledge was self perceived prior knowledge compared to their actual knowledge. Actual prior knowledge could have be en more predictive of a change in attitude or change in risk perception compared to the perceived prior knowledge. Additionally , the study sought to measure a change in attitude and a change in risk perception. Since the variables were determined using a pretest posttest design, there was the possibility of pretest treatment interaction (Ary et al., 2010). The effect of pretest tr eatment interaction may have caused minimal changes from the posttest and pretest responses due to prior exposure. This interaction may explain why the changes in attitude and risk perception were so marginal in this study. There was also not a substantial amount of time between the pretest and posttest. A lack of attitude change may be a result of limited time to process the information since a change in attitude requires multiple exposures to a message over time (Perloff, 2003; Petty et al., 2009). Altern atively, an attitude change may have just Conclusions from this study were limited to the sources used and cannot be generalized to all sources used to communicate abo ut genetically modified food. Alternative government and industry sources may have had a different effect on changes in attitude and risk perception than organizations used in the study. Another limitation for this study was that it only examined message s ource as a peripheral cue. Other peripheral cues, such as number of arguments, could have affected the dependent variables (Petty et al., 2009). Similarly, questions asked throughout the instrument may have also been
137 they were prompted with a number of food safety related questions before completing this part of the questionnaire. Recommendations Future Research This study supported that consumers used the peripheral pathway of the ELM when forming attitudes toward ge netically modified food, similar to other agricultural studies (Frewer et al., 1997; Goodwin, 2013; Meyers, 2008). To gain a better understanding of the pathway used when presented with a message about genetically modified food, researchers should utilize thought listing procedures to gain a greater understanding for how consumers process these messages (Petty et al., 1993). Prior knowledge was measured in this study, but relevance/motivation to process was not. Gathering information on these variables will give a more holistic understanding for how consumers move through the ELM when assessing information regarding genetically modified food (Petty et al., 2009). Also, not using a pretest posttest design would eliminate issues associated with pretest treatme nt interactions (Ary et al., 2010). Instead, only gathering attitude and risk perception data after exposure to a message could provide results more representative of final attitude and risk perception. When asked about changes in risk perception, it was n ot as clear which pathway respondents used when forming perceptions of risk. Further research specifically addressing risk communication needs to be conducted to identify how consumers are assessing associated risks with genetically modified food or agricu ltural technology in general, especially since the regression model did not account for much variance for changes in risk perception. Alternatives to the ELM should be explored since the conceptual model did not appear to accurately predict changes in risk perception. The
138 possibility exists that a new theoretical framework is needed to explain risk perceptions related to morally contentious issues. The ELM does aide in understanding shifts in attitude when communicating about genetically modified food. The peripheral pathway appeared to be used, but there is room for further research. This study only provided the name of the organization for the experiment. Adding a brief description of the organization, the brand logo, or organizational values may yield di fferent results and provide greater understanding for how peripheral cues operate. Other sources could also be explored to see their affect as well. For instance, a popular blogger, politician, or restaurant chain may prove to yield different results from this study and give a greater understanding of the influence of message sources. Additionally, collecting source credibility data before exposure to the message may give a more realistic understanding of the variable since the message itself may have influ enced perceived source credibility for this research. Sources are only one element of peripheral cues, and others should be studied as well. Researchers should utilize qualitative or mixed methods strategies to examine the quality and quantity of arguments , or imagery associated with a message. This information would give researchers a greater understanding of how consumers use the peripheral pathway to assess persuasive communication. The affect of prior knowledge on both attitude and risk perception shou ld also be further examined. Since this study used a self reported assessment, an actual useful. Also regarding prior knowledge, research should explore best practices for communicating factual information to the public. Since a greater level of prior knowledge
139 was not associated with more positive views of genetically modified food, agricultural communicators need to realize that an increase in knowledge will not increase acceptance of genetically modified food (McFadden & Lusk, 2015; Verdurme & Viaene, 2003). Discussing concerns and values related to the technology could prove more effective in changing attitudes and risk perception than simply stating statistics and facts . Replacing the message in this study with a value driven message could provide needed insight into attitude formation, and have a stronger influence on changes in attitude and risk perception ( Krause et al., 2015 ) . Further research is needed to evaluate h ow consumers form attitudes toward morally contentious science and technology issues to better understand how to communicate with the public about these topics. Research should explore what platforms consumers use and trust the most (e.g. social media, we b, television, or print), both for seeking information about their food, as well as general information. Understanding how these platforms are used will allow communicators to strategically place messages for effective communication. There is also the poss ibility that the platform itself could serve as a peripheral cue and influence attitude change. Different types of appeals should also be tested to see which ones best facilitate changes in attitude and risk perception when communicating information on a c ontentious topic, such as genetically modified food. This will be important for future communication development since only informing people about the topic does not result in the desired positive attitudes (McFadden & Lusk, 2015; Verdurme & Viaene, 2003). The Shannon and Weaver communication model can also be used to examine other sources of noise that distract the receiver from the intended message (Lee &
140 Baldwin, 2004). Genetically modified food has been surrounded by debate, and the different sources of noise should be explored. Media coverage, personal values, and conflicting scientific findings may prove to distort messages about genetically modified food. Identifying sources of noise can help communicators develop messages which would lessen the degre e of distortion of the intended message. A content analysis of who is currently communicating about genetically modified food, along with how successful their communication is, would be insightful to how the agricultural industry is communicating about genetically modified food compared to opponents. Understanding who has been communicating about the technology could help make stronger connections to the data found in this study about source credibility and give agricultural communicators ideas for colla borations, which would have a positive impact for the industry. Demographic relationships with attitude and risk perception change should also be further explored. Specifically, value based demographics should be assessed in addition to the general demogra phics used in this study. Examining where respondents were raised, their political or religious beliefs, and even the type of diet they follow could give a greater insight into how they interpret a message. This study could be replicated on a larger scale to see if there were regional differences when assessing the credibility of different sources and determine if differences in attitudes and risk perceptions were present in different areas of the United States and the world. This research should also be r eplicated with other contentious agricultural and natural resource issues, such as irradiation of food or fracking. However, if this study were to be replicated, manipulation checks should be used to
141 ensure respondents viewed and paid attention to the trea tment. The methods used in this study could also be expanded outside of the agricultural industry to the larger scientific community. Consumer skepticism is not contained to agriculture alone, and risk and attitude formation needs to be further explored re garding scientific advancements if new technologies are to succeed (MacFie, 2007). Industry and Practitioners This study provided valuable insight into how consumers form attitudes and risk perceptions after receiving persuasive communication about genetic ally modified food. One of the most significant findings was the model for change in attitude and change in risk perception should be different. Changes in attitude were more dependent on source credibility than prior knowledge, and conclusions about chang e in risk perception could not be made. Agricultural communicators need to consider this when developing messages or persuasive communication campaigns. Using a highly credible source would help create larger changes in attitude. Even though AgLabs has had minimal press coverage, it was viewed just as credible as the FDA and USDA, and Green Giant did not have a statistically different credibility score either. Communicators and extension agents should understand that the message sources were perceived simil arly by respondents when communicating about genetically modified food. However, source credibility was a significant predictor of change in attitude. The target audience should be considered when selecting a source for the information since people may vie w the credibility of the same source differently, and using a distrusted source could lead to undesired effects. For instance, producers may view certain industry sources as credible while consumers do not.
142 Agricultural communicators, and the agricultural industry in general, need to recognize that increasing consumer knowledge will not necessarily increase favorable attitudes toward genetically modified food. Attitude formation consists of more than only gaining more knowledge, and cultural values and demo graphics also need to be accounted for. For example, women typically hold more negative attitudes toward genetically modified food (Irani, 2001; Pounds, 2014, Verdurme & Viaene, 2003), but men showed a smaller change in attitude when compared to women. Cha nges in attitude were positive for both genders, and communication should target women in order to increase their attitudes toward genetically modified food. Since consumers who purchased food for a spouse were predicted to have a larger change in attitude than those who did not, creating family focused communication about genetically modified food could be a successful strategy. Additionally, placing advertisements in magazines targeting housewives would be good platforms to reach the intended audience. H aving a high school diploma or less was also a significant predictor of change in attitude; however, this group only accounted for a small portion of the population and practical recommendations cannot be made until further research is done. Communication campaigns should not produce one message and expect it work universally for the population. The differences in demographic characteristics illustrated how different communication strategies will be needed for different population segments. The influences o n changes in risk perception were different than changes in attitude. The results were inconclusive as to what route respondents used when assessing risks, and the ELM may not have been used at all. Different considerations
143 need to be made when communicati ng about risks associated with genetically modified food to the public. Further research was needed before practical recommendations could be made. These recommendations for practitioners could be extended to other contentious issues in science and agricul ture. Post Hoc Key Findings Two post hoc tests were ran to further examine the affect of persuasive communication on attitude and risk perception of genetically modified food. The first test used a regression model to predict change in risk perception with all variables from the conceptual model except for message source. The model was significant, but only accounted for 7.2% of the variance in change in risk perception ( F (3, 511) = 1.924, p = .010, R 2 = .072). Men were predicted to have smaller changes in risk perception compared to women, respondents with a race categorized as other were predicted to have a smaller change than white respondents, and those earning between $50,000 and $74,999 were predicted to have a larger change in risk perception than re spondents who earned $25,000 to $49,999 annually. Additionally, as prior knowledge increased, change in risk perception decreased. Source credibility was not a significant predictor of change in risk perception. A second regression was ran to examine how w ell the variables in the conceptual final attitude toward genetically modified food. Additionally, prior risk perception was added as a moderating variable (Frewer, Howard, & Shepherd, 1998). This model was significant and accounted for 74.6% of the variance ( F (3, 511) = 68.845, p < .000, R 2 = .746). Source credibility and prior risk perception were both significant predictors, and final attitude became more positive as these variables increased. Generation X, Young Baby Bo omers, and Old Baby Boomers were
144 predicted to have a more positive final attitude compared to the Millennial Generation or younger, and respondents earning $24,999 or less a year were predicted to have more positive final attitudes than those earning betwe en $25,000 and $49,999. The final demographic predictor was respondents who purchased food for themselves. These respondents were predicted to have more negative attitudes than those who did not purchase food for themselves. Post Hoc Implications Theoretic al Implications A post hoc analysis of risk perception (message source omitted) showed that some demographic characteristics were predictors, as was prior knowledge, but source credibility was not. The R 2 for this model was small (R 2 = .072), and no furthe r conclusions could be made other than that initial perceptions of risk were retained. In fact, the R 2 was so small, that it supported previous findings in the study that risk perception was not operating within the ELM or the conceptual model. The post ho c analysis examining the final attitude yielded a much higher R 2 ( R 2 = .746) than any of the regression models used to fulfill the objectives. The ELM does illustrate changes in attitude, but this research showed that examining the change in attitude as a dependent variable was not as informative as studying the final attitude when using the model. More positive prior risk perception and source credibility were both predicted to positively increase attitudes toward genetically modified, which aligned with p rior research (Frewer, Howard, & Shepherd, 1998; Frewer et al., 1999). Risk perception may have represented motivation to process information in the ELM due to personal relevance. Respondents who perceived less risks were given higher scores and predicted to have more positive final attitudes toward genetically modified food.
145 Respondents who viewed risk perceptions positively may have not viewed the message as personally relevant and used the peripheral route to process the information. Additionally, prior knowledge was not a significant predictor of final attitude, and source credibility was, which would also indicate that respondents were using the peripheral processing route to assess the information (Petty et al., 2009). This finding supported previous r esearch (Frewer et al., 1997; Goodwin, 2013; Meyers, 2008). Demographics, prior risk perception, and source credibility were identified as important variables used model. Pr actical Implications Regarding change in risk perception, an increase in prior knowledge led to a smaller change in risk perception in the post hoc analysis. Again, this supports previous literature that consumers with a greater understanding of science we re not more likely to accept genetically modified food (McFadden & Lusk, 2015; Verdurme & Viaene, 2003). The findings also showed that source credibility was not a significant predictor of changes in risk perception, which was different from the data exami ning change in attitude. Research had been inconsistent on the effect of income on perceptions of genetically modified food ( Antonopoulou et al., 2009; Verdurme & Viaene, 2003), but the post hoc analysis for this study showed that respondents with an annu al income between $50,000 and $74,999 showed a greater change in risk perception than those in lower income bracket. Finally, differences in attitudes toward genetically modified food have been identified amongst races (Irani et al., 2001). This study iden tified respondents grouped as other to have a smaller change in risk perception compared to
146 white respondents. These differences in demographics could be the result of cultural differences in values (Verdurme & Viaene, 2003). The post hoc regression ran for final attitude also presented valuable practical implications. The first implication was that prior knowledge was not a predictor of final attitude and source credibility was, thus supporting previous literature and regression models in this study (McF adden & Lusk, 2015; Verdurme & Viaene, 2003). Additionally, prior risk perception was a significant predictor. This finding was consistent with prior research (Frewer, Howard, & Shepherd, 1998). Similar to the objectives studied in this research, demograph ic characteristics were important predictors of final attitude. Most importantly, older generations, excluding the Silent Generation or older, were predicted to have more positive final attitudes compared to the Millennial Generation or younger, which conf licted with previous literature ( Antonopoulou et al., 2009). Additionally, respondents who purchased food for themselves were predicted to have a more negative final attitude compared to those who did not purchase food for themselves. One of the income cat egories was also a predictor, but did not make up for as large of the population as the previously discussed categories. Post Hoc Recommendations Future Research Risk perception should be studied as a moderating variable for final attitude toward genetica lly modified food rather than a dependent variable. Understanding influences of persuasive communication on risk perception is still important, but so is its influence on attitudes toward genetically modified food. The post hoc analysis also supported that an alternative theory or model should be used when researching risk perceptions specifically.
147 Since respondents who purchased food for themselves was a predictor of final attitude, further research looking at food purchasing and consuming behaviors shoul d be studied to see how well they predict final attitude toward genetically modified food. Additionally, researching final risk perception and final attitude toward genetically modified food would be more beneficial than studying changes when using ELM to guide the study. Industry and Practitioners When communicating about specific risks, using a credible source will have limited effect, and increasing knowledge may decrease changes in risk perception. Discussing concerns of the consumers and framing the m essage based on values could be more effective when discussing risks (Frewer, Howard, & Shepherd, 1998; Krause et al., 2015 ). The post hoc analysis showed that an increase in knowledge led to smaller changes in risk perception. Agricultural communicators s hould focus on the concerns of the consumer (Frewer, Howard, & Shepherd, 1998) or deliver context driven communication (e.g. citrus greening in Florida) to elicit positive final attitudes and perceptions from consumers (Pounds, 2013). Avoiding logic and fa ct based messages may prove to be more effective in evoking positive final attitudes. This study showed that older generations were predicted to have more positive final attitudes toward genetically modified food than Millennials, and people purchasing for themselves had more negative final attitudes than those who did not. Communicators and extension agents should develop different communication campaigns for older and younger consumers. Additionally, appropriate mediums should be used for the targeted age category. For instance, Millennials may be more receptive to receiving communication online or through social media than older generations.
148 Since people who purchased food for themselves were predicted to have more negative final attitudes toward genetica lly modified, point of purchase advertising for genetically modified food should be avoided to keep these consumers from actively thinking about genetically modified food while making purchases. Framing the messages around the direct tangible benefits gene tically modified food offers consumers could make the message more relevant and elicit more positive thoughts. Summary This study sought to determine the affect of persuasive communication on ion of genetically to guide the study. The results indicated that source and source credibility matter much more when examining change in attitude. Change in risk perception yiel ded different results, and conclusions from the objective could not be made. The findings illustrated that the peripheral pathway was likely used by consumers when forming general attitudes, but this was not necessarily the case when discussing perceptions of risk. In fact, changes in risk perception did not appear to operate withi n the ELM. When communication efforts aims to effect general attitudes, care should be taken when selecting information sources. While the source used for the message was identified as important, prior knowledge should also be considered. However, in the p resence of source credibility, prior knowledge was no longer a predictor of attitude change, but it was shown to be associated with a smaller change in attitude. The agricultural and biotechnology industry needs to realize that increasing knowledge alone w ill not lead to increased positive perceptions of genetically modified food. Demographics, and likely personal values, also played a role in how effective
149 communication can be on a topic like genetically modified food. Post hoc analysis used prior risk per ception as a moderating variable in a regression model to predict final attitude. The model was concluded to be a much better fit that the ones used to fulfill the objectives. As perceptions of risk perception increased favorably, the final attitude was pr edicted to become more positive. Further research is needed to explore additional influences on risk perception and attitudes to aid agricultural communicators in developing effective communication campaigns.
150 APPENDIX A IRB APPROVAL
154 A PPENDIX B INSTRUMENT USED FOR STUDY
156 Note. This example shows the FDA, but respondents were randomly assigned to the FDA, USDA, Green Giant, or AgLabs.
157 Note. This example shows the FDA, but respondents were randomly assigned to the FDA, USDA, Green Giant, or AgLabs.
158 APPENDIX C EXPERIMENTAL TREATMENT
160 APPENDIX D COMPLETE SURVEY INSTRUMENT
168 Note. Respondents were randomly shown one of these two messages .
188 REFERENCES Abbott, E. A., Lucht, T., Jensen, J. P., & Jordan Conde, Z. (2001). Handling of GM crops issues by the mass media . Paper presented at the Illinois Crop Protection Technology Conference, Chicago, IL. Retrieved from http://ipm.illinois.edu/education/proceedings/icptcp2001.pdf#page =35 Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes , 50 , 179 211. doi:10.1016/0749 5978(91)90020 T Ajzen, I., & Fishbein, M. (1977). Attitude behavior relations: A theoretical analysis and review of em pirical research. Psychological Bulletin , 84 , 888 918. doi:10.1037//0033 2909.84.5.888 Alport, G. W. (1935). Attitudes. In C. Murchinson (Ed.), A handbook in social psychology (pp. 798 844). Worcester, MA: Clark University Press. Anderson, S., Ruth, T., & Rumble, J. (2014). Public opinion of food in Florida . PIE2011/12 17 Gainesville, FL: University of Florida/IFAS Center for Public Issues Education. Retrieved from http://www.piecenter.com/wp content/uploads/2014/12/Food Panel Report_2014_Final_4.pdf Antono poulou, L., Papadas, C. T., & Targoutzidis, A. (2009). The impact of socio demographic factors and political perceptions on consumer attitudes towards genetically modified foods: An econometric investigation. Agricultural Economics Review , 10 (2), 89 103. Retrieved from http://www.eng.auth.gr/mattas/10_2_7.pdf Ary, D., Jacobs, L. C., & Sorensen, C. (2010). Introduction to research in education (8 th ed). United Kingdom: Wadsworth Cengage Learning. Axsom, D., Yates, S., & Chaiken, S. (1987). Aud ience response as a heuristic cue in persuasion. Journal of Personality and Social Psychology, 53 , 30 40. doi:10.1037//0022 35126.96.36.199 (2013). Summary report of the a apor task force on non probability sampling. Journal of Survey Statistics and Methodology , 1 (2), 90 143. doi: 10.1093/jssam/smt008 Barnett, J., Cooper, H., & Senior, V. (2007). Belief in public efficacy, trust, and attitudes toward modern genetic science. Risk Analysis , 27 , 921 933. doi:10.1111/j.1539 6924.2007.00932.x Baron, R. M., & Kenny, D. A. (1986). The moderator mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personalit y and Social Psychology , 51 (6), 1173 1182. doi:10.1037/0022 35188.8.131.523
189 Beharrell, B., & Denison, T. J. (1995). Involvement in a routine food shopping context. British Food Journal, 97 (4), 24 29. doi:10.1108/00070709510085648 Bem, D. J. (1964). Self pe rception theory. In L. Berkowitz (Ed.), Advances in experimental social psychology (pp. 1 62). New York, NY: Academic Press. Benbrook, C. M. (2012). Impacts of genetically engineered crops on pesticide use in the United States: The first eight years. Envi ronmental Sciences Europe , 24 (24), 1 13. Retrieved from http://www.enveurope.com/content/24/1/24 Black, R. E., Vistora, C. G., Walker, S. P., Bhutta, Z. A., Christian, P., Onis, M. D., . . . Uauy, R. (2013). Maternal and child undernutrition and overweight in low income and middle income countries. The Lancet , 382 (9890), 427 451. Retrieved from http://www.sciencedirect. com/science/article/pii/S014067361360937X Bove, J. M. (2012). Huanglongbing and the future of citrus in Sao Paulo State, Brazil. Journal of Plant Pathology , 94 (3), 465 467. Retrieved from http://www.cabi.org/cabdirect/FullTextPDF/2013/20133025591.pdf Bowerm an, B. L., & O'Connell, R. T. (1990). Linear statistical models: An applied approach . Belmont, CA: Duxbury Press. Bredahl, L. (2001). Determinants of consumer attitudes and purchase intentions with regard to genetically modified food Results of a cross n ational survey. Journal of Consumer Policy , 24 , 23 61. doi:10.1023/A:1010950406128 Brinol, P., Petty, R. E., & Tormala, Z. L. (2004). Self validation of cognitive responses to advertisements. Journal of Consumer Research, 30 , 559 573. doi:10.1086/380289 Brock, T. C. (1967). Communication discrepancy and intent to persuade as determinants of counterargument production. Journal of Experimental Social Psychology , 1 , 650 654. doi:10.1016/0022 1031(67)90031 5 Cacioppo, J. T., Petty, R. E., & Sidera, J. (1982). The effects of a salient self schema on the evaluation of proattitudinal editorials: Top down versus bottom up message processing. Journal of Experimental Social Psychology, 18 , 324 338. doi:10.1016/0022 1031(82)90057 9 Caffrey, A. (2014, May 21). March a gainst Green Giant to shout "Hell no to GMO" for second year in a row. San Antonio Current [San Antonio]. Retrieved from http://sacurrent.co m/news/march against Green Giant to shout hell no to gmos for second year 1.1689580 Campbell, D. T., & Stanley, J. C. (1963). Experimental and quasi experimental designs for research . Chicago, IL: Rand McNally.
190 Carlson, F., Frykblom, P., & Lagerkvist, C. J . (2007). Consumer benefits and bans on genetically modified food An empirical analysis using choice experiment. American Journal of Agricultural Economics , 89 (1), 152 161. Retrieved from https://ideas.repec.org/p/hhs/gunwpe/0129.html Center for Disease C ontrol. (2001). CDC report to FDA: Investigation of human health effects associated with potential exposure to genetically modified corn . Retrieved from http://www.cdc.gov/nceh/ehhe/Cry9cReport/pdfs/cry9creport.pdf Chaiken, S. (1987). The heuristic model of persuasion. In M. Zanna, J. Olsen, & C. Herman (Eds.), Social influence: The Ontario symposium (pp. 3 39). Hillsdale, NJ: Erlbaum. Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and systematic informa tion processing within and beyond persuasion context. In J. Uleman, & J. Bargh (Eds.), Unintended thought: Limits of awareness, intention, and control (pp. 212 252). New York, NY: Guilford Press. Chaiken, S., Wood, W., & Eagly, A. H. (1996). Principles of persuasion. In E. Higgins, & A. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 702 742). New York, NY: Guilford Press. Chassy, B. (2003). The role of agricultural biotechnology in world food aid . Retrieved May 19, 2014, from http: //www.4uth.gov.ua/usa/english/tech/ijee0903/chassy.htm Chassy, B. M. (2007). The history and future of GMOs in food and agriculture. Cereal Foods World , 52 (4), 169 172. doi:10.1094/CFW 52 4 0169 Chaussee, J. (2014, May 29). California GMO labeling push fai ls second time after lawmakers reject bill. Huffington Post . Retrieved from http://a gmo bill_n_5408147.html Chen, S., & Chaiken, S. (1999). The heuristic systematic model in its broader context. In S. Chaiken, & Y. Trope (Eds.), Dual process theories in s ocial psychology (pp. 73 96). New York, NY: Guilford Press. Chetelat, R. T., DeVerna, J. W., & Bennett, A. B. (1995). Introgression into tomato (Lycopersicon esculentum ) of the L. chmielewskii sucrose accumulator gene (sucr ) controlling fruit sugar compo sition. Theoretical and Applied Genetics , 91 , 327 333. doi:10.1007/BF00220895 Chipotle. (2015, April 27). Press release Chipotle becomes the first national restaurant company to use only non GMO ingredients. Retrieved from http://ir.chipotle.com/phoenix.z html?c=194775&p=irol newsArticle&ID=2040322 Chipotle. (n.d.). Chipotle: Videos [Video file]. Retrieved from http://www.chipotle.com/en us/fwi/videos/videos.aspx?v=1
191 Cho, M. L., Kim, H. K., Choi, H. W., Buchanan, B. B., & Lemaux, P. G. (2000). Endosperm spe cific GFP expression driven by barley D hordein promoter and its inheritance in transgenic barley and wheat plants. In Vitro Cellular and Developmental Biology Animal , 36 , A36. Retrieved from http://eurekamag.com/research/034/851/034851245.php#close Condit , C. M. (2010). Public attitudes and beliefs about genetics. The Annual Review of Genomics and Human Genetics , 11 , 339 359. doi:10.1146/annurev genom 082509 141740 Cook, T. D., & Campbell, D. T. (1979). Quasi experimentation: Design and analysis issues for field settings . Chicago, IL: Rand McNally. Dillman, D., Smyth, J., & Christian, L. (2009). Turbulent times for survey methodology. In D. Dillman (Ed.), Internet, mail, and mixed mode surveys. (3rd ed., pp. 1 228). Hoboken, NJ: John Wiley & Sons, Inc. Doer ing, C. (2014, March 3). Chipotle's farm satire upsets agriculture industry . Retrieved June 2, 2014, from http://www.usatoday.com/story/news/politi cs/2014/03/03/chipotle show angers farmers/5874435/ Dona, A., & Arvanitoyannis, I. S. (2009). Health risks of genetically modified foods. Critical Reviews in Food Science and Nutrition , 49 (2), 164 175. doi:10.1080/10408390701855993 Dubois, S. (2014, May 21 ). Jackson County, Oregon, approves GMO ban ABC News . Retrieved June 2, 2014, from http://abcnews.go.com/Health/wireStory/jackson county oregon approves gmo ban 23812490 Durant, J., Bauer, M.W., & Gaskell, G. (1998). Biotechnology in the public sphere . London: Science Museum. Earle, T. C., & Cvetkovich, G. T. (1995). Social trust: Toward a cosmopolitan society . Westport, CT: Praeger. Ells, S. (2014, January 28). Chipotle is saying no to GMOs. Huffington Post . Retrieved from http://www.huffingtonpost.c om/steve ells/chipotle gmos no_b_4063994.html Evans, G., & Durant, J. (1995). The relationship between knowledge and attitudes in the public understanding of science in Britain. Public Understanding of Science , 4 , 57 74. Retrieved from http://pus.sagepub.com/content/4/1/57.full.pdf+html Fazio, R. H. (1989). On the power and function of attitudes: The role of attitude accessibility. In A. Pratkanis, S. Breckler, & A. Greenwald (Eds.), Attitu de structure and function (pp. 153 179). Hillsdale, NJ: Lawrence Erlbaum Associates.
192 Fazio, R. H., & Zanna, M. P. (1981). Direct experience and attitude behavior consistency. Advances in Experimental Social Psychology , 14 , 162 202. doi:10.1177/014616727800400109 Fernandez Cornejo, J., Wechsler, S., Livingston, M., & Mitchell, L. (2014). Genetically engineered crops in the United States (Report No. 162). Retrieved from U.S. Dept. of Agriculture, Economic Research Service website: http ://www.ers.usda.gov/media/1282246/err162.pdf Field, A. (2013). Discovering statistics using IBM SPSS statistics (4th ed.). Thousand Oaks, CA: SAGE. Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and r esearch . Reading, MA: Addison Wesley Pub. Co. Florida House of Representatives. (2015). SB 416 Labeling of genetically engineered foods. Retrieved from http://www.myfloridahouse.gov/Sections/Bills/billsdetail.aspx?BillId=53440& Folta, K. (2012). "Talk n erdy to me" about GMO food & science A colorful adventure [Video file]. Available from http://janiceperson.com/agriculture/ag awareness/gmo food scient ists talk nerdy to me/ Food and Agriculture Organization of the United Nations. (2015). The state of food insecurity in the world 2015 . Retrieved from http://www.fao.org/hunger/en/ Food and Drug Administration. (2014). Genetic Engineering . Retrieved from http://www.fda.gov/AnimalVeterinary/DevelopmentApprovalProcess/GeneticEngi neering/ Frewer, L. J., Howard, C., & Aaron, J. I. (1998). Consumer acceptance of tr ansgenic crops. Pesticide Science , 52 , 338 393. doi:10.1002/(SICI)1096 9063(199804)52:4<388::AID PS740>3.0.CO;2 F Frewer, L. J., Howard, C., Hedderley, D., & Shepherd, R. (1997). The elaboration likelihood model and communication about food risks. Risk Analysis , 17 (6), 759 769. doi:10.1111/j.1539 6924.1997.tb01281.x Frewer, L., Howard, C., Hedderley, D., & Shepherd, R. (1999). Reactions to information about genetic engineering: Impact of source characteristic, perceived personal relevance, and persuasive ness. Public Understanding of Science, 8 (35), 35 50. doi:10.1088/0963 6625/8/1/003 Frewer, L., Howard, C., & Shepherd, R. (1998). The influence of initial attitudes on responses to communication about genetic engineering in food production. Agricultural and Human Values , 15 , 15 30. Retrieved from http://link.springer.com/article/10.1023/A:1007465730039
193 Frewer, L., Scholderer, J., & Bredahl, L. (2000). Communicating about the risks and benefits of genetically modified foods: Effects of different informatio n strategies . Risk Analysis, 23 ,6, 1117 1139 Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14641888 Funk, C., Raine, L., Smith, A., Olmsted, K., Duggan, M., & Page, D. (2015). Public and scien tists' views on science and society . Retrieved from Pew Research Center website: http://www.pewinternet.org/files/2015/01/PI_ScienceandSociety_Report_012915. pdf Gaskell, G. (2003). Europeans and biotechnology in 2002: A report to the EC Directorate General for Research from the project 'Life Sciences in European Society' (Report No. QLG7 CT 1999 00286). Retrieved from The Directorate website: http://ec.europa.eu/public_opinion/archives/ebs/ebs_177_en.pdf George, D., & Mallery, M. (2010). Using SPSS for Windows step by step: A simple guide and reference . Boston, MA: Allyn & Bacon. Gerland, P., Raftery, A. E., Secviklova, H., Li, N., Gu, D., Spoorenberg, T., . . . Wilmoth, J. (2014). World population stabilization unlikely this century. Science , 356 (6306), 234 237. doi:10.1126/science.1257469 GMO Answers. (2014). What is a GMO? An introduction | GMO Answers . Retrieved June 1, 2014, from http://gmoanswers.com/explore GMO Inside. (2014). FAQs: GMO inside . Retrieve d May 19, 2014, from http://gmoinside.org/faqs/ Gonsalves, D. (1998). Control of papaya ringspot virus in papaya: A Case Study. Annual Review of Phytopathology , 36 , 165 205. doi:10.1146/annurev.phyto.36.1.415 Gonsalves, D., Ferreira, S., Manshardt, R., Fit ch, M., & Slightom, J. (2000). Transgenic virus resistant papaya: New hope for controlling papaya ringspot virus in Hawaii. Plant Health Progress . doi:10.1094/PHP 2000 0621 01 RV Goodman, D. & DuPuis, E. M. (2002). Knowing food and growing food: Beyond the production consumption debate in the sociology of agriculture. Sociologia Ruralis, 42 (1), 5 22. do i: 10.1111/1467 9523.00199 Goodwin, J. N. (2013). Taking down the walls of agriculture: Effect of transparent communication and personal relevance on attitudes and trust within the Elaboration Likelihood Model (Doctoral Dissertation). Retrieved from http://ufdcimages.uflib.ufl.edu/UF/E0/04/53/18/00001/GOODWIN_J.pdf . Green Giant. (2014). Green Giant at a glance . Retrieved June 2, 2014, from http://www.green giant.com/whoweare/pages/default.aspx
194 Grice, J., & Lawrence, G. (2003). Consumer surveys of biotechnology: Asking the questions until we get the answers we want or empowering the public to express their opinions. Paper presented at the Agrifood Research ne twork Conference, Akaroa, New Zealand. Hall, C., & Moran, D. (2006). Investigating GM risk perceptions: A survey of anti GM and environmental campaign group members. Journal of Rural Studies , 22 , 29 37. doi:10.1016/j.jrurstud.2005.05.010 Hallman, W. K., Hebden, W. C., Aquino, H. L., Cuite, C. L., & Lang, J. T. (2003). Public perceptions of genetically modified foods: A national study of American knowledge and opinion (Report No. RR 1003 004). Retrieved from Food Policy Institute website: h ttp://core.ac.uk/download/pdf/6435317.pdf Hallman, W. K., & Metcalfe, J. (1991). Public perceptions of agricultural biotechnology: A survey of New Jersey Residents . Retrieved from Food Policy Institute, Cook College, Rutgers, The State University of New Je rsey. website: http://www.foodpolicyinstitute.org Henig, R. M. (2000). Monk in the garden: The lost and found genius of Gregor Mendel . New York, NY: Houghton Mifflin. Hinchee, M. A., Connor Ward, D. V., Newell, C. A., McDonnell, R. E., Sato, S. J., Gasser, C. S., . . . Horsch, R. B. (1988). Production of transgenic soybean plants using agrobacterium mediated DNA transfer. Bio/technology , 6 , 915 922. doi:10.1038/nbt0888 915 Hoban, T. J. (1995). The constru ction of food biotechnology as a social issue. In D. Maurer & J. Sobal (Eds.), Eating agendas: Food nutrition as social problems (pp. 189 209). Hawthorne, NY: Aldine de Gruyter. Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953). Communication and persu asion: Psychological studies of opinion change . New Haven, CT: Yale University Press. Hovland, C. I., & Weiss, W. (1951). The Influence of Source Credibility on Communication Effectiveness. Public Opinion Quarterly, 15 , 635 650. doi:10.1086/266350 Huffman, W., Rousu, M., Shogren, J. F., & Tegene, A. (2004). The effects of prior beliefs and learning on consumers' acceptance of genetically modified foods. Journal of Economic Behavior and Organization , 63 (1), 193 206. Retrieved from http://www.sciencedirect.com/science/article/pii/S0167268106000369
195 Whom do you trust? The influence of culture, gender, and g eography on consumer perceptions of GMO labeled products. Paper session presented at the 17th annual conference of the Association of International Agricultural Extension Education, Baton Rouge, LA. Retrieved from https://www.aiaee.org/index.php/proceeding s/127 2001 baton rouge louisiana/1381 whom do you trust the influence of culture gender and geography on consumer perceptions James, C. (2007). Global status of commercialized biotech/GM crops, 2007: Executive summary (35). Retrieved from International Service for the Acquisition of Agri Biotech Applications (ISAAA) website: isaaa.org Kalton, G., & Flores Cervantes, I. (2003). Weighting methods. Journal of Official Statistics , 19 (2), 81 97. Retrieved from http://www.jos .nu/articles/article.asp Kantor, K. (2013, April). GMO's Pros and Cons. CNN iReport. Retrieved from http://ireport.cnn.com/docs/DOC 957532 Karole, G. (2014, May 15). AgLabs VP on using technology to give farmers, consumers more choices . Retrieved June 2, 2014, from http://www.fo xbusiness.com/business leaders/2014/05/15/dupont pioneer vp on using technology to give farmers consumers more choices/ Klumper, W., & Qaim, M. (2014). A meta analysis of the impacts of genetically modified crops. Plos One , 9 (11), e111629. Retrieved from http://www.plosone.org/article/fetchObject.action?uri=info:doi/10.1371/journal.pon e.0111629&representation=PDF Krause, A., Meyers, C., Irlbeck, E., & Chambers, T. (2015). What side are you on? An examination of the persuasive message factors in proposition 37 videos on YouTube. Paper presented at the 2015 American Association for Agricultural Education, San Antonia, TX Langer, G. (20 13). Poll: Skepticism of Genetically Modified Foods. ABC News . Retrieved from http://abcnews.go.com/Technology/story?id=97567 Lee, K. C., & Baldwin, J. R. (2004). History of "speech communication" research: models and messages. In J. Baldwin, S. Perry, & M. Moffitt (Eds.), Communication theories for everyday life (pp. 55 73). Boston, MA: Pearson/Allyn and Bacon. Lee, A., & Lee, E. B. (1939). The fine art of propaganda: A study of Father Coughlin's speeches . New York, NY: Harcourt, Brace. Lemaux, P. G. (200 8). Genetically engineered plants and foods: A scientist's analysis of the issues (Part I). Annual Review of Plant Biology , 59 , 771 812. doi:10.1146/annurev.arplant.58.032806.103840
196 Lewis, P. (1991, June 16). Mutant food create risks we can't yet guess . Re trieved June 2, 2014, from http://www.nytimes.com/1992/06/16/opinion/l mutant foods create risks we can t yet guess 330092.html Lidskog, R. (1996). In science we trust? On the relation between scientific knowledge, risk consciousness and public trust. Acta Sociologica , 39 (1), 31 58. Retrieved from http://www.jstor.org/stable/4194804?seq=1#page_scan_tab_contents Lockie, S., Lawrence, G., Lyons , K., & Grice, J. (2005). Factors underlying support or opposition to biotechnology among Australian food consumers and implications for retailer led food regulation. Food Policy , 30 , 399 418. doi:10.1016/j.foodpol.2005.06.001 Losey, J., Raynor, L., & Carter, M. E. (1999). Transgenic pollen harms monarch larvae. Nature , 399 , 214. doi: 10.1038/20338 Lusk, J. L., Jamal, M., Kurlander, L., Roucan, M., & Taulman, L. (2005). A meta analysis of genetically modified food valuation studies. A Meta Analysis of Ge netically Modified Food Valuation Studies , 30 (1), 28 44. Retrieved from http://ageconsearch.umn.edu/handle/30782 Ma, J. K., Drake, M. W., & Christou, P. (2003). The production of recombinant pharmaceutical proteins in plants. Nature Review Genteics , 4 , 794 804. doi: 10.1038/nrg1177 MacFie, H. J. (2007). Consumer led food product development . Cambridge, UK: Woodhead. Maletta, H. (2007). Weighting. SPSS Tools . Retrieved from http://www.spsstools.net/Tutorials/WEIGHTING.pdf March against Green Giant . (2014, May 28). Retrieved June 2, 2014, from http://www.march against Green Giant .com/ Marks, L. A., Kalaitzandonakes, N., Allison, K., & Zakharova, L. (2002). Time series analysis of risk frames in media communication of agrobiotechnology. In V. Santaniello, R. E. Evenson, & D. Zilberman (Eds.), Market development for genetically modified agricultural products (pp. 217 225). Wallingford, UK: CABI. Marks, L. A., Kalaitzandonakes, N. G., Allison, K., & Zakha rova, L. (2003). Media coverage of agrobiotechnology: Did the butterfly have an effect? Journal of Agribusiness , 21 (1), 1 20. Retrieved from http://ageconsearch.umn.edu/bitstream/14 674/1/21010001.pdf McCroskey, J. C. (1997). An introduction to rhetorical communication (7th ed.). Boston, MA: Allyn and Bacon.
197 McCullum Gomez, C., & Palmer, S. (2010, July). Genetically engineered foods update. Environmental Nutrition , 33 (7). Retrieved fr om http://go.galegroup.com/ps/i.do?id=GALE%7CA230060422&v=2.1&u=gain40375 &it=r&p=AONE&sw=w&asid=68d94544ba5d398152 176e91be2b3d4a McFadden, B. R., & Lusk, J. L. (2015). Cognitive biases in the assimilation of scientific information on global warming and genetically modified food. Food Policy, 54, 35 43. doi:10.1016/j.foodpol.2015.04.010 McGuire, W. J. (1969). The nature of attitudes and attitude change. In S. Fiske, D. Gilbert, & G. Lindzey (Eds.), The handbook of social psychology (2nd ed., pp. 136 314). Reading, MA: Addison Wesley. Menard, S. W. (1995). Applied logistic regression analysis . Thousand Oaks, CA: Sage Publications. Meyers, C. A. (2008). The agricultural angle: Effect of framing agricultural biotechnology messages on attitudes and intent to publish within the Elaboration Likelihood Model (Doctoral Dissertation). Retrieved from htt p://gradworks.umi.com/33/34/3334488.html Miller, G. R. (1980). On being persuaded: Some basic distinctions. In M. Roloff, & G. Miller (Eds.), Persuasion: New directions in theory and research (pp. 11 28). Beverly Hills, CA: Sage Publications. Miller, J. D., Annou, M., & Wailes, E. J. (2003). Communicating biotechnology: Relationships between tone, issues, and technology in U.S. print media coverage. Journal of Applied Communications , 87 (3), 29 40. Moon, W., & Balasubramanian, S.K. (2001). A mult i attribute model of public acceptance of genetically modified organisms. Paper presented at the Annual Meeting of American Agricultural Economics Association. Chicago, IL Morgan, J. M., & Gramann, J. H. (1989). Predicting effectiveness of wildlife educati on programs: A study of students' attitudes and knowledge toward snakes. Wildlife Society Bulletin , 17 (4), 501 509. Retrieved from http://www.jstor.org/stable/3782720 Napier, T. L., Tucker, M., Henry, C., & Whaley, S. R. (2006). Consumer attitudes toward GMOs: The Ohio experience. Journal of Food Science , 69 (3), 79 76. doi:10.1111/j.1365 2621.2004.tb13344.x National Academy of Sciences (1987). Introduction of recombinant DNA engineered organisms into the environment: Key issues . National Academy Press. National Agricultural Statistics Service. (2011). Florida agricultural overview. Retrieved from http://www.nass.usda.gov/Statist ics_by_State/Florida/index.asp
198 National Agricultural Statistics Service. (2012). Florida agricultural overview. Retrieved from http://www.nass.usda.gov/Quick_Stats/Ag_Overview/stateOverview.php?state=F LORIDA National Research Council (1989). Field testing genetically modified organisms: Framework for decisions . National Academy Press. Nelson, C. H. (2001). Risk perception, behavior, and consumer response to genetically modified organisms: Toward understanding American and European public reaction. American Behavioral Scientist, 44 (8) 1371 1388. doi:10.1177/00027640121956737 Nichols, J. (2014, May 21). Beating Green Giant in the food fight: Oregon counties vote to ban GMO crops | The Nation [Web log post]. Retrieved from http://www.thenation.com/blog/179962/beating Green Giant food fight oregon counties vote ban gmo crops Nicolia, A., Manzo, A., Veronesi, F., & Rosellini, D. (2014). An overview o f the last 10 years of genetically engineered crop safety research. Critical Reviews in Biotechnology, 34 (1), 77 88. doi:10.3109/07388551.2013.823595 Osgood, C. E., Suci, G. J. & Tannenbaum, P. H. (1971). The measurement of meaning . Chicago, IL: University of Illinois Press. Paine, J. A., Shipton, C. A., Chaggar, S., Howells, R. M., Kennedy, M. J., Vernon, G., . . . Drake, R. (2005). Improving the nutritional value of Golden Rice through increased pro vitamin A content. Nature Biotechnology , 23 , 482 487. d oi:10.1038/nbt1082 Perloff, R. M. (2008). The dynamics of persuasion: Communication and attitudes in the 21st century (3rd ed.). New York, NY: Routlage. Petty, R. E., Brinol, P., & Priester, J. R. (2009). Mass media attitude change: Implications of the ela boration likelihood model of persuasion. In J. Bryant, & M. Oliver (Eds.), Media effects: Advances in theory and research (3rd ed., pp. 125 164). New York, NY: Routledge. Petty, R. E., & Cacioppo, J. T. (1979). Issue involvement can increase or decrease pe rsuasion by enhancing message relevant cognitive responses. Journal of Personality and Social Psychology , 37 , 1915 1926. Retrieved from http://psycnet.apa.org/doi/10.1037/0022 35184.108.40.2065 Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes to attitude change . New York, NY: Springer/Verlag.
199 Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Central and peripheral routes to advertising effectiveness: The moderating role of involvement. Journal of C onsumer Research , 10 (2), 135 146. doi:10.1086/208954 Petty, R. E., Haugtvedt, C., & Smith, S. M. (1995). Elaboration as a determinant of attitude strength: Creating attitudes that are persistent, resistant, and predictive of behaviour. In R. Petty, & J. Kr osnick (Eds.), Attitude strength: Antecedents and consequences (pp. 93 130). Mahwah, NJ: Erlbaum. Petty, R. E., & Wegener, D. T. (1998). Matching versus mismatching attitude functions: Implications for scrutiny of persuasive messages. Personality and Socia l Psychology Bulletin, 24 , 227 240. doi:10.1177/0146167298243001 Pew Research Center. (2009). Public praises science; Scientists fault public. Retrieved from http://www.people press.org/2009/07/09/public praises science scientists fault public media/ Phill ips, T. (2008). Genetically modified organisms (GMOs): Transgenic crops and recombinant DNA technology. Nature Education , 1 (1), 213. Retrieved from http://www.nature.com/scitable/nated/topicpage/genetically modified organisms gmos transgenic crops nbsp 732 Poortinga, W., & Pidgeon, N. F. (2005). Trust in risk regulation: Cause or consequence of the acceptability of GM food? Risk Analy sis , 25 , 199 209. doi:10.1111/j.0272 4332.2005.00579.x Pounds, K.L. (2014) purchasing intentions, and labeling perceptions of GMOs. University of Florida, Gainesville, Fl Priester, J. R., & Petty, R. E. (1995). Source attributions and persuasion: Perceived honesty as a determinant of message scrutiny. Personality and Social Psychology Bulletin, 21 , 639 656. doi:10.1177/0146167295216010 Priester, J. R., & Petty, R. E. (2003) . The influence of spokesperson trustworthiness on message elaboration, attitude strength, and advertising effectiveness. Journal of Consumer Psychology, 13 , 408 421. Retrieved from https://msbfile03.usc.edu/digitalmeasures/priester/intellcont/2003 JCP PriesterPetty 1.pdf Putnam, A. (2014). Citrus Greening . Florida Farm Bureau Federation . Retrieved from http://www.floridafarmbureau.org/node/3154 . Rabin, R. C. (2014). Information not on the label. The New York Times. Retrieved from http://well.blogs.nytimes.com/2014/05/26/information not on the label/?_r=0
200 Roe, B., & Teisl, M. (2007). Genetically modified food labeling: The impacts of message and messenger on consumer perceptions of labels and products. Food Policy , 32 , 49 66. doi: 10.1016/j.foodpol.2005.12.006 Ronteltap, A., van Trijp, J. C., Renes, R. J., & Frewer, L . J. (2007). Consumer acceptance of technology based food innovations: Lessons for the future of nutrigenomics. Appetite , 49 , 1 17. doi:10.1016/j.appet.2007.02.002 Rothenberg, S., & Becker, M. (2004). Technical assistance programs and the diffusion of envi ronmental technologies in the printing industry: The case of SMEs. Business and Society, 43 , 366 397. doi: 10.1177/0007650304268801 Rumble, J. N., & Leal, A. (2013). Public opinion of food in Florida . PIE2012/13 15 Gainesville, FL: University of Florida/IFAS Center for Public Issues Education. Retrieved from http://www.piecenter.com/wp content/uploads/2014/05/foodreport.pdf Russia Today (2014, May 29). California lawmakers reject GMO labeling bill . Re trieved June 2, 2014, from http://rt.com/usa/162384 california gmo labeling law/ Satran, J. (2013). Citrus greening forces Florida growers to trust a controversial savior. Huffington Pos t . Retrieved from http://www.huffingtonpost.com/2013/08/30/citrus greening_n_3780984.html . Sears, M. K., Hellmich, R. L., Stanley Horn, D. L., Oberhauser, K . S., Pleasants, J. M., Mattila, H. R., . . . Dively, G. P. (2001). Impact of Bt corn pollen on monarch butterfly populations: A risk assessment. Proceedings of t he National Academy of Sciences , 98 (21), 11937 11942. doi:10.1073/pnas.211329998 Senauer, B. ( 2013, August). Considering the mandatory labeling of genetically engineered foods in the U.S. Retrieved from University of Minnesota, Food Policy Research Center website: http://www.foodpolicy.umn.edu/policy summaries andanalyses/ge labeling/index.htm Shannon, C. E., & Weaver, W. (1949). The mathematical theory of communication . Urbana: University of Illinois Press. Sheskin, D. J. (2004). Handbook of parametric and nonparametric statistical procedures (3rd ed.). CRC Press LLC., Boca Raton, FL. Siegrist, M. (2000). The influence of trust and perceptions of risks and benefits on the acceptance of gene technology. Risk Analysis , 20 , 195 203. doi:10.1111/0272 4332.20 2020 Siegrest, M. (2008). Factors influencing public acceptance of innovative food technologies and products. Food Science and Technology , 19 , 603 608. doi: 10.1016/j.tifs.2008.01.017
201 Slovic, P. (1987). Perceptions of Risk. Science , 230 , 280 285. Retrieved from http://www.uns.ethz.ch/edu/teach/0.pdf Snyder, M. (1974). Self monitoring of expressive behavior. Journal of Personality and Social Psychology , 30 , 526 537. Retrieved from http://www.communicationcache.com/uploads/1/0/8/8/10887248/self monitoring_of_e xpressive_behavior.pdf Sparks, P., & Shepherd, R. (1994). Public perceptions of the hazards associated with food production and food consumption: An empirical study. Risk Analysis , 14 , 79 86. Retrieved from doi: 10.1111/j.1539 6924.1994.tb00291.x Sparks, P., Shepherd, R., & Frewer, L. J. (1994). Gene technology, food production, and public opinion: A UK study. Agriculture and Human Values , 11 (1), 19 28. doi:10.1007/BF01534445 Staats, A. W., & Staats, C. (1958). Attitudes established by classical co nditioning. Journal of Abnormal and Social Psychology , 67 , 159 167. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/13563044 Takeda, S., & Matsuoka, M. (2008). Genetic approaches to crop improve ment: responding to environmental and population changes. Nature Reviews Genetics , 9 , 444 457. doi:10.1038/nrg2342 Telg, R., & Irani, T. A. (2012). Agricultural communications in action: A hands on approach. Clifton Park, NY: Delmar, Cengage Learning Tesse r, A. (1993). The importance of heritability in psychological research: The case of attitudes. Psychological Review , 100 , 129 142. Retrieved from http://www2.psych.ubc.ca/~schall er/528Readings/Tesser1993.pdf The New York Times. (2015a, May 25). Search: GMOs. Retrieved from http://query.nytimes.com/search/sitesearch/?action=click&contentCollectionÂ®ion =TopBar&WT.nav=searchWidget&module=SearchSubmit&pgtype=Homepage#/ genetically+engin eered+food/from20140917to20140926/ The New York Times. (2015b, May 29). Search: AgLabs. Retrieved from http://query.nytimes.com/search/sitesearch/?action=click&contentCollectionÂ®ion =TopBar&WT.nav=searchWidget&module=SearchSubmit&pgtype=Homepage#/ dupont+pio neer/from20000101to20150101/ The New York Times. (2015c, May 29). Search: Green Giant. Retrieved from http://query.nytimes.com/search/sitesearch/?action=click&contentCollectionÂ®ion =TopBar&WT.nav=searchWidget&module=SearchSubmit&pgtype=Homepage#/ Green Giant /from20000101to20150101/ United States Census Bureau . (2014, March 27). State and County Quick Facts. Retrieved from http://quickfacts.census.gov/qfd/states/12000.html
202 University of Illinois Extension. (2001 ). Agronomy day The Morrow plots: A landmark for agriculture. Retrieved from http://agronomyday.cropsci.uiuc.edu/2001/morrow plots/ van Kleef, E. V., Frewer, L. J., Chryssochoidis, G. M., Houghton, J. R., Korzen Bohr, S., Krystallis, T., . . . Rowe, G. (2006). Perceptions of food risk management among key stakeholders: Results from a cross European study. Appetite , 47 , 46 63. doi:10.1016/j.appet.2006.02.002 Ver beke, W. (2005). Agriculture and the food industry in the information age. European Review of Agricultural Economics, 32 (3), 347 368. Retrieved from doi: 10,1093/eurrag/jbi017 Verbeke, W., & Vackier, I. (2004). Profile and effects of consumer involvement i n fresh meat. Meat Science, 67 , 159 168. doi:10.1016/j.meatsci.2003.09.017 Verbeke, W., & Ward, R. W. (2006). Consumer interest in information cues denoting quality, traceability and origin: An application of ordered profit models to beef labels. Food Quality and Preference, 17 (6), 453 467. doi:10.1016/j.foodqual.2005.05.010 Verdurme, A., & Viaene, J. (2003). Exploring and modelling consumer attitudes towards genetically modified food. Qualitative Market Research: An International Journal , 6 (2), 95 110. doi:10.1108/13522750310470109 Wachenheim, C., & Rathge, R. (2000). Societal Perceptions of Agriculture (Report No. 449). Retrieved from Agribusiness and Applied Economics Report website: http://ageconsearch.umn.edu/bitstream/23541/1/aer449.pdf Weed Scienc e Society of America. (2015). International survey of herbicide resistant weeds . Retrieved from http://www.weedscience.org/summary/home.aspx Weigold, M. F. (2001). Communicating science: A review of the literature. Science Communication, 23 (2), 164 193. doi: 10.1177/1075547001023002005 Whalen, D. J. (1996). I see what you mean: Persuasive business communication . Thousand Oaks, CA: Sage Publications. Whitaker, B. K., & Dyer, J. E. (2000). Identifying sources of bias in agricultural news reporting. Journal of Agricultural Education, 41 (4), 125 133. doi: 10.5032/jae.2000.04125 Wood, W., Rhodes, N., & Biek, M. (1995). Working knowledge and attitude strength: An information processing analysis. In R. Petty, & J. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 283 313). Hillsdale, NJ: Lawrence Erlbaum Associates.
203 World Health Organization. (2009). 20 questions on genetically modified (GM) foods. URL . htp: //www.who.int/foodsafety/publications/biotech/20questions/en/. Last accessed 1.26.14 World Health Organization. (2015). Micronutrient deficiencies Vitamin A deficiency. Retrieved from http://www.who.int/nutrition/topics/vad/en/ Zickuhr, K. (2010). Generat http://www.pewinternet.org/files/old media//Files/Reports/2010/PIP_Generations_and_Tech10 .pdf Zimbelman, R. G., Wilson, L. L., Bennett, M. B., & Curtis, S. E. (1995). Public image of animal agriculture in the United States. Livestock Production Science, 43 (2), 153 159. doi: 10.1016/0301 6226(95)00040 R
204 BIOGRAPHICAL SKETCH Taylor Ruth was born in 1991, and grew up in the outskirts of Jacksonville, Florida in the small town of Fleming Island. Taylor is the only child of David and Jean Ruth. Taylor attended the University of Florida pursuing a Bachelor of Science in m icrobiolo gy and cell s cie nces. After becoming a sister in Sigma Alpha, a professional agricultural sorority, she learned about the agricultural education and communication depart ment, and eventually pursued a minor in agricultural c ommunications. Once she graduated with her bachel Taylor took two semesters off to research gen etic mutations in maize in the H orticulture Sciences D epartment at the University of Florida. In the s pring of 20 14, Taylor began her enrollment in the Master of Science program for agricultural education and communication, specializing in agricultural communication. She spent the majority of her time researching consumer perceptions related to food and natural resources in the state of Florida through the Center for Publi c Issues Education (PIE Center). Taylor will begin the doctoral program at the University of Florida in agricultural educ ation and communication in the f all of 2015. She has accepted an assistantship through the PIE Center, and will continue researching co nsumer perceptions related to agricultural issues.