Citation
Differences in Calibration of Reading Comprehension and Self-Efficacy across Achievement Levels and Learning Disability Status

Material Information

Title:
Differences in Calibration of Reading Comprehension and Self-Efficacy across Achievement Levels and Learning Disability Status
Creator:
MOLITCH-HOU, TAMARA E. ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Academic learning ( jstor )
Banduras ( jstor )
Calibration ( jstor )
Disabilities ( jstor )
Educational psychology ( jstor )
High school students ( jstor )
Learning disabilities ( jstor )
Reading comprehension ( jstor )
Self concept ( jstor )
Special needs students ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Tamara E. Molitch-Hou. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
8/1/2006
Resource Identifier:
84564539 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

DIFFERENCES IN CALIBRATION OF READING COMPREHENSION AND SELFEFFICACY ACROSS ACHIEVEMENT LEVE LS AND LEARNING DISABILITY STATUS By TAMARA E. MOLITCH-HOU A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS IN EDUCATION UNIVERSITY OF FLORIDA 2003

PAGE 2

ii ACKNOWLEDGMENTS I would like to acknowledge th e students and staff of Interlachen and Prairie View Elementary Schools. Their eager participat ion was the foundation of this project. I appreciate the flexibility and enthusiasm of th e teachers in allowing me to work with their students. In addition, I tha nk Rob Klassen from Simon Fras ier University and Dr. John Kranzler from the University of Florida, for their early input into th e design of my study, as well as guidance in con ceptualizing calibration. I thank Sarah Graman for her assistance in the data collection process. I w ould also like to exte nd my gratitude to Dr. Randall Penfield and Mike Padilla for their critic al statistical advice. I would also like to thank my parents for their support. They have encouraged me throughout my education and I could not have achieved so much wit hout their love behind me. Finally, I would like to express my gratitude to my committee members, Dr. Waldron and Dr. Davis. Dr. WaldronÂ’s support and guidance throughout the process helped me to maintain my serenity while completing a thes is along with the other pressure s of graduate school. Dr. Davis contributed to this project from its inception and helped me to follow it through successfully. When I struggled to describe the concept I was interested in, she was there to name it and direct me to an established literature. Her assist ance in developing the project, guiding the research process, and helping me to push through writing has been indispensable.

PAGE 3

iii TABLE OF CONTENTS Page ACKNOWLEDGMENTS .................................................................................................. ii LIST OF TABLES...............................................................................................................v ABSTRACT....................................................................................................................... vi CHAPTER 1 INTRODUCTION ............................................................................................................1 2 REVIEW OF THE LITERATURE ..................................................................................4 Self-Efficac y .................................................................................................................4 Calibration ................................................................................................................... . 5 Overconfidence and Underconfidence in Calibration...........................................6 Group Differences in Calibration..........................................................................7 Calibration Across Academic Domains................................................................9 Purpose of the Current Study........................................................................................9 Research Questions.....................................................................................................11 Hypothese s ..................................................................................................................11 3 METHODOLOGY .........................................................................................................12 Participants .................................................................................................................12 Measure s .....................................................................................................................12 Procedur e ....................................................................................................................14 Recruit m ent of Participants .................................................................................14 Infor m ed Consent Proces s ...................................................................................14 Data Collectio n ....................................................................................................15 Assign m ent of Participant Group s .......................................................................17 Scoring the Data ..................................................................................................18 Analyses ..............................................................................................................19 4 RESULTS ..................................................................................................................... . .21 Group Diff e rences in Dependent Variable s ................................................................21 Reading Self-Concep t ..........................................................................................22

PAGE 4

iv Reading Comprehension Total Scores ................................................................23 Reading Self-Efficacy Total Sc o r es ....................................................................23 Group Diff e rences in Calibration ...............................................................................24 Use of the Covariate: Reading SelfC oncep t .......................................................24 Mean Bia s ............................................................................................................24 Mean Accurac y ....................................................................................................25 Item Accuracy .....................................................................................................25 5 DISCUSSIO N .................................................................................................................2 7 Calibration in Reading Co m prehensio n ......................................................................27 Mean Bia s ............................................................................................................2 8 Mean Accurac y ....................................................................................................28 Total Item Accurac y ............................................................................................29 Sum m ary of Calibration Results .........................................................................29 Reading Self-Concep t .................................................................................................30 I m plications ................................................................................................................31 Li m itations of the Study .............................................................................................32 Future Directions for Researc h ...................................................................................3 4 Interventions for I m proving Calibratio n .....................................................................35 Increasing S e lf-Regulation ..................................................................................35 Instructional Method s ..........................................................................................36 APPENDIX A ACAD E MIC SELF DESCR I PTION QUES T I ONNAIR E I .........................................38 B W I AT-II R ECORD FORM ................................................. . ..........................................39 C LIKERT SCAL E ............................................................................................................41 LIST OF REFERENCE S ...................................................................................................42 BIOGRAPHICAL SKETCH .............................................................................................45

PAGE 5

v LIST OF TABLES Table page Table 1. Mean and standard deviation for all m ea s ures by grade, learning disability status, and a chieveme n t level . ..................................................................................22 Table 2. Significance levels for R e ading Self-Concept as a covariate. .............................24

PAGE 6

vi Abstract of Thesis Presented to the Graduate School of th e University of Florida in Partial Fulfillment of the Requirements for th e Degree of Master of Arts in Education DIFFERENCES IN CALIBRATION OF READING COMPREHENSION AND SELFEFFICACY ACROSS ACHIEVEMENT LEVE LS AND LEARNING DISABILITY STATUS By Tamara E. Molitch-Hou August 2003 Chair: Nancy L. Waldron Cochair: Heather A. Davis Major Department: Educational Psychology The purpose of the current study is to explore the phenomenon of calibration of reading comprehension achievement and read ing self-efficacy in students of different achievement levels, with and without lear ning disabilities. Mi scalibration can limit academic success due to a lack awareness of wh en and how to ask for assistance due to not recognizing weaknesses. Knowledge about how calibration functions for students of different abilities, as well as suggested in terventions to improve calibration, can help school instructors to improve the accuracy with which these students perceive their abilities and consequently maximize specializ ed instruction time. Self-efficacy and calibration research are explor ed in depth. Results from the current research study comparing self-perception of reading compre hension skills in upper elementary students with and without learning disabilities are discussed. Findings s uggest that there are significant differences in calibration levels for students of different achievement levels.

PAGE 7

vii Students of higher achievement demonstrate better calibration, th an those of lower reading ability. Additionally, typical stude nts demonstrate better calibration than do students with learning disabiliti es. Implications for instru ction and intervention based on these results are described.

PAGE 8

1 CHAPTER 1 INTRODUCTION This study explores the constr uct called calibratio n as it occurs across students of varying achievement levels. Calibration is defined as a match between self-efficacy judgments and actual achievement in a partic ular subject area. For the current study, calibration is examined in the subject area of reading comprehension. Students of low, average and high achievement levels, as well as students with lear ning disabilities are compared in their calibration levels. I became interested in the concept of calibration when working on an academic assessment and intervention for a student who was struggling in reading. Despite consistent low performance on reading task s, the student with whom I was working believed she was a good reader. She st ruggled with decoding, fluency and comprehension. She became easily frustrated with grade-level reading tasks. Although she struggled through her reading assignments and was the lowest performing reader in her class, she reliably repor ted that she was a good reader. On specific tasks she would predict that she had gotten 90-100% correct wh en she had only answered correctly for less than 50% of questions. Sh e denied offers for help in reading even though she could not effectively use strategies that she had le arned in class. While designing interventions, I focused on her specific reading deficits. We worked on tasks to improve decoding and fluency. These remediation tasks were not me t with success. The student continued to struggle and continued to lack awareness of he r deficits. I then shifted the focus from

PAGE 9

2 specific reading interventions to what I might now think of as calibra tion interventions. I helped the student to become more awar e of her own weaknesses in reading through sharing some assessment results. We set goals for reading so that she could monitor her own progress. I also reinforced her ability to recognize when she could and could not read particular words, sentences, or passage s. We then worked on making a list of strategies for reading and she was rewarded for being able to remember strategies and picking the appropriate one, in addition to actually using it. In adding these metacognitive components to the reading inte rventions the student began to appreciate the relevance of the reading strategies I was teaching. At the time I did not recognize this as an issue of miscalibration. However, I did begin to questio n how much of the problems students experience academically might be due to a lack of awareness regarding their own abilities. I spoke to a member of the department facu lty, Dr. Davis, about this concept. She referred me to some literature that had named and researched this very concept. I began a review of the literature on calibration and found that it o ffers a potentially informative explanation for why some stude nts may have a particularly difficult time in school. Many students may simply be unaware of their own areas of weaknesses and may therefore lack interest, know ledge, or motivation regarding what they should do to improve. Although calibration has been explored, there is still a lot of room to expand on what currently exists in the literature. The earliest discussion I found regarding this construct is in Blumfield, Pintrich, Meece, and Wessels (1982), although not by the term calibration. Many researchers disc uss this self-evaluation of ability, however the earliest use of the term “calibration” is in Schraw, Po tenza, and Nebelsick-Gullet (1993). At this

PAGE 10

3 point I have not found any discussion of how calibration functions for reading comprehension. There has been some research regarding calibration in mathematics and writing. In the learning disabili ties research there is consensu s that students with learning disabilities have severe deficits in metac ognition, however there is no specific discussion of calibration problems among these deficits (Wong, 1995). There is minimal discussion of calibration for students with learning disabilities in the calibration literature. These gaps in the understanding of calibration in reading compre hension and for students with learning disabilities led me to the curre nt study. I compare st udents of different achievement levels, with and w ithout learning disabilities in their calibration of reading self-efficacy and reading comprehension. In doing so, I hope to contribute information regarding the difficulties that some students face in reading comprehension. I also offer interventions that may aid students in b ecoming more accurately calibrated.

PAGE 11

4 CHAPTER 2 REVIEW OF THE LITERATURE In discussing relevant literature to this investigation, self-efficacy, self-concept, and calibration will be addressed. Self-efficacy represents the area with the strongest research background and contribut es substantially to an unde rstanding of calibration. For this reason, self-efficacy is covered first. It is followed by a discussion of the existing literature on calibration. Calibra tion will be addressed as a ge neral construct, as well as how it occurs for different groups of individual s according to previous research findings. Finally, a discussion of calibra tion in different academic domains, particularly reading comprehension, is included. The chapter concludes with a discussion of the purpose, research questions, and hypotheses of the current study. Self-Efficacy Bandura defines self-efficacy as, “beliefs in one’s capabilities to organize and execute the courses of action required to manage prospective situations” (1997, p.2). Self-efficacy, as it relates to academic performa nce, is a critical area of study because it can predict a sizeable portion of student performa nce. In some circumstances, beliefs in one’s abilities may outweigh actual abilities because of the strong influence they have over performance outcomes. People choose to cultivate their competencies in different areas. According to Bandura (1997), this will re sult in self-beliefs that differ for different areas of functioning. Therefore, based on their experiences, knowledge, and skills, individuals will develop positive and negative beliefs about their efficacy in different areas. Effective functioning requires adequate skills and also feelings of efficaciousness

PAGE 12

5 regarding competence (Bandura, 1997). For example, Pajares (1996a) found that selfefficacy beliefs made an independent contribu tion to performance of gifted students in mathematics. More specifically, he found th at low self-efficacy caused some students to believe tasks were more difficult than th ey actually were, which produced stress, depression and poor problem solving skills (1996a). Bandura suggests that efficacy beliefs “affect thought processes, the level and persistency of mo tivation, and affective states“ (1997, p. 39) which all a ffect performance outcomes. Due to the influential role that self-efficacy beliefs have on performan ce, it becomes important to consider the accuracy of those beliefs. Frequently indi viduals hold beliefs about their own abilities that are not accurate, and then form opinions about themselves, or make choices about how to behave based on thes e erroneous beliefs. Calibration When personal beliefs about abilities are ma tched to actual abilities, individuals are said to be well-calibrated (Pajares & Kran zler, 1995). Accurate calibration is rare, especially in children, but improves with de velopmental maturation (Pajares & Kranzler, 1995; Bouffard, Markovits, V ezeau, Boisvert, Dumas, 1998) . Young children tend to be inaccurate in the direction of extreme overc onfidence (Blumfield et al., 1982; Bandura, 1986). Blumfield and colleagues hypothesize th at this is because they have little understanding of their personal limits. As metacognitive knowledge develops, children begin to acquire a more realistic underst anding of their own limitations and human limitations. However, depending on persona l experience, calibration improves at different rates for individua ls across different points in development, as does metacognitive ability.

PAGE 13

6 Blumfield et al. (1982) found that the t ypes of tasks and feedback used in elementary school classrooms can prolong ove rconfident beliefs. Their explanation is that feedback and evaluation are not always contingent on actual performance but on teacher perceptions of students and attempts to increase self-esteem. Although intended to encourage success, this behavior on the teache rÂ’s part may be to the detriment of actual achievement. Blumfield and colleagues (1982) point out that young children may not be extremely logical in analyzing sources of beha vior. They are more likely to judge their efficacy based on performance of others, or on task difficulty, rather than on personal competence at a task (Blumfield, 1982). These authors suggest that, with age, calibration abilities will improve as a function of deve lopmental factors. Bandura (1986) supports this notion, explaining that in general, ca libration improves along with metacognitive development with age; both are related to development of ability and competence in specific academic areas as well. Overconfidence and Underconfidence in Calibration Calibration may be important because it enab les students to participate in effective comprehension monitoring during a learning or testing situation (Schraw, Potenza, & Nebelsick-Gullet, 1993). That is, a successful student is able to monitor his or her comprehension as s/he goes through a partic ular instructional e xperience. Accurate calibration is ideal; however, it may come at the cost of lower optimism and lower levels of the functional goals of self-efficacy, includ ing effort and persiste nce (Pajares, 1996a). According to Bandura (1986), the most successf ul performance is achieved when selfefficacy judgments slightly exceed actual capabilities, in other words, slight overconfidence. Overconfidence tends to pus h students to attempt more challenging tasks. Extreme overconfidence, however, is not advantageous because it reflects an

PAGE 14

7 inability to recognize oneÂ’s own weaknesses. This may cause inefficient use of the skills a student does have and a lack of awareness of when and how to seek assistance. People who demonstrate overconfidence take on situatio ns that they cannot handle that can result in physical and emotional risk -taking (Bandura, 1997). U nderconfidence, alternatively, can be detrimental because it may limit what a student is willing to attempt to do. This can result in lost opportu nities, and underdevelopment of potential, resulting in failure and regret (Bandura, 1997). Bandura (1986) ex plains that stronger self-efficacy makes a person more likely to undertake demanding tasks, persist at them, and achieve success. Group Differences in Calibration Prior research has shown that different populations of students demonstrate varying levels of self-efficacy, as well as calibration. For example, research in mathematics has shown that gifted students are more accurately calibrated than typical students; however, they tend to err slightly in the direction of overconfidence in their abilities (Pajares & Kranzler, 1995). For these students, ove rconfidence may encourage them to pursue challenging tasks. This degree of overc onfidence is commensurate with BanduraÂ’s description of the optimal le vel. Lower achieving students have been found to be more accurately calibrated than students of aver age achievement in both math and writing (Pajares & Kranzler, 1995; Shell, Colvin, & Bruning, 1995). They are more able to recognize what they do not know or understand, possibly due to consistent feedback from teachers. Unlike students of globally low achieve ment, it has been shown across several studies that students with lear ning disabilities tend to be have significan tly higher selfefficacy beliefs, and are thus overconfident (Klassen, 2002). Graham and Harris (1989) explain there is a great deal of research to suggest that stude nts with learning disabilities

PAGE 15

8 “have difficulty accurately assessing or pr edicting their performance capabilities” (p. 212). Shell et al. (1995) supported this notio n, having found that students with learning disabilities tend to overestimate their efficacy while low achievers were more accurately calibrated in writing tasks. One explanation for why student s with learning disabilities may be inaccurately calibrated is that it may reflect the discrepancy between their cognitive functioning and their performance in specific academic domains. This explanation stems out of the de finition that is typically used to classify students with learning disabilities (Klassen, 2002). The clas sification of learning disability typically necessitates of split of at least one standard deviation be tween cognitive ability (IQ) and achievement in an academic area. The inability to recognize their weakne sses may also be an example of poor metacognitive skills. Students w ith learning disabilities tend to have difficulty engaging metacognitive processes that would be used to actively monitor performance for different tasks (Butler, 1998). Butler (1998) explains that metac ognitive problems of students with learning disabilities are much more pronounced than those of peers. Evidence suggests that students with le arning disabilities may have di fficulty with a variety of metacognitive tasks including: recogniz ing task requirements, selecting and implementing strategies, monitoring and adjusting performance, motivation, and emotional control (Butler, 1998). Pajares a nd Kranzler (1995) suggest that improving calibration will help students to recognize wh at they do and do not know, so that they learn how to utilize cognitive strategies in the problem solving process. They recommend attempting to improve calibration wh ile taking care not to lower self-efficacy or self-confidence.

PAGE 16

9 Calibration Across Academic Domains A good deal of research has examined some aspect of calibration in writing (Shell et al., 1995; Sawyer, Graham, & Harris, 1992; Klassen, 2002b) and mathematics (Pajares, 1997; Pajares & Graham, 1999; Pajares & Kr anzler, 1995). However, there are few studies of calibration in the area of re ading achievement, specifically reading comprehension. Reading comprehension is among the most important academic skills to master in early school years because it is a re quisite skill for so ma ny other subject areas (Mastropieri & Scruggs, 1997). It has been defined as “a process of constructing meaning from written texts, based on a complex coordination of a number of interrelated sources of information” (Mastropieri & Scruggs, 1997, p. 197). Stude nts who struggle in reading tend to be less awar e of when they are appropria tely understanding what they read, likely due to a lack of self-monitori ng (Garner, 1980, as cited in Pressley, 1998). Therefore, not only are they not comprehending a piece as well as better readers, but they are also in a poor position to make decisi ons about when to make extra efforts to comprehend better (Pressley, 1998). It may be difficult for any student to gauge how well s/he has understood a particular piece in comparison to peers. For this reason, calibration may be an essential piece of info rmation about how a student is performing in reading comprehension. If this lack of self -awareness prevents a student from asking for help, or utilizing strategies that may alre ady be known, performance accordingly suffers. Purpose of the Current Study The purpose of the current study is to expl ore the role of cali bration in reading comprehension for students of different achie vement levels. Calibration assessment has the potential to guide numerous interventions . A student who is capable of accurately assessing what strategies to use, what s/he does or does not know, and when and where to

PAGE 17

10 seek assistance, will be much more successf ul than one who presses on blindly. For example, Mastropieri and Scruggs (1997) found that students who were trained in selfmonitoring procedures in addition to reading comprehension strategi es showed enhanced performance. This type of calibration-improving instruction can help a student to expand skills across all academic areas. Several researchers suggest that further ex ploration of calibration in students of lower abilities and w ith learning disabilities is needed (Pajares, 1996a; Klassen, 2002a). The present study examines cal ibration in reading compre hension for students of different skill levels. Specifically, th ree categories of reading comprehension achievement (low-achieving, average achievi ng, and high achieving) are compared, as well as students with learning disabilities in terms of level of ca libration. The reading comprehension and self-efficacy measures are domain specific to reading comprehension, as in the past omnibus measures have b een known to weaken effects (Pajares, 1996b). Calibration is measured by comparing pr edicted performance (self-efficacy) on comprehension questions with actual perfor mance on matched comprehension questions regarding the same set of read ing comprehension tasks. Results from a broader reading self-c oncept measure are included to examine studentsÂ’ global feelings of self-worth as re aders and to compare this to their perceived self-efficacy on more specific tasks. Bandur a conceptualized self-concept as a global perception of the self (1986). This broad c onstruct is related to and may be made up of more narrow perceptions of the self, such as self-efficacy (Bandalos, Yates, & Thorndike-Christ, 1995). It is expected th at students determine some of their selfefficacy judgments based on how they have performed in that area in the past, as

PAGE 18

11 represented by their self-concept in that area (Bandalos et al., 1995). For these reasons, it was predicted that reading self-concept woul d predict self-efficacy judgments and would contribute to calibration. Research Questions I. How does calibration differ across achievement levels in reading comprehension using reading se lf-concept as a covariate? II. How does calibration differ for students with learning disabilities and students not identified with learni ng disabilities in reading comprehension using reading self-concept as a covariate? Hypotheses I. Students with learning disabi lities will be overconfident in their abilities. II. Students without learning disabilities w ill be more accurately calibrated than students with learning disabilities. III. Positive self-concept will predict overconfidence in students with lowachievement and students with learning disabilities. Negative self-concept will predict more accurate calibration for students with low-achievement and students with learning disabilities , and will predict underconfidence for students with average achievement students.

PAGE 19

12 CHAPTER 3 METHODOLOGY Participants Participants in the study included 169 students (95 female, 74 male) from two public elementary schools in rural communities of North Central Florida. Participants included 98 fourth-grade and 71 fifth-grade students. Of th e participants, there were 127 Caucasian students, 25 African American st udents, 15 Hispanic students, and 2 Asian students. There were 25 student s identified as learning disabled. 1 Measures Reading self-concept was measured using a modified version of the Academic SelfDescription Questionnaire I (ASDQ-I), produced by Marsh at the Self-concept Enhancement and Learning Facilitation (S ELF) Research Centre, 1999. Only the reading-related questions were given to the students, yielding a 9-question survey with an eight point Likert type respons e. This measure is comparable to the mathematics selfconcept scale from the Academic Self-Descr iption Questionnaire II used by Pajares and Graham (1999) for the same purpose. One item was eliminated from analyses for the ASDQ-I because item analysis found that it di d not contribute strongl y to the reliability of the scale. Correlations for item 8 ranged from -.03 to .15 with other items. 1. Students with learning disabilities were identified as having a Specific Learning Disability according to definitions provided in the Florida State Regulations. This identification did not specify whether the disability was in Reading.

PAGE 20

13 A modified version of the Wechsl er Individual Achievement Test, 2nd Edition (WIAT-II; 2001), was used to measure Readi ng Comprehension achievement. This test was selected because its validity and reliabi lity were established with a large normative sample including a subset of students with learning disabilities. The reliability of the WIAT-II Reading Comprehension subtest is stro ng, with an alpha coefficient of .96 for the grades involved. Additionally, validity of the WIAT-II was established in comparison with other individually admi nistered achievement tests and group administered achievement tests. Validity comparisons with the Stanford Achievement Tests, 9th edition, found a correlation of .75 for Total Reading. It was valuable that strong validity was establishe d in comparison to group-admi nistered tests because the achievement groups for this study were divided based on the comparable groupadministered, Florida Comprehensive Achi evement Test. Questions from the WIAT-II were selected across 2nd through 8th grade achievement expectations. This range was used to ensure that students of all achieve ment levels would be able to answer some questions correctly, and that no students would be able to answer all of the questions correctly. In using this set of questions, all students answer ed questions that ranged from easy to difficult. Selected items typically i nvolved reading aloud and answering direct or inferred questions from the text. These questions were also the basis for the self-efficacy questions. Self-efficacy was measured through questions that were designed to directly correspond to the comprehension questions from the WIAT-II. Efficacy judgments must assess the same skills to be used in the actual performan ce task of comparison (Pajares & Kranzler, 1995). The questions took the standard form , “How confident are you, on a scale from 1

PAGE 21

14 to 6, that you can __(question)___?” (for ex amples, see Appendix B). A six-point Likert rating scale was developed as a visual aide for answering the questions (see Appendix C). Procedure Recruitment of Participants The University of Florida Institutional Review Board approved the research study and all appropriate consent pr ocedures in July, 2002. A pproval to recruit schools was given by county officials for both countie s involved in August, 2002. The primary investigator then presented th e project to school principals. The principals arranged for brief presentations to the four th and fifth grade teachers. All teachers at the two schools agreed to the participation of their classes. Students recruited were enrolled in fourth or fifth grades (8-11 years old), general and special education. Each classroom of fourth or fifth graders, including exceptional student education self-contained classrooms, was given an explanation of the research study a nd consent forms with an attached summary of participation to take home. Informed Consent Process Letters were sent home to parents that de scribed the research project in detail. They were asked to return a signed and dated consent form that allowed them the option of agreeing or refusing to participate in the study. Only students whose parents consented, were asked to participate. Parent s had the option of wit hdrawing their student at any time. Students who returned consen t forms, with or without permission, were rewarded with the choice of a piece of candy or a sticker. This was done to encourage them to remember to ask their parents and to re turn the forms. The investigator also used a verbal assent script before participation to ask the students whether or not they wanted

PAGE 22

15 to participate. The investigator describe d the study including what the student was expected to do. The students were informed that they did not have to answer any questions and could cease partic ipation at any time. They we re then asked to sign a form to indicate whether they were w illing to participate. Only t hose who assented took part in the study. Data Collection Data collection occurred during Fall, 2002. Two measures were administered to all students who gave consent and assent. The first measure, the Academic SelfDescription Questionnaire (see Appendix A) was given in small classroom groups by the primary researcher. The survey was admini stered in classrooms or in the hallway depending on the number of participants per classroom. It took approximately fifteen minutes to administer. Students answered nine questions designed to measure their reading self-concept. The que stions addressed how the stude nts felt about themselves as readers using a Likert scale from 1 (definitely fa lse) to 8 (definitely tr ue). All directions were read orally and further explanation wa s given for the Likert scale on the survey. Examiners went through the following proce ss to ensure student understanding of the survey, “On this survey you are going to be answering questions on a scale from 1 to 8. Has anybody every asked you how much you like something, from 1 to 10? Well this is sort of like that. Let me give you an example. If I put this sent ence on the board [write sentence if possible] ‘I love mushroom pizza’. Raise your hand if you totally agree with this sentence, okay, you would all be 8 on this scale. Now raise your hand if you don’t agree at all with this sentence, you hate mushroom pizza, you would be 1. And does anybody feel somewhere in the middle, you don’t love mushroom pizza but you might eat it, or eat it and pick the mushrooms off? You would be somewhere in the middle, like

PAGE 23

16 a 4 or a 5”.” Further explanation was given if the students did not appear to understand the directions. As each question was asked, students were reminded of what each end of the scale meant for that question. Potentiall y confusing vocabulary was explained further for items 1 and 9, including the words “hopeless” and “satisfied.” The second measure was administered indivi dually by the primary investigator or a trained assistant. The student s were taken to a quiet area outside the classroom, typically a part of the hallway or an empty classr oom where there were no interruptions. They were individually administered selected questions from the Reading Comprehension subtest of the Wechsler I ndividual Achievement Test, 2nd Edition (WIAT-II) (see Appendix B for types of questions). Prior to answering each comprehension question, the students were asked to predict whether they believed they could answer the question correctly on a Likert Scale of 1 to 6. They were reminded of the Likert Scale explanation from the previous instrument. Examiner instructions were as follows: “I am going to ask you some reading questions. Some might be easy for you a nd others might be hard for you, you probably won’t know the answer to all of the questi ons. Before each question, I am going to ask you to tell me how confident you are that you can give me the right answer to a question. Do you know what confident means? Confident is how sure you are about something. For example, I am confident that fire is hot, but I am not confident that it will always be sunny outside. Can you tell me some things you are confident about ? You are going to tell me how confident you are that you can answer correctly on a scale from 1 to 6 (see Appendix C). Look at this scale, see how it says “Not confident at all” at the 1 and “Extremely Confident” at the 6? When I as k how confident you are, you say or point to

PAGE 24

17 the number that says how confident you are. [r ead all choices and point to numbers] It’s okay to tell me that you don’t think you will kno w the answer. After that part, you are going to try to answer the question. I cannot tell you if you get th e question right or wrong. Do you understand?” Each session took approximately 15 minutes per student. They gave a self-efficacy rating from 1 to 6 and were then asked the reading comprehension question. Reading Comprehens ion responses were scored 0, 1, or 2 based on the prescribed WIAT-II scoring procedures. Assignment of Participant Groups Students who returned positive consent and al so assented to participate made up the four groups of participants. All data collection procedures occurred before determine group assignment. Therefore, examiners were blind to achievement levels and learning disability status. The mean and standard devi ation were calculated for the participants of the study and matched closely th e percentile score distributi on of a normal distribution. Information regarding FCAT scores and ps ychoeducational classifi cation were obtained, with informed consent from parents, through a review of educational records. The first group included students who had been identified with a specific lear ning disability by the school district, which included students across low and average achievement levels. The second group, was made up of students with lo w-achievement in reading. These students were identified by percentile scores that fell greater than one standard deviation below the mean on the reading portion of the 2001 Florida Comprehensive Assessment Test (FCAT). The third group, was made up of students with average achievement. These students were identified by performance with in one standard devi ation above or below the mean on the reading portion of the 2001 FCAT. The fourth group, was made up of

PAGE 25

18 students with high achievement. These stude nts were identified by performance greater than one standard deviation above the mean on the reading portion of the 2001 FCAT. Scoring the Data Achievement level. Students were compared across achievement levels as determined by their performance on the previous yearÂ’s Florida Comprehensive Assessment Test and a review of school reco rds. Based on percentile ranks on the FCAT or identification of a learning disability, they fell into one of the following categories: low achievement, average achievement, high achie vement, and identified with a learning disability. Reading self-concept . Scores on the Academic Self -Description Questionnaire-I were used to globally assess studentsÂ’ feeli ngs about their own reading skills. Selfconcept scores were determined by calcula ting a mean of the ASDQ-I ratings. The reading self-concept scores also functioned as a covariate, representing how much of the variance in calibration is accounted for by reading self-concept. Reading comprehension performance . The WIAT-II Reading Comprehension produced a composite score that was compared across students. All of the selected questions were administered to all students. Questions ranged from second to eighth grade levels to ensure that some ques tions would be easy and challenging for all participants. Performance on this test pr oduced a task-specific reading comprehension score, calculated by adding the reading comprehension scores for all questions. Reading comprehension self-efficacy . Students answered Likert-type questions about their confidence in answering the comprehension questions from the WIAT-II correctly, before actually answer ing the questions. The Likert scale ranged from 1 (not at

PAGE 26

19 all confident) to 6 (extremely confident) fo r each question. A tota l self-efficacy score was obtained by adding the va lues of these scores. Calibration. Using methods described by Pa jares (1996a), th ree calibration scores were calculated based on results from a comparison of the Reading Comprehension Performance and Reading Comprehension Self-Efficacy scores as described above. The first, a mean-bias score reflected the direction of errors in judgment. For this score, performance was sc ored so that a 6 was a correct answer, a 3 was a one-point (partial credit) answer, a nd a 1 was an incorrect answer. Actual performance (1, 3 or 6) was subtracted from predicted performance (1 through 6). The mean of the bias scores reflects overconfid ence or underconfidence. Overconfidence is reflected in a more positive score, underconfidence is reflected in a more negative score. The second calibration score was mean-accuracy , computed by taking the mean of the absolute value of all of the bias scores. This score demonstrates the magnitude of judgment error. Finally, an item-accuracy sc ore was calculated to determine the number of times that efficacy judgment and perfor mance were the same. Students received a point for every time they expressed confiden ce (5 or 6) and answered correctly, every time they expressed unsure confidence (3 or 4) and earned partial credit, and every time they expressed lack of confidence (1 or 2) and answered incorrectly. Analyses One-way Analysis of Covariance (ANC OVA) procedures were conducted to determine group differences for reading co mprehension, reading self-efficacy, and the three types of calibration, while controlling for readi ng self-concept. In this model, the independent variable, achievement level, in cluded four levels: le arning disability, low achievement, average achievement, and high achievement. The dependent variables for

PAGE 27

20 the analyses were the thr ee calibration score, mean-bias, mean-accuracy, and itemaccuracy. Reading Self-Concept was used as a c ovariate. BonferroniÂ’s tests were used to evaluate pairwise differences in the means of each group. A significance level of .05 or less was used to determine statistical significance levels for all analyses.

PAGE 28

21 CHAPTER 4 RESULTS Group Differences in Dependent Variables The purpose of the study was to determine differences in calibration levels across achievement levels and for students with learni ng disabilities. More specifically, it was hypothesized that students with learning disabilities would be less calibrated than typical low and average achieving students, in the di rection of overconfidence. Additionally, it was hypothesized that students of higher achie vement levels would be more accurately calibrated than lower achievement levels. One-way ANCOVA procedures were used to determine group differences along the dimensi ons described. Reading self-concept acted as covariate because it was thought to be em bedded within the four groupings and was expected to contribute to calibration levels. Data analyses began with a screening of the data for differences based on gender, race, and grade. There were no significant gr oup differences by gender or race for any of the variables. As anticipated, significant differences by grade were found for reading comprehension (F(1, 156) =4.12, p = .04, 2= .03). Table 1 presents descriptive statistics across grade, learning disability status, and achievement grouping.

PAGE 29

22 Table 1. Mean and Standard Deviation for all measures by grade, learning disability status, and achievement level. Group (N) ASQQI Mean Total SE Total RC Score Mean Bias Mean Accuracy Total Item Accuracy Grade 4th (98) 5.94(1.62) 88.00(10.73) 21.22(4.71)* .92(.65) 1.46(.52) 10.68(2.55) 5th (71) 5.77(1.47) 92.58(9.54) 22.89(4.72)* .91(.66) 1.42(.50) 11.10(2.35) SLD Typical (144) 6.03(1.45)* 90.74(10.06)* 22.68(4.40)* .86(.62)* 1.40(.50)* 11.03(2.45) SLD (25) 4.94(1.83)*cd 85.24(11.69)*d 17.56(4.56)*cd 1.26(.71)*cd 1.71(.51)*cd 9.84(2.41)d Ach. Low (31) 5.32(1.61)d 88.81(8.85)d 20.55(4.52)d 1.15(.62)c 1.64(.62)d 10.16(2.54)d Average (82) 5.93(1.45)ad 88.87(10.37)d 22.48(4.18)ad .75(.65)ab 1.38(.45)a 10.98(2.46) High (31) 7.01(1.16)abc 97.61(7.21)abc 25.35(3.54)abc .84(.44)ab 1.19(.40)ab 12.06(1.93)ab Note. High-achieving students were removed in comparisons of SLD and Typical. Note. * Denotes significant di fference at p < .05. Note. Subscripts denote significant differences found in post hoc Bonferonni pairwise comparisons at p < .05. a. Significantly different from LD b. Significantly different from Low c. Significantly different from Average d. Significantly different from High Reading Self-Concept Reading self-concept was compared by achie vement level using mean scores from the ASDQ-I. Bonferonni pairwise compar isons were used to determine post-hoc differences. Analyses showed that the high achieving group had significantly higher reading self-concept than all of the othe r groups. Additionally, the average achieving group had significantly higher reading self-c oncept than the learning disabled group. There were no significant differences between the low achieving and the learning

PAGE 30

23 disabled groups for reading self-concept. The ANOVA results were significant (F(3, 165) =11.76, p = .000, 2 = .18) for achievement group. Refer back to Table 1 for self-concept score comparisons. Reading Comprehension Total Scores Reading comprehension was compared by achievement level using total accuracy scores from the WIAT-II. The ANOVA results were significant (F(3, 165) =17.53, p = .000, 2= .24) for achievement group. Bonferonni pairwise comparisons were used to determine post-hoc differences. Analyses showed that the high achieving group had significantly higher reading comprehension sc ores than all other groups. The average achieving group, had significantly higher scores than the learning disabled group. There was no significant difference in reading comp rehension scores for the low achieving and learning disabled group. Refer back to Table 1 for significance levels for reading comprehension score comparisons by group. Reading Self-Efficacy Total Scores Reading Self-Efficacy was compared by ach ievement level using the sum of all self-efficacy ratings from the WIAT-II. The ANOVA results were significant (F(3, 165) =8.69, p = .00, 2 = .14) for achievement group. Bonf eronni pairwise comparisons were used to determine post-hoc differences. Anal yses showed that the high achieving group had significantly higher self-efficacy scores th an all of the other groups. There were no significant differences for self-efficacy across th e other groups. Refer back to Table 1 for self-efficacy score comparisons.

PAGE 31

24 Group Differences in Calibration Use of the Covariate: Reading Self-Concept When examining the correlation betw een reading self-efficacy and reading comprehension, findings revealed significan t results (r= .3, p=.000) , suggesting students with higher self-concept also had higher sel f-efficacy predictions. Because reading selfconcept was expected to predict calibration, due to its correlation with self-efficacy, it was included for analyses as a covariate fo r the three calibration scores. Reading selfconcept, as measured by the ASDQ-I did not significantly predict any of the calibration differences. See Table 2 for results. Table 2. Significance levels for Reading Self-Concept as a covariate. Measure of Calibration Reading Self-Concept Mean Bias F (3,165) = 3. 79, p = .053 Mean Accuracy F (3, 165) = 1.10, p = .297 Total Item Accuracy F (3, 165) =.41, p= .521 Mean Bias Results from the ANOVA conducted for mean bias scores according to achievement level revealed significant differences (F (3,165) = 6.05, p = .001, 2 = .099). All groups were found to be overconfident. The average achieving group was the least biased ( X .75), followed by high achieving ( X .84), low achieving (mean = 1.15) and learning disabled ( X 1.26). A Bonferonni test wa s used to make pairwise comparisons for the results of the four gr oups. The average achieving students were significantly less biased than the low achieving (p .02) and learning disabled groups (p .003). The difference between the mean bias of the average and high achieving students was not significant. To make a co mparison of learning disabled and typical students, high achieving students were rem oved because there were no students with

PAGE 32

25 learning disabilities in that category. The ANOVA revealed significant differences (F (1, 167) =7.936, p = .001, 2 = .05) for learning disability st atus (see Table 1). That is, students with learning disabili ties demonstrated significantly more overconfidence than other low and average achieving students. Mean Accuracy The ANOVA revealed significant differences (F (3, 165) =7.37, p = .000, 2 =.118) for achievement levels on the mean accura cy scores. The high achieving group was found to be most accurate ( X 1.19), followed by average achieving ( X 1.38), low achieving ( X 1.64) and learning disabled ( X 1.71). A Bonferonni test was used to make pairwise comparisons for the mean accu racy across the four groups. It was found that the high achieving group was significantl y more accurate than the low achieving and learning disabled groups p .002 and p .001, respectively. The difference in accuracy for the high and average achieving groups wa s not significant. The average achieving group was significantly more accurate than the learning disabled group p .023. To make a comparison of learning disabled a nd typical students, high achieving students were removed because there were no students w ith learning disabilities in that category. The ANOVA revealed significant differences (F (1,167) =5.59, p = .005, 2 = .05) for learning disability status (s ee Table 1). That is, student s with learning disabilities demonstrated a significantly greater degree of inaccuracy than other low and average achieving students. Item Accuracy For total item accuracy th e ANOVA was significant (F (3, 165) = 5.12, p = .002, 2 = .118) for achievement levels on total item accuracy scores. The high achieving group

PAGE 33

26 was found to have the highest number of accurate predictions ( X 12.06), followed by average achieving ( X 10.98), low achieving (mean = 10.16) and learning disabled ( X 9.84). A Bonferonni test was used to make pairwise comparisons for the total item accuracy of the four groups. These analyses showed that the high achieving group predicted significantly more accurate item s than did the low achieving and learning disabled groups p \224 .012, and p \224 .004, respectively. There were no other significant differences for number of correct predicti ons. To make a comparison of learning disabled and typical students, high achieving students were removed because there were no students with learning disabilities in that category. Lastly, there were no significant differences found (F (1, 167) = 2.61, p = .077, K2 = .03) for learning di sability status (see Table 1). That is, students with learning disabilities did not de monstrate significantly fewer instances of accurately predicting their performance. .

PAGE 34

27 CHAPTER 5 DISCUSSION T his study explored the phenomenon of calibration in the area of reading comprehension for students of different achievement levels, with and without learning disabilities. The purpose of this study was to measure degree of calibrati on in readin g comprehension. Calibration i s defined as a match between perceived self efficacy for a given task and actual ability to perform that task. Students were divided by achievement level in reading based on performance on a group administered ac hievement test. They were also grouped by learning disability status, as identified in their educational records. Student s predicted their ability to answer reading comprehension questions correctly and then answered those questions, thus yielding a meas ure of achievement level and self efficacy . This chapter presents a discussion and interpretation of the results of the study as they relate to previous findings in the area of calibration, as well as the research questions and hypotheses. It also addr esses theoretical implications of the findings. The chapter concludes with a discussion of future directions for research and limitations of the current study. Calibration in Reading Comprehension The first goal of this study was to measure cali bration for the area of reading comprehension. For this study, students were asked to respond to reading comprehension questions from the Wechsler Individual Achievement Test II. After reading a statement or passage, students were asked to predict whethe r or not they could answer a specific

PAGE 35

28 question through confidence ratings. They then answered the questions. Calibration was measured through a comparison of their confidence rating and their actual performance. Calibration was compared for different ac hievement levels and for students with and without learning disabilities, using three different scores . Mean Bias The mean bias calibration score represented whether student s were underconfident or overconfident about their abilities. It was hypothesized that the learning disabled group would be most overconfident about their abilities. All groups were found to be overconfident about their abilities. This means that most students, across all achievement level s, predicted that they would perform better o n the comprehension questions than they actually did. The average achieving groups was found to be the least biased. This means that they displayed a more balanced pattern of responses (underconfidence and overconfidence) than did any of the other groups . As predicted, the learning disabled group was more biased toward overconfidence than the students without learning disabilities . This supports results from other calibration studies, and research on metacognitive skills of students with learning disabi lities (Klassen, 2002; Graham & Harris, 1989; Shell et al., 1995) . They have more difficulty assessing their ability to perform tasks and typically judge that they can perform more than they are able. Mean Accuracy The mean accuracy score represents deg ree of inaccuracy in making predictions about performance. Mean accuracy is measured through the absolute value of mean bias scores, representing total inaccuracy, regardless of direction of judgment (underconfidence or overconfidence). For this measure of calibration, the high achieving

PAGE 36

29 students were found to be most accurate, followed by the average achieving, low achieving and learning disabled groups. This suggests that although they may have demonstrated more overconfident scores than the average ac hievers (see mean bias), their predictions were closer to their actual performance. They were overconfident, but still closer to predicting their actual abilities than the other three groups. For mean accuracy, the students with learning disabilities wer e significantly less accurate than students without learning disabilities . That is, the degree to which they were over and underconfident was much greater than students without learning disabilities . Similar to the mean bias results, this supports previou s findings that students with learning disabilities are less able to judge their performance than other students (Klassen, 2002; Shell et al., 1995) . Total Item Accuracy The total item accuracy calibration score represent the number of times that studen t s accurately predicted their performance (whether correct, incorrect, or unsure). The high achievers made the most correct predictions followed by the average achievers, the low achievers and the students with learning disabilities . The high achieving s tudents correctly predicted their performance significantly more times than did the low achievers and the students with learning disabilities . Additionally, the students without learning disabilities made significantly more accurate predictions than did t he students with learning disabilities . Summary of Calibration Results In reading comprehension, all students were found to be overconfident. This is consistent with findings that most students are generally overconfident in their abilities (Pajares & Miller, 1994). High achieving students were the most accurate in terms of

PAGE 37

30 degree of overconfidence, and in total number of accurate predictions. Lower achieving and students with learning disabilities were less able to predict when they would be able to answer questions correctly. One explanation for this outcome is that students are more accurate in predicting their performance on easy tasks ( Schraw, Potenza, & Nebelsick Gullet, 1993). Although the questions included a range of easy and difficult quest ions, there were more easy questions for the higher achieving students. Across all three measures of calibration, comparisons of non learning disabled low and average achieving students, to students with learning disabilities found that students with lear ning disabilities are less well calibrated. They are more biased toward overconfidence and had fewer accurate predictions of their performance, even when achievement level was controlled for by removing the high achieving students. This supports previou s research findings that students with learning disabilities are less well calibrated than other students, and have poorer metacognitive abilities than other students (Shell, 1995; Butler, 1998). Reading Self Concept An interesting finding from this study pertains to the potential relationship between domain specific self concept and self efficacy constructs. Bandura’s conceptualization of self concept is as a global perception of the self (1986). However, it has been suggested that this broad construct is developed out of more specific perceptions of the self, such as self efficacy (Bandalos, Yates, & Thorndike Christ, 1995). Previous research has found that self concept for a specific area contributes to self efficacy because students base their perfor mance p redictions on how they have performed in that area in the past (Bandalos, Yates, & Thorndike Christ, 1995). For these reasons, it was predicted that reading self concept would predict self efficacy judgments and would contribute to

PAGE 38

31 calibration. St udents with high reading self concept were expected to be overconfident. Positive self concept would then predict accurate calibration for those who actually had high achievement . Those with lower achievement and higher self concept would be more overcon fident. Students with lower self concept were expected to be underconfident if they had high achievement , and accurate if they had lower achievement . Students were administered a reading self concept measure to test these hypotheses. Across all groups a nd all measures of calibration, reading self concept did not predict calibration. It was found that reading self concept was moderately correlated with self efficacy. However, this relationship did not affect calibration results. For these reasons, conc erns regarding lowering self concept through interventions to improve calibration may not be valid. Instead, s elf concept appears to be an independent construct from calibration. Implications Overall , all of the groups were found to be overconfident. Additionally, calibration was found to be independent of reading self concept. Bandura concluded that slight overconfidence is the best tool for academic persistence, motivation, and success. All students we re overconfident, so increasing confidence does not need to be a focus of intervention for calibration. Additionally, fears that lowering self efficacy to increase calibration will impact self concept may prove to be unfounded. The lack of relationship between self concept and calibration suggest tha t students are not basing their confidence judgments on self concept. Therefore, improving calibration can be attempted without detrimental effects on self esteem, optimism and effort. The groups that demonstrated the p oorest calibration were the low a chieving and learning disabled groups. Their overconfidence, as compared to their actual performance,

PAGE 39

32 is more extreme. For this reason, calibration for these groups may be an area of concern. In targeting interventions to improve calibration, the focu s should be on these students rather than high and average achieving students. Improving calibration for these students assumes certain qualities about classroom instruction for students who tend to struggle more. For calibration to be an effective tool, students must be performing at a level that is commensurate with their ability. For these students this necessitates some level of individualized instruction at their performance level. This is of course not always the case. The question then remains, what is the ideal level of calibration for a student who is struggling but truly does not have the capability to succeed in a classroom where instruction is targeted at students of average achievement levels? Further exploration of how calibration can be useful in different instructional settings might aide an understanding of this issue. Limitations of the Study Future explorations of calibration within the domain of reading comprehension may want to consider the limitations of this study. For example, one limitation to the research design was the identification of students with learning disabilities through educational records. Students’ records identified them as having a Specific Learning Disability but did not include in which academic area the stud ents disability fell. For the purposes of this study, it was assumed that most of these students identified as having a specific learning disability probably had a disability in reading. This decision reflected previous research indicating that the major ity of students with learning disabilities have a disability in the area of reading, or in reading and another area (Wong, 1995). However, this may not have been the case. Future studies may choose alternate means for identifying these students.

PAGE 40

33 Future studies may want to consider how the nature of the task, and the subsequent calibration scores, may differ for students of different achievement levels and disability status. For example, the questions used to elicit reading comprehension performance sco res ranged from 2 nd to 8 th grade achievement levels. This range was selected to ensure that all students would be capable of answering a set of questions but that no student would be able to answer all of the questions correctly. In examining the results , it was proposed that calibration scores for the high achieving students might be attenuated. This is because it is easier for students to accurately assess their own performance on “easy questions . ” In this design, t here were simply more questions the students in the high achieving group could be completely confident they would earn a correct score on than there were for students in the low achieving group. This may have contributed to their stronger calibration scores. A broader range of questions, a nd that are asked out of difficulty order might address this concern for future studies. However, even after modifying the range of questions, it may be that students with different ability and disability status approach the calibration task differently. For example, students who have a history of struggling with reading, may approach all reading tasks with the assumption of difficulty or failure based on their prior experiences of not being successful with the tasks adults give them. In contrast, stude nts who are more successful readers are likely to approach reading tasks with an optimism not only because they have been successful, but because adults expectations tend to match their actual skill level. By constraining the task such that all students ap proach the questions with a sense of ambiguity regarding their potential success may help to ensure calibration scores hold the same meaning for students across different achievement and disability

PAGE 41

34 groups. For example, presenting the task as challenging t o students in the higher achieving group, (e.g. “ These questions are high school level reading. ”) might force them to question whether they could answer all of the questions. Additionally, a task that appeared less like what students typically do in schoo l might encourage all students to approach the tasks without a presumed assumption of success or ability level. Final concerns relate to the contribution of other factors to explain efficacy judgments. Examiners noted that some students appeared to rate themselves as accurate in an attempt to please the examiners. Confounding variables such as a desire to please the examiner, or a desire to appear confident or smart may have biased how certain students responded. It is not known whether this type of be havior occurred more in some groups over others. However, these factors may have contributed to calibration scores. Future Directions for Research Until recently there has been little research on calibration in students with learning disabilities and in the domain of reading. This project represents the beginning of an exploration in both areas. Metacognitive skills such as the ability to assess one’s own performance should be further researched for populations with learning disabilities as this repres ents a major component of how and why these individuals struggle academically. Specifically, calibration can be further explored for these students for other areas, academic and otherwise. Assessing one’s own performance is critical in areas of social co mpetence, physical activities, communication, as well as academic areas. These are some of the many areas in which students with students with learning disabilities tend to struggle (Wong, 1995). Additionally, future studies might address how students pe rceive their performance on actual classroom tasks such as tests and homework to gain further understanding of how this phenomenon functions in context. Finally, an additional

PAGE 42

35 component that would be interesting to research, is the basis that students use to form their performance judgments. Some students may be motivated by outside factors that influence how they rate themselves. This might address some components that could bias efficacy judgments, such as desire to please an adult, competition with pe ers, personal satisfaction, actual appraisal of the questions etc. Measurement of these factors could occur throughout the efficacy judgments, embedded within the task, or they could be addressed after the task. There are many directions to be explored i n the area of calibration and related constructs. Interventions for Improving Calibration Several methods have been suggested by various researchers to improve calibration. Typically the strategies that are incorporated into interventions are not limite d to calibration, but would be beneficial in improving metacognition, self regulated learning, and other characteristics of strong students. They also tend not to emphasize any components that would run the risk of decreasing self efficacy or self esteem . Increasing Self Regulation Teaching students to set personal goals for performance helps them to take responsibility for learning (Stone, 2000). If they take ownership in the goals, the responsibility for learning is transferred from the teacher to the students increasing involvement in achievement outcomes (Henderson, 1986). Schraw, Potenza, and Nebelsick Gullet suggest that educators should look for new ways to promote self control such as encouraging self monitoring of improvement. It is also imp ortant for students to have an understanding of why they want to improve. To address this, s tudents can be asked to generate arguments for why they should improve (1993).

PAGE 43

36 Encouraging students to test their own knowledge can help them to process informat ion more deeply because they are asked to challenge their knowledge (Stone, 2000). They will learn to recognize what they do and do not know and what they need to continue to study. Self testing is a skill used by highly successful students but many stud ents do not naturally use this tactic and can be taught to use it more effectively (Stone, 2000). Instruction al Methods Teachers can construct texts or lessons so that test questions and examples are embedded with in them. This will help students to be a ble to predict material that will be included on a test . This will aide them in developing self tests and in knowing what to study ( Walczyk & Hall, 1989). Calibration is best when students are challenged in their tasks but they are able to reach attaina ble goals. Stone (2000) believes that a goal of approximately 80% correct is a level that will be effective for m otivation and improvement . Teachers can design tests to elicit approximately this success rate based on student knowledge. Students should b e encouraged to generate supporting and opposing arguments to support or refute their answers in order to encourage critical analysis of their thinking process (Schraw, Potenza, & Nebelsick Gullet, 1993). In one study, s tudents who generated opposing argu ments processed material more deeply and could demonstrate a more coherent understanding of the material than other students (Stone, 2000). Research has shown that incentives affect calibration of performance. Students rewarded with external incentives for accurate predictions not only made more accurate predictions, they also performed more successfully (Schraw, Potenza, & Nebelsick Gullet, 1993). There is some debate over whether feedback affects performance in the

PAGE 44

37 same way. Feedback from examiners d id not have the same effects as rewarding self generated feedback through incentives (Schraw, Potenza, & Nebelsick Gullet, 1993). Stone found that feedback appeared to play a prominent role in improving calibration (Stone, 2000). However, across studies, feedback had to address the process of learning and problem solving rather than just the outcomes. Feedback on the process and the outcome was effective in reducing overconfidence and increasing calibration on harder tasks, particularly for older childre n (Newman & Wick, 1987). Process feedback helps students to realize what strategies can be used in order to reach goals which enhances calibration (Newman & Wick, 1987).

PAGE 45

38 APPENDIX A ACADEMIC SELF DESCRIPTION QUESTIONNAIRE-I ASDQ-I : READING Name _____________________________ Grade _____________ Examiner __________________________ Date ______________ This is a chance for you to consider how you think and feel about yourself. This is not a test. There are no right or wrong answers, and everyone w ill have different answers. Be sure that your answers show how you feel about yourself. We w ill keep you answers private and not show them to anyone. On this page are a series of statements that are more or less true (or more or less false) descriptions of you. Please use the following 8 poin t scale to indicate how true or false each item is as a description of you. Respond to the items as you now feel even if you felt differently at some other time in your life. Try to avoid leaving any items blank. 1 Definitely False 2 False 3 Mostly False 4 More False than True 5 More True than False 6 Mostly True 7 True 8 Definitely True ____ 1. I have always done well in Reading classes. ____ 2. Work in Reading classes is easy for me. ____ 3. Compared to others my age, I am good at Reading classes. ____ 4. I get good grades in Reading classes. ____ 5. Work in reading classes is easy for me. ____ 6. I learn things quickly in Reading. ____ 7. I am hopeless when it co mes to Reading classes. ____ 8. It is important for me to do well in Reading classes. ____ 9. I am satisfied with how well I do in Reading classes.

PAGE 46

39 APPENDIX B WIAT-II RECORD FORM # __________ Reading Comprehension and Self Efficacy Sample Record Form Name ___________________________________ Date ______________ Examiner ________________________________ Grade _____________ Verbal Instructions: I am going to ask you some reading questions . Some might be easy for you and others might be hard for you, you probably won’t kno w the answer to all of the questions. Before each question, I am going to ask you to tell me how confident you are that you can give me the right answer to a question. Do you know what confident means? [explain confident:] Confident is how sure you are about something. For example, I am confident that fire is hot, but I am not confident that it will always be sunny outside. Can you tell me some things you are confident about? [explain further if student cannot think of anything or does not mention appropriate things] You are going to tell me how confident you are that you can answer correctly on a scale from 1 to 6. Look at this scale, see how it says “Not confident at all” at the 1 and “Extremely Confident” at the 6? When I ask how confident you are, you say or point to the number that says how confident you are. [read all choices and point to numbers] It’s okay to tell me that you don’t think you will know the answer. After that part, you are going to try to answer the questio n. I cannot tell you if you get the question right or wrong. Do you understand?

PAGE 47

40 Examiner Notes: On read-aloud items, allow the particip ant to read before asking the second part of the question. Write questionable responses in Notes section Follow all instructions and note query items (Q) WIAT Item # Sample Self Efficacy Question -as question is read, point to Likert card SE Rating Sample Actual Question write appropriate points RC Score Read these sentences aloud. How confident are you, from 1 to 6, that you can point to ______? Okay, now point to the picture that matches the sentences. Read these sentences aloud. How confident are you, from 1 to 6, that you can tell me ____? Okay, now tell me __________. How confident are you, from 1 to 6, that you can read this sentence and do what it says? Okay, now read this sentence aloud and do what it says. Read this aloud. How confident are you, from 1 to 6, that you can tell me what ____ means? Okay, now tell me what _____ means in the last sentence. Read this aloud. How confident are you, from 1 to 6, that you can tell me why ______ occurred? Okay, now tell me why _____ occurred in the story. Read this aloud. How confident are you, from 1 to 6, that you can predict what might happen as a result of the event described? Okay, what would you predict might happen as a result of the event described? Notes: ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________

PAGE 48

41 1 2 3 4 5 6 Not Really Confident (I might get the answer wrong) Somewhat Confident (I might get the answer right) Extremely confident (I know theanswer) Not at All Confident (I don't know the answer) APPENDIX C LIKERT SCALE

PAGE 49

42 LIST OF REFERENCES Bandalos, D., Yates, K, & Thorndike-Christ, T. (1995). Effects of math self-concept, perceived self-efficacy, and attributions for failure and success on test anxiety. Journal of Educational Psychology, 87, 611-623. Bandura, A. (1986) Social foundations of thought and acti on: A social cognitive theory. Englewood Cliffs, NJ: Prentice Hall. Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman. Blumfield, P.C., Pintrich, P.R., Meece, J., & We ssels, K. (1982). The formation and role of self perceptions of ability in elementary classrooms. The Elementary School Journal 82(5) . 401-420. Bouffard, T., Markovits, H., Vezeau, C., Bois vert, M., & Dumas, C. (1998). The relation between accuracy of self-percep tion and cognitive development. British Journal of Educational Psychology, 68, 321-330. Butler, D.L. (1998). Metacognition and l earning disabilities. In Wong, B. (Ed.), Learning about learning disabilities (pp. 277-301). San Diego, CA: Academic Press. Butler, D.L., & Winne, P.H. (1995). Feedback and self-regula ted learning: A theoretical synthesis. Review of Educati onal Research, 65(3), 245-281. Graham, S. & Harris, K.R. (1989). Improvi ng learning disabled students’ skills at composing essays: Self-ins tructional strategy training. Exceptional Children, 56, 3, 201-214. Henderson, R.W. (1986). Self-regulated lear ning: Implications for the design of instructional media. Contemporary Educational Psychology, 11, 405-427. Klassen, R. (2002a). A question of calibration: A review of the self -efficacy beliefs of students with learning disabilities. Learning Disability Quarterly, 25 , 88-103. Klassen, R. (2002b). Writing in early adolescenc e: A review of the role of self-efficacy beliefs. Educational Psychology Review, 14 , 173-203. Marsh, H. (1999). Academic self-description questionnaire I. Milpe rra, Australia: Selfconcept Enhancement and Learning Facilitation Research Center.

PAGE 50

Mastropieri, M.A., & Scruggs, T.E. (1997). Best practices in promoting reading comprehension in students with learning disabilities: 1976-1996. Remedial and Special Education, 18 , 197-213. Newman, R.S., & Wick, P.L. (1987). Effect of age, skill and performance feedback on childrenÂ’s judgments of confidence. Journal of Educational Psychology, 79, 115119. Pajares, F. (1996a). Self-efficacy beliefs a nd mathematical problem-solving of gifted students. Contemporary Educational Psychology 21 . 325-344. Pajares, F. (1996b). Self-efficacy beliefs in academic settings. Review of Educational Research 66(4) . 543-578. Pajares, F. (1997). Current di rections in self-efficacy re search. In M. Maehr & P.R. Pintrich (Eds.), Advances in motivation and achievement. Volume 10, (pp. 1-49). Greenwich, CT: JAI Press. Pajares, F., & Graham, L. (1999) Self-effi cacy, motivation constructs, and mathematics performance of entering middle school students. Contemporary Educational Psychology 24 . 124-139 Pajares, F., & Kranzler, J. (1995). Self-effi cacy beliefs and general mental ability in mathematical problem solving. Contemporary Educational Psychology 20 . 426443. Pressley, M. (1998). Reading instruction that works: The case for balanced teaching . New York: Guilford Press. (pp. 59-80). Sawyer, R., Graham, S., & Harris, K.R. (1992) . Direct teaching, stra tegy instruction, and strategy instruction with explicit self-r egulation: Effects on learning disabled studentsÂ’ compositions and self-efficacy. Journal of Educational Psychology, 84, 340-352. Schraw, G., Potenza, M.T., & Nebelsick-Gull et, L. (1993). Constraints on the Calibration of Performance. Contemporary Educational Psychology, 18, 455-463. Shell, D., Colvin, C., & Bruning, R. (1995) . Self-efficacy, attributions, and outcome expectancy mechanisms in reading a nd writing achievement: Grade-level and achievement-level differences. Journal of Educational Psychology, 87, 386-398. Stone, N.J. (2000). Exploring the Relations hip between calibrati on and self-regulated learning. Educational Psychol ogy Review, 12(4), 437-475. Wechsler Individual Achievement Test, 2nd Edition. (2001). San Antonio, TX: The Psychological Corporation.

PAGE 51

44 Walczyk, J.J. & Hall, V.C. (1989). Effects of examples and embedded questions on the accuracy of comprehension self-assessments. Journal of Educational Psychology, 81, 435-437. Wong, B. (Ed.). (1995). Learning about learning disabilities . San Diego, CA: Academic Press.

PAGE 52

45 BIOGRAPHICAL SKETCH Tamara Molitch-Hou was born in Boston, Massachusetts. She moved to River Forest, IL, in 1984 and considers herself a “Chicagoan.” She has two younger brothers, Ethan and Michael Molitch-Hou. Tamara attended Emory University in Atlanta, GA, for her undergraduate studies. There, she gradua ted magna cum laude w ith a Bachelor of Arts in psychology in 2000. She minored in li nguistics and Italian studies. Tamara has fulfilled her desire to work with children in many capacities includi ng tutoring, acting as a Hepzibah “Big Sister,” as a paraprof essional in exceptional student education classrooms, as a summer camp counselor, as a docent naturalist at th e Florida Museum of Natural History, and now through her choice of career. She became interested in working with disadvantaged and disabled chil dren when working as the director of an after-school program at the Ni cholas House, a family shelter in Atlanta. There she realized the profound imp act that attention, care and enthusiasm could have in children’s lives. She began graduate study in the School Psychology Program in the fall of 2001. There she developed the skills and interests that will allow her to work as a school psychologist in the public school system. She will complete the School Psychology Program in May of 2004 after completing a ye ar of internship w ith Hillsborough County Schools.