|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 MEMORY SELF EFFICACY AND STEREOT YPE EFFECTS IN AGING By BENJAMIN ANDR BENSADON A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010
2 2010 Benjamin Andr Bensadon
3 To my Grandmother
4 ACKNOWLEDGMENTS I thank the members of my supervisory committee, Drs. Kenneth G. Rice, Bonnie Moradi, and Tanya Koropeckyj Cox, for their constructive feedback and helpful input, and my mentor Dr. Robin L. West, whose guidance, support, mentorship, and editorial assistance, have played an integral role in enabling me to complete this study.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGU RES ................................ ................................ ................................ .......... 8 ABSTRACT ................................ ................................ ................................ ..................... 9 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 11 Backgr ound ................................ ................................ ................................ ............. 11 Self Stereotyping Theory ................................ ................................ ........................ 13 Stereotype Threat ................................ ................................ ................................ ... 16 Mediators and Moderators ................................ ................................ ...................... 24 Memory Self Efficacy ................................ ................................ .............................. 29 2 CURRENT STUDY ................................ ................................ ................................ 36 Hypothesis 1 ................................ ................................ ................................ ........... 37 Hypothesis 2a ................................ ................................ ................................ ......... 38 Hypothesis 2b ................................ ................................ ................................ ......... 38 Hypothesis 3 ................................ ................................ ................................ ........... 39 Hypothesis 4 ................................ ................................ ................................ ........... 39 3 METHODS AND PROCEDURES ................................ ................................ ........... 41 Over all Design ................................ ................................ ................................ ........ 41 Participant Information ................................ ................................ ............................ 41 Cognitive Function ................................ ................................ ................................ .. 42 Memory Anxiety ................................ ................................ ................................ ...... 43 Memory Self Ef ficacy ................................ ................................ .............................. 44 Stereotype Effects ................................ ................................ ................................ .. 44 Implicit Activation ................................ ................................ .............................. 45 Explicit Activation ................................ ................................ ............................. 46 Memory Performance ................................ ................................ ............................. 47 Explorator y Measures ................................ ................................ ............................. 48 Manipulation Check ................................ ................................ ................................ 50 4 RESULTS ................................ ................................ ................................ ............... 52 Overview ................................ ................................ ................................ ................. 52 Hypothesis 1 ................................ ................................ ................................ ........... 54
6 Hypothesis 2a ................................ ................................ ................................ ......... 55 Hypothesis 2b ................................ ................................ ................................ ......... 56 Hypothesis 3 ................................ ................................ ................................ ........... 56 Hypothesis 4 (St ructural Equation Model) ................................ ............................... 57 Exploratory Analyses ................................ ................................ .............................. 59 NS = p >.05 ................................ ................................ ................................ ............. 67 5 DISCUSSION ................................ ................................ ................................ ......... 71 Age ................................ ................................ ................................ ......................... 71 Stereotype Condition ................................ ................................ .............................. 75 Memory Self Efficacy ................................ ................................ .............................. 81 Memory Anxiety ................................ ................................ ................................ ...... 83 Exploratory Analyses ................................ ................................ .............................. 85 Study Limitations ................................ ................................ ................................ .... 90 6 CONCLUSION ................................ ................................ ................................ ........ 93 Clinical Implications ................................ ................................ ................................ 93 Theore tical Implications ................................ ................................ .......................... 95 Research Implications ................................ ................................ ............................. 96 APPENDIX A RECALL LIST (POSITIVE) ................................ ................................ ................... 100 B RECALL LIST (NEGATIVE) ................................ ................................ .................. 101 C RECALL LIST (AGING NEUTRAL) ................................ ................................ ...... 102 D PHONE SURVEY SCRIPT ................................ ................................ ................... 103 LIST OF REFERENCES ................................ ................................ ............................. 104 BIOGRAPH ICAL SKETCH ................................ ................................ .......................... 110
7 LIST OF TABLES Table page 3 1 Sample Characteristics ................................ ................................ ....................... 51 3 2 Exploratory Items ................................ ................................ ................................ 51 4 1 Baseline Subjective Age as a Function of Condition and Age ............................ 62 4 2 Baseline Memory Domain Identification as a function of Condition and Age ...... 62 4 3 Baseline Anxiety as a Function of Condition and Age ................................ ........ 63 4 4 Memory Performance as a Function of Condition and Age ................................ 64 4 5 Self efficacy as a Function of Condition and Age ................................ ............... 64 4 6 Anxiety as a Function of Condition and Age ................................ ....................... 65 4 7 Manipulation Check as a Function of Condition and Age ................................ ... 65 4 8 Correlations Between Predictor and Memory Performance Variables ................ 66 4 9 Path Coefficients as a Function of Age ................................ ............................... 67
8 LIST OF FIGURES Figure page 2 1 Model of hypothesized interrelations among variables when positive or negative stereotypes are activated ................................ ................................ ..... 40 4 1 Measurement Model ................................ ................................ ........................... 68 4 2 Predicted Model with Younger and Older Adults ................................ ................ 68 4 3 Best Model with Younger and Older Adults. Note: path coefficients are listed in Table 4 9. ................................ ................................ ................................ ........ 69 4 4 Final Model with Older Adults Only ................................ ................................ ..... 69 4 5 Age Differences in Memory Anxiety Change After Baseline ............................... 70
9 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MEMORY SELF EFFICACY AND STEREOTYPE EFFECTS IN AGING Benjamin Andr Bensadon December 2010 Chair: Kenneth G. Rice Cochair: Robin L. West Major: Counseling Psychology Stereotypes are oversimplified generalizations about people that are taken to be veridical P sychologists have explored their influence on behavior, including task performance. Recently, this research has incorporated aging stereotypes (e.g. wise, frail) Regarding behavior, several s tereotype theor ies ( e.g., Steele, 1997 ; Levy, 2003) have e xplicated this belief performance link. Empirically, s cholars have in duced or emphasized stereotypes and examined subsequent effects on performanc e, demonstrat ing such emphasis can threaten and lead to poorer performance by older adults. Another set of beliefs, self efficacy ( Bandura, 1997) has also been examined in relation to performance across various task domains including cognitive aging This work has focus ed on memory, given its increasing p ersonal relevance as people age, and memory self efficacy (MSE) As in other domains, r esults have shown that MSE can have a determining impact on memory performance Further, research has consistently shown significant age differences, both in memory performance and MSE, favoring the young Surprisingly, stereotypes and self efficacy, although both linked to performance outcomes, have rarely been examined together. Th e present investigation attemp ted to
10 fi ll this void by simultaneously examining aging stereotype s MSE, and memory performance. Older and younger adults were exposed to aging stereotypes (positive or negative) or a control condition, asked to rate their MSE, and then perform ed memory tasks. Memory related anxiety, perceived memory importance, and several other participant demographic s were also captured In analyses of variance examining the impact of stereotype condition and aging, aging showed significant ef fects on MSE and memory perform ance, with older adults obtaining lower scores. Unexpectedly, stereotype c ondition did not affect MSE or performance R ecall scores were comparable across negative ste reotype, positive stereotype, and control conditions. Structural equation models revea led only weak relationships among these variables for younger adults. For older adults, however, MSE predicted memory performance and anxiety exerted an indirect e ffect on performance by influencing MSE For both young and old anxiety was comparable at baseline and increased significantly during the study. This effect was stronger for older adults. Cumulatively th ese results illustrate that memory has increasing personal relevance as people age, and underscore the key role s of memory specific anxie ty and self confidence (e.g., MSE) in predicting memory performance.
11 CHAPTER 1 INTRODUCTION Background is, people categorizing themselves and/or others into groups, often v ia stereotyping (e.g., Fisk & Taylor, 1991), is an economic and efficient way to organize a complex world. One such category for grouping individuals is age. Throughout the modern world, societies in which youth is emphasized or even glamorized, such a s the United States, are fraught with age related biases (Kite & Johnson, 1988; Kite et al., 2005). Ageism, a term originally coined nearly three decades ago by Butler (1969), refers to evaluation of someone based solely on age. Focusing on the potential ly negative the most condoned, institutionalized forms of prejudice in the world especially in the United States ay also be heavy costs to such heuristics. Levy (2009) has posited ageism may be qualitatively different fro m other forms of discrimination. In es sence, other forms of prejudice, such as racist or sexist messages may target someone from birth, thus allowing for or perhaps necessitating coping strategies, often adopted from other in group members, as a way to defend against them given their perceived self relevance. Because people do not r in life, they may be relatively unprepared to resist the effects of ageist stereotypes, which may make such messages more easily internalized, and all the more powerful Interestingly, empirical results do reflect the existence of multiple stereotypes of
12 stereotype valence appears mixed (e.g., Hummert, 1990; Hummert et al., 1995). In other words, such stereotypes may not be exclusively negative. Both young and old endorse positive (i.e., nurturing) and negative (i.e., forgetful) traits as typical of older adults. Though this balance of stereotypes may appear comforting, age based negative stereotypes are dominant in many youth focused nations (e.g., Wentura & Brand stdter, 2003; Desrichard & Kopetz, 2005). The past two decades have witnessed mounting empirical interest in the relationship between stereotyped beliefs and subsequent behavior. Theoretically, it stands to reason that negative stereotypes about older adult memory could inhibit or be detrimental to performance (e.g., Steele, 1997; Hess et al., 2003), but empirical findings in support of this relationship have been As discussed later in this paper, some possible re asons for these mixed results may include inconsistent operationalization and conceptualization of precisely how stereotypes should be activated, what influence they exert on subsequent cognitive performance, mechanisms through which this impact occurs, an d what determines whether or not stereotype activation leads to an effect at all. To address this complexity, researchers have developed theoretical models to explain the impact of stereotypes on performance. Arguably t he two foremost theoretical expla nations for the stereotype performance link are self stereotyping theory (e.g., Levy, 2003) and the concept of stereotype threat (e.g., Steele, 1997). A self stereotype is internalized, can be activated either with or without conscious awareness, and occu rs in response to both positive and negative stereotypes. Stereotype threat is not internalized, occurs only with awareness (i.e., consciously) and in response to
13 negative stereotypes, and exerts its effect by increasing anxiety. For purposes of this disc ussion, the impact of stereotypes on performance, from either perspective, will be called a stereotype effect. Both theories will now be consider ed in greater detail. Self Stereotyping Theory Early research focusing on stereotypes aimed to explicate their influence on perception (Jost & Banaji, 1994). More recently, however, increased scientific attention has focused on the latter. One of th e most consistent and methodologica lly diverse programs of research aimed at self perception in aging has been that of Levy (1996; 2000; 2003) and colleagues (e.g., Levy & Langer, 1994, Levy et al., 2000). According to Levy (2003), aging self stereotypes are distinguished by three characte ristic elements: They originate in the form of aging stereotypes as early as childhood and are reinforced in adulthood. Aging self stereotypes, as well as aging stereotypes, can operate below awareness (unconsciously). In old age, aging stereotypes become aging self stereotypes Freudeman, West, & Viverito, 2006), recent research has examined the impact of agin g stereotypes on older adult memory performance (e.g., Levy, 2003; Hess et a l., 2007; memory, Levy and Langer (1994) compared aging attitudes and memory performance in three cultures presumed to have contrasting valenced views of aging -Amer ican (negative view), American deaf (more neutral view), and Chinese (positive view). Four types of memory performance were measured; immediate, learned, delayed, and
14 probed recall. When recruited, participants were not told that the study consisted of memory tests. adults were found to hold the most favorable views of aging followed by the American deaf and then American hearing participants. Recall perform ance mirror ed this order, that is, Chinese performed best, followed by deaf and then hearing Americans. This result was true across all four recall tasks. Though an impressive result, t he precise mechanisms through which this occurred were not elucidated a recurrent theme in stereotype literature. The authors posited that internalization of differential beliefs in the two cultures explained their results (Levy & Langer, 1994). Consistent with other work exa mining the potential malleability of stereotypes (e.g., Blair, 2002; Guo, Erber, & Szuchman, 1999), Levy (1996) attempted to clarify these cross cultural findings (Levy & Langer, 1994) and more directly illustrate the possibility of performance benefits fo r positive stereotypes. She measured performance both before and after the use of either negative or positive (e.g., wise, sage) primes. Priming occurred via both explicit and implicit manipulations and results showed that onl y the latter affected memory performance. Consistent with her hypotheses, overall, negative primes reduced, whereas positive primes increased emphasis on personal relevance. To the young participants, the aging stereotypes were not considered personally relevant and therefore did not lead to self stereotyping. In a similar vein, Yoon e t al. (2000) utilized a sample of English speaking Canadians and recent Chinese Canadian immigrants. Across age, Chinese immigrants
15 showed more favorable views toward the aged than the Anglophone non immigrant Canadians. Among younger adults, both groups (Chinese immigrant and non immigrant) showed comparable memory performance. Chinese older adults, however, performed significantly better than non immigrant older adults, and both groups of older adults performed worse than their younger counterparts on most of the memory tasks, consistent with Levy and Langer (1994). The more positive views toward aging and smaller age differences in performance among those of Chinese origin, suggest a potentially important impact of cultural or societal views on perfor mance. However, this differences, nor were such differences consistent across all the memory tests. This arlier work (Yoon et al., 2000). Fields, and Hertzog (2002) included a neutral prime condition in addition to the positive and negative age stereotypes. Consistent with work, negative primes led to decreased performance among older adult participants and there were no significant priming a positive age stereotype did not increase older adu The authors attributed this to their small sample size and limited statistical power (Stein et al., 2002). Other empirical work has in fact shown the potential performance boost of positive stereotype effects, in some cases remov ing significant age differences in memory performance among young and older adults (e.g., Hess et al., 2003).
16 Perhaps the most convincing support for self stereotyping has been its broad reach across various functional domains beyond memory (e.g. Levy, 20 03). Some of this work has examined and empirically supported the relationships between older adult negative self stereotyping and increased physiological response to stress (Levy et al., 2000), poorer handwriting (Levy, 2000), and slowed gait speed (Haus dorff, Levy, & Wei, 1999). Stereotype Threat An alternative to self (e.g., Steele & Aronson, 1995 ; Steele, 1997). Like self stereotyping, th i s theory emphasizes the personal or self relevan ce of the negative stereotype. However, this approach emphasizes negative stereotypes only and frames stereotype effects in terms of stereotype threat. For threat to occur and actually impact performance, according to Steele (1997), the assessment context must be perceived as diagnostic of the stereotyped ability, and the individual must also attach high importance to this ability (e.g., domain identification ; Steele, 1997). Theoretically, these conditions are both necessary and sufficient for performance to be affected. Some empirical evidence has also illustrated such a pattern (e.g., Hess et al., 2003). Less clear, however, has been empirical support for the theoretically relevant mechanisms (e.g., anxiety, motivation) ubsequent performance. recognition that a negative group stereotype could apply to ones elf in a given
17 most with a particular domain are most likely to consider that domain self definitional and consequently are most vulnerable to threat. Further, the theory speculates that increases in perceived threat may relate to anxiety and motivation which in turn can impair performance. In fact, Steele and colleagues (e.g., Steele, 1997; Spencer et al., 199 9 ; Steele & Aronson, 1995) have demonstrated this empirically with gender related stereotypes and math performance, and racial stereotypes and academic achievement. I now consider empirical studies in aging conducted in accordance with the stereotype threat theoretical framework. In a study manipulating task instructions, Chasteen et al. (2005) attempted to both elicit and then directly measure stereotype threat. The authors compared two tasks, one memoriza tion and the other impression formation. Participants in the memorization mpression that for ol der adult participants, the memorization condition would induce threat, consistent with aging stereotypes about memory decline, and lead to age related differences in recall performance. Measures of perceived threat, anxiety, and self efficacy were also i ncluded in this study. Consistent with their hypotheses, the authors did find age differences in recall, and stereotype threat did mediate the relationship between age and memory performance, with older adults perceiving greater aging and memory related s tereotype threat than the younger participants. In addition, anxiety was higher for those in the memorization condition across all ages, but it did not mediate the relation between instruction type and performance. Self efficacy did not
18 mediate the link between instruction type and memory performance although those in the memorization condition reported lower self efficacy (Chasteen et al., 2005). Desrichard and Kopetz (2005) employed a similar design and obtained a comparable pattern of results. Tasks were described as either orientation or memory based. Older adult memory performance was significantly worse when instructions emphasized the memorial component Although the authors did not measure anxiety, they did offer the possibility that their memo ry emphasis condition may have induced a high level of anxiety related to memory stereotypes of the elderly. Such anxiety, the authors speculated, may have impacted those lowest in memory self efficacy, and mediated the decrease in observed task performan ce (Desrichard & Kopetz, 2005). Perhaps most prolific in conducting premise have been Hess and colleagues (e.g., Hess et al., 2003; 2004; 2007). Hess et al. (2003) elicited stereotype effects through the use of four fictitious read prior to recall. ability as able need to become dependent on others due to diminishing skills. This was the negative stereotype condition. The other two analogous reports illustrated more positive discoveries about the relationship between memory and aging. To then measure and con firm that stereotypes were activated, participants viewed word pairs on a computer monitor each consisting of a prime (young or old) and a trait. They were then told to ignore the first word (prime) and rate as good or bad the second word in each pair. P articipant responses were timed. Faster judgments of negative traits, and slower judgments of
19 considered evidence of automatic (implicit) activation of stereotypes. The auth ors consistent words, which served as stereotype relevant cues, thus facilitating more rapid retrieval. In addition, prior to entering the test environment, Hess et al. (2 003) had participants complete the Metamemory in Adulthood Achievement (MIA Ach; Dixon & Hultsch, 1984) subscale which assesses value that individuals place on their memory ability, and the Metamemory in Adulthood Anxiety (MIA Anx ; Dixon & Hultsch, 1984) subscale, an explicit measure of anxiety associated with memory performance. Memory performance was subsequently assessed via a free recall task using a 30 item list of moderate to high frequency words from six semantic categories. Strategy use was measu red by clustering at recall. What did the authors find? Older adults who read negative reports about aging and memory had poorer recall while those who read positive reports had better recall, a finding illustrative of the typical impact of threat. Also consistent with stereotype threat theory (Steele, 1997), performance was most negatively affected for older adults in the negative condition who also expressed investment in their memory ability. No significant effects were observed for anxiety, contrary In another study, Hess et al. (2004) provided participants with sets of five words and asked them to complete a scrambled sentence task. For each set, participants had to construct a sentence with four of the five words. To activate stereotypes, the words were organized according to stereotypicality (age stereotype relevant) and valence
20 (positive or negative). Aging negative (e.g., cranky), positive (e.g., distinguished), and neutral (e.g., book) words w ere printed on index cards. In addition to condition assignment based on these words, participants were further divided into aware and unaware conditions. Index cards in the aware condition contained words highlighted in yellow, and participants were tol d that two thirds of these words contained a trait often associated with views of older people, and they were instructed to use and underline these words in forming their sentences. Participants in the unaware condition did not receive such instructions n or were their index cards highlighted. After this task, all participants were given a memory recall test. What did the authors find? As hypothes ized older adults performed worse when exposed to negative than to positive stereotype primes This findin g was strongest for those in the unaware condition. Performance of participants in the aware condition was not significantly affected by prime valence (positive vs. negative) A mong participants exposed to positive stereotypes, performance was comparable across age. Y ounger adult pe rformance was not significantly affected by exposure to aging stereotypes (Hess et al., 2004). In a study utilizing an exclusively older adult sample, Hess et al. (2007) divided participants into two groups, young old (ages 60 to 70 years) and old old (71 to 82 years). All participants were informed that the study was examining predictors of memory performance. However, t he authors then manipulated task instructions such purpose was to better understand why younger participants had outperformed older adults in prior research,
21 free from age related bias, as evidenced by comparable scores between younger and condition were also asked to indicate their age on a piece of paper Using their previous memory recall task (Hess et al., 2003) the authors again found significant stereotype threat effects on performance. Interestingly, the most negative impact occurred for younger members of the sample (aged 60 to 70) and those with higher education levels (Hess et al., 2007). In a mixed age sample divided into young, middle aged, and older adults, Andreoletti and Lachman (2004) assigned participants to one of three experimental groups and then informed them about a memory test. One group (aging stereotype) was told that prior research had s hown that young adults generally outperform older adults on the test. The other group (counterstereotype) was told that prior research had shown no age differences on test performance and that all ages could remember the same number of words. In addition participants in both groups were told to pay attention to which test they were taking as they would be asked to let the researcher know later in the session. A third group received no information about performance and served as a control condition. Aft er recall, as a manipulation check, the authors assessed how accurate participants were in recalling the information given to them at session onset. The majority (91%) of participants correctly identified whether they had been given stereo t ype or counters tereotype information. Perhaps mo re importantly, and consistent with the studies above, performance was lower for those in the stereotype than the counterstereotype condition across all ages, for those with higher education. Participants lowest in educat ion, across both stereotype and
22 counterstereotype conditions, showed lower recall than the control group (Andreoletti & Lachman, 2004). This apparent moderating role of education will be discussed in more detail later. Considering the above literature, it is apparent that although self stereotyping (e.g., Levy, 2003) and stereotype threat (Steele, 1997) theories share some overlapping characteristics, they are nonetheless distinct Both theories attempt to account for the potentially negative impact of stereotypes on performance, and are applicable to older adults and cognitive performance (e.g., memory). They differ, however, in that self stereotyping theory (e.g., Levy, 2003) po sits both positive and negative stereotypes can have an impact, and this can occur consciously and/ or unconsciously, whereas stereotype threat, as the name suggests, focuses primarily on the negative (threatening) impact of negative stereotypes. In additi on, stereotype threat posits such effects are activated via conscious awareness, and will impact performance via increased anxiety, though it is clear from the articles reviewed above that empirical support for this has been mixed. In light of the differe nces between these two views one study attempted to directly compare the two theoretical approaches vis vis older adult memory performance self stereotyping accounts o f behavioral assimilation to age stereotypes and to case, refer red to actions or performance conforming to or consistent with a particular stereotype. In their s ample of late middle aged adults (48 62 years; M = 54), the
23 authors focused on the extent to which participants identified with the particular age group (i.e., middle aged vs. older adult). What did the authors find? Most relevant, and consistent with self stereotyping theory (Levy, 2003), participants who believed their performance was being compared cantly worse on word recall than those who were m no mention of a comparison group was made at all (i.e., control group). The authors interpreted this stereotypes about elders and memory loss, thus contributing to diminis hed more anxious while engaging in the recall tests. Though increased anxiety is addressed more directly by stereotype threat (e.g., Steele, 1997), the fact that this was purportedly stereotyping. Therefore, the authors concluded that their results yielded support for self stereotyping but not tical supremacy between self stereotyping and stereotype threat concepts, the authors suggested a third perspective, specifically in reference to the posited importance of stereotyped group membership. Activation of any stereotype they asserted, can pote ntially lead to behavior consistent with that stereotype, regardless of group identification. In fact, though a bit less popular than the other two theories, some empirical evidence does support this idea (e.g., Bargh, Chen, & Burroughs, 1996 ; Wheeler & P etty, 2001 ).
24 the key issue may be differential vulnerability, that is, those who actually identify with or feel they belong to the stereotyped group are most susceptible to the stereo type effects. Presumably, those for whom the stereotyped group is more personally relevant may have a lower threshold for threat, and thus are more likely to be powerfully affected by memory performance appears to be highly salient (Dark Freudeman, West, & Viverito, 2006), should constitute a group particularly susceptible to the impact of age related memory stereotypes and the concomitant anxiety associated with acti vities they perceive as diagnostic of memory ability. Barring some exceptions, this theoretical link between memory salience and concomitant vulnerability to aging stereotypes and their subsequent impact on memory performance, has been confirmed empirical ly as well (e.g., Levy, 1996; Stein, Blanchard Fields, & Hertzog, 2002). Mediators and Moderators The literature reviewed thus far suggests consensus is lacking, both theoretically and empirically, regarding the impact of stereotype effects on performan ce. It seems safe to conclude that positive, negative, conscious, and unconscious stereotype effects can have an impact on individuals and their subsequent performance. But is this impact direct or are other variables equally or even more relevant for ex plaining stereotype effects? Attempts to identify the precise mechanisms through which stereotypes exert their impact on performance have also proven inconclusive. Consistent with earlier, not everyone degree. Rather, most relevant are the individual differences in salience of a particular
25 domain identified, the situational relevance of the stereotype is threatening because it threatens diminishment in a domain that is self definitional...For the less domain identified, this recognition is less threatening or not threatening at all, because it threatens something that is less self the intuitive appeal of this view empirical support for the importance of domain identification has been mixed Among those who are most domain identified, Steele (1997) theorizes their performance may be compromised by situational, threat 168). Though he has found some empirical support for this theoretical claim in his own work f ocusing on gender and racial differences in achievement test scores (e.g., Spencer et al., 199 9 ; Steele & Aronson, 1995), evidence for the relevance of anxiety in aging literature has been mixed (e.g., Hess et al., 2007), with some studies even showing hig her anxiety levels among younger adults (e.g., Chasteen et al., 2005). In terms of cognitive aging, investigators have become increasingly aware of the need to empirically investigate these and other possible mediators and moderators relevant to the stereotype performance link (e.g., Chasteen et al., 2005 ; Shapiro & Neuberg, 2007 ). Recent empirical attempts to identify such mechanisms will be considered now. explored the potential moderating impact of age, anxiety, education, and concerns memory performance. The authors utilized a n exclusively older adult sample which they
26 dichotomized into yo ung old (60 70 yrs.) and old old (71 82 yrs.). Those in the threat condition were informed that the primary purpose of the study was to explain why young and old perform so differently on memory tests, and were then asked to indicate their age on a pi ece of paper. Those in the non threat condition were told the test they were taking was free of age related bias, so as to avoid any contaminating factors, and that adults of all ages had performed similarly in the past (Hess et al., 2007). Results show ed that age education, and stigma consciousness all moderated the impact of threat on performance Consistent with other cognitive aging studies (e.g., Neupert et al., 2006), those highest in education were most susceptible to threat. In addition, the i mpact of the other moderating variables was particularly evident in the yo unger part of their exclusively older adult sample. There are several plausible explanations for why this would be the case for the young old participants. Perhaps the transition f rom middle to adult life is a particularly sensitive time period. Perhaps those on the cusp of older adulthood place greater importance on their cognitive (memory, in this case) abilities and interpret any consciously perceived lapses as indicative of impact. Those who were exposed to threat made more cautious, conservative predictions of their performance. Why? Perhaps they were bracing themselves for potentially p oor performance, which may be a testament to the degree of perceived threat they felt The authors attempted to explain this result in terms of motivation and its influence on task approach (Hess et al., 2007 ). Specifically, positive stereotypes were pos ited to elicit a promotion focus, that is an attempt by participants to maximize performance even at the risk of errors. Conversely, negative stereotypes were thought
27 to lead to prevention focus, a more cautious approach that may sacrifice peak performance in favor of minimizing errors. Though this promotion/prevention dichotomy seems feasible and is rooted in prior literature (e.g., Seibt & Frster, 2004), in terms of motivation, an alternative explanation seems plausi ble. Among those exposed to negative stereotypes, increased perceived threat may lead to increased motivation for high performance. That is, participants who feel particularly threatened and by extension, challenged, can respond in one of two ways. They may be motivated to exert extra effort to succeed in spite of the challenge, or, they may feel unmotivated, if they feel that additional effort would be futile given stereotype s might lead to promotion. It is puzzling that this possibility was not strategy use (a possible indication of motivated effort) was determined to account for approximately 58% of the variance associated with stereotype threat effects in recall. Considering the theoretical importance Steele (1997) ascribes to anxiety, Hess et important. The author s attributed this to their methodology. Specifically, Hess et al. (2007) co ncluded that given the broad (non memory related) anxiety measure used, accurately captured re sponses to the working memory task. As cited earlier, in a study of young, middle aged, and older adults, Andreoletti and Lachman (2004) also considered the moderating impact of education. Specifically, the authors examined its potential influence on res ilience and susceptibility to effects of
28 memory aging stereotypes. Consistent with prior studies examining possible performance benefits of positive stereotypes (e.g., Levy, 2003), the authors hypothesized that memory recall would be higher among those wh o received counterstereotype information than those receiving either stereotype or unrelated However, their findings for the role of education directly contrasted th ose of Hess et al (2007). Among those low est in education, recall performance was weaker for both stereotype and counterstereotype information conditions, as compared to the control group. How did the authors interpret these results? They argued that t he more highly educated participants were not susceptible to negative stereotypes and were most able to benefit from positive (counter) stereotypes. Conversely, those with less education did not carefully process the distinguishing information presented b y experimenters. Taken together, the authors concluded, mere exposure to age based stereotype information, regardless of its valence (i.e., positive vs. negative), is sufficient to activate threat and diminished per formance (Andreoletti & Lachman, 2004). Data cited from the above studies demonstrates that in addition to effects of the stereotypes themselves, other variables may be important in determining their influence, positive and/or negative, on performance. The preponderance of evidence suggests o lder adult memory may be increasingly susceptible over time to stereotype effects, but is this applicable to all older adults ? Is performance affected equally? What role, if any, might individual differences play? Why may some individuals be more easily influenced by stereotype effects than others? These questions seem to lack empirical answers. In
29 fact, much of the theoretical work on stereotypes and cognition seems to address these nuances only peripherally. Aside from a few exceptions (e.g., Chaste en et al., 2005; Desrichard & Kopetz, 2005), another form of beliefs, self efficacy, has received minimal consideration in the extant literature examining stereotypes and cognitive aging. The relevance of memory self efficacy (MSE) to aging stereotype eff ects and memory performance will now be considered in detail. Memory Self Efficacy efficacy is defined as produ that people are active, agentic contributors, rather than solely determiners or recipients of what actually happens to them. People are simultaneously agents and objects at once acting on the environment and reflecting and acting on themselves. Self efficacy neither a global self amic and malleable, subject to changes in task demands, situational determinants, social context, and individual developmen construct is theoretically conceptualized and most empirically valid and measurable at the domain level (e.g., memory self efficacy; MSE). Self efficacy levels in one domain of function, such as public speaking, would not be expected to significantly correlate with performance in a different domain (e. g., physical exercise ). Does MSE vary by age? Indeed, the empirical data suggest that the answer is yes. In their comprehensive review of the literature, Berry & West (1993) found strong converging evidence for a poorer sense of MSE in older adults relative to young adults.
30 Thi s has been observed across several different types of measures including MSE factor scores derived from the Metamemory in Adulthood questionnaire (Cavanaugh & Poon, 1989; Hertzog, Dixon, & Hultsch, 1990; Hultsch, Hertzog, Dixon, & Davidson, 1988), single item self efficacy predictions for memory tasks (Hertzog et al., 1990 ; Rebok & Balcerak, 1989), and multilevel, task specific measures of MSE strength & level (Berry et al., 1989; West et al., 2002; West & Yassuda, 2004; West et al., 2005). What about th e relationship between MSE and performance? Perhaps the most compelling and consistent evidence among older adults has been found by West and her colleagues (e.g., West et al., 1984; Berry et al., 1989; West et al., 1992; Berry & West, 1993; Welch & West, 1995; West et al., 1996; West et al., 2001; West et al., 2003; West & Yassuda, 2004; West et. al., 2005 ; Bensadon et al., 2007; West et. al., 2007; West et al., 2008 ). S tudies exploring the impact of goal setti ng and feedback on memory have often found that older adults exhibit lower MSE and poorer memory performance th an their younger counterparts. But which comes first? Though some data have been correlational in nature, Bandura (1993) explains the beliefs performance relationship in terms of recipr ocal causation, that is, a bidirectional chain whose links include individuals, their environment, and their behavior. Others (e.g., Berry & West, 1993) have referred to this as a feedback loop where performances influence future judgments of self efficac y that, in turn, influence future performance. In spite of this bidirectionality, self efficacy theory tends to emphasize the preeminence of beliefs. Initially, it is posited, individuals hold beliefs about their capacity in particular domains, even pri or to performing relevant behaviors (Bandura, 199 3 ). Such beliefs derive from four main sources; enactive mastery experiences,
31 which serve as indicators of capability; vicarious experience, which alters efficacy beliefs via transmission of competence and comparison with attainments of others; verbal persuasion and social influences (including stereotypes) that communicate to the individual that he or she has certain capabilities; and physiological and affective states, from which individuals judge their ow n capability, strength, and vulnerability to dysfunction (Bandura, 1997). These beliefs are theorized to then have a direct impact on performance. Cervone and others (e.g., Cervone et al., 2001) have offered empirical support for this concept, demonstrating the unique contribution of self efficacy to performance, independent of ability and past performance a ccomplishments. This emphasis on beliefs is also somewhat consistent with memory training literature. Surprisingly few memory interventions have focused on enhancing self efficacy, opting instead to target performance and related strategies directly (Fl oyd & Scogin, 1997). Though they have had success in improving memory performance, such interventions have rarely been able to elicit accompanying (and enduring) changes in beliefs (e.g., Rapp et al., 2002; Woolverton et al., 2001). Guided 1997) theoretical principles, West, Bagwell, and Dark Freudeman (2008) were successful in developing a multifactorial training program designed to improve both memory self efficacy and performance. The program achieved these goals and the increased MSE le vels were found to predict memory performance post intervention. As stated above, i t seems that efficacy beliefs may be more predictive of performance than vice versa. Similar directional patterns have been evidenced in t he memory stereotype literature Older participants tend to display lower baseline memory performance than their younger counterparts (e.g., Stein, Blanchard Fields, & Hertzog,
32 2002). Further, virtually all of the memory stereotype literature cited thus far that has utilized mixed age s amples (i.e., both young and older adults), has found that activation of age adults (e.g., Levy, 1996; Chasteen et al., 2005). Though there are several theoretical explanations as to why this may occur, the MSE literature reviewed above suggests that memory as a domain is more salient for older adults, who often have lower levels of MSE (e.g., West & Yassuda, 2004). Similarly, the stereotype literature reviewed above (e.g., Steele 1997; Levy, 2003) suggests that the impact of stereotyping is greatest for those who identify with the stereotyped group and view the stereotyped domain as personally relevant. Taken together, it seems curious that most research examining the relationsh ip between stereotypes and memory performance, and MSE and memory performance has been conducted separately. The preponderance of evidence suggests that both types of beliefs, MSE and stereotypes, are relevant to subsequent performance. What appears unc lear, however, is which of these sets of beliefs, if either, carries more weight and is more influential vis vis memory performance. Strangely, the interplay of these constructs, at least by na me, has rarely been investigated. This may be a result of t he fact that work explicitly addressing stereotype effects on memory performance is still relatively nascent. Prior cognitive aging literature, both theoretical and empirical, that has addressed memory beliefs and their impact on performance, may have bee n examining stereotypes without necessarily identifying them as such. Studies by Dixon and Hultsch (1983) and Cavanaugh et al. (1983), have found that older adults say that memory is important to them, rate themselves as having an adequate memory, but sti ll report more
33 instances of memory failure than do younger adults. It seems reasonable that although these studies did not explicitly refer to (or measure) them by name, memory and aging iefs, particularly regarding their own memory self assessment and performance expectations. An every day example of this seems illustrative. An elderly woman meets someone she knows at a restaurant and is unable to recall ame. Aside from being embarrass ed, she may become exceedingly discouraged and unsure of herself, and perhaps due to stereotype threat, she may become anxious that she is than an isolated instance of forgetfulness, the older adult may suffer a reduction in self efficacy, which, in turn, may decrease the likelihood of attending that or any restaurant in the future, due to fear of similar embarrassment. As Bandura (1997) points out s themselves, come to activate a sense of incompetence that impairs future performance But where do so such belief systems and expectati ons come from? What is guiding them? Certainly actual decrement may occur, especially over time (e.g., Baltes, 1997; Freund & Ebner, 2005). Objective decline aside, stereotypes may also be contributing to these expectations of both self and others. Bec ause of the stereotypical expectation across ages that memory inevitably declines in late life, older adults may interpret each memory lapse as further evidence of age related cognitive decline (self stereotyping) while younger individuals free of such exp ectations can and do attribute the same lapses to other phenomena unrelated to
34 decline, such as low interest, stress, or distraction. Gerontologists have referred to this particu lar (Zarit et al., 1981; Erber, 1999; Erber et al., 1996; Erber et al., 1992; Erber & Rothberg, 1991; Erber et al., 1990). This coincides with most of the stereotype literature, both theoretical and empirical, which has consistently described and demonstr ated differential effects on performance across age. In spite of the potential theoretical relevance of self efficacy, surprisingly few studies of stereotype effects and memory aging have incorporated MSE (e.g., Chasteen (1997) theoretical conceptualization (e.g., Desrichard & Kopetz, 20 05). In their 2005 study, Desrichard and Kopetz manipulated task instructions and compared the subsequent memory performance of young and older adults. Tasks were explicitly presented as either memory related or non memory related. When instructions emp hasized the memory component of the task, older adult performance decreased along with MSE. This did not occur for younger adults. In addition, when instructions did not emphasize memory, this relationship was not significant. Further, the authors found that older adult performance expectations, a construct accounted for by efficacy theory, were sensitive to task instructions and acted as a partial mediator of the link between stereotype effects and performance. Once again, this was not significant for younger adults. In fact, for one memory task, explicit mention of the memory component was associated with increased performance expectations for the young. Overall, the authors interpreted their results as evidence for ster eotype threat, though they did not explicitly attempt to measure degree of threat.
35 These stereotypes and memory performance. instructions also sought to examine self efficacy as a potential mediator of stereotype threat effects on performance. As mentioned earlier, the authors did not utilize a specific, comprehensive MSE measure (e.g., MSEQ 4). Rather, a brief, general scale was unsure if I have the ability to do well on this task). Unlike much of the MSE literature cited above, there were no significant age differences in self efficacy, whic h calls into question the validity of their scale.
36 CHAPTER 2 CURRENT STUDY In spite of progress in understanding and disentangling the interrelationships among aging stereotypes, memory beliefs and performance, several gaps in the extant literature remain, particularly regarding variables that influence stereotype effects. The current research, pat terned in part after Hess et al. (2004), attempt ed to fill some of these gaps. The study ha d several primary aims. First, this study simultaneously examine d the effects of stereotypes on older adult memory self efficacy and memory performance. To the au do so. In addition, activation of both positive and negative stereotypes allow ed for the direct test of self stereoty pe effects on performance. As mentioned earlier, self efficacy theory postulates that poor performance in a specific domain results from low self efficacy. Stereotype threat theory, however, posits other variables such as domain identification and anxiet y as factors determining extent of threat and resulting performance decrement. The current study empirically test ed both theories. To reiterate, although recent articles investigating aging stereotype effects have begun to consider the self efficacy cons truct (e.g., Chasteen et al., 2005), only one ( see Desrichard & Kopetz, 2005) has used a comprehensive measure of domain specific memory self efficacy in accordance with assertions, based on stereotype threat theory, which have yielded mixed empirical results (e.g., Hess et al., 2003; Hess, Hinson, & Statham, 2004; Bargh, Chen, & Burrows, 199
37 value placed on the domain, and resulting anxiety, all contribute to the occurrence of threat and can impact performance (Steele, 1997). In addition to explicitly measuring anxi ety, the current study used both young and older adult participants, for whom memory and related stereotypes are posited to be differentially salient, as well as explicit measures of domain identification. Another strength of this research pertains to met hodology. At present consensus is lacking as to which type of stereotype activation is most effective and meaningful. Extant research has attempted to activate stereotypes both implicitly (e.g., Levy, 2003) and explicitly (e.g., Chasteen et al., 2002). Given the current lack of a methodological ed both approaches so as to enhance the activation process. Furthermore, a manipulation check w as utilized to assess whether stereotype effects were, in fact, actually per ceived by study participants. This is a step that has not been universally employed in the extant literature (e.g., Desrichard & Kopetz, 2005). Finally, other potentially relevant variables w ere assessed, so as to gain as comprehensive a picture as poss ible when accounting for factors that contribute to the stereotype performance relationship. Figure 2 1 provides a model of the hypothesized interrelations among the respective constructs when positive or negative stereotypes are activated. The model inc ludes the three la tent factor s -MSE, memory performance, and memory anxiety as well as two predictor variables, age and stereotype condition. Specific hypotheses follow. Hypothes is 1 Consistent with Levy (2003), it was hypothesized that older adult memor y performance would be lowered by negative stereotype activation and increased by
38 positive stereotype activation. Specifically, in comparison to older adults in a control group, memory recall performance would be lower among those exposed to negative ster eotype activation, and higher among those exposed to positive stereotype activation. Conversely, memory performance of younger adults would not be significantly altered by stereotype activation such that recall scores across all conditions -positive stere otype, negative stereotype, and control -would not be significantly different. This was predicted to be evidenced by a 2 way interaction between age group and stereotype condition, with memory performance as the dependent variable. Hypothesis 2a It was a lso expected that stereotype condition would have a similar impact on MSE. A 2 way interaction between age group and stereotype condition was hypothesized with MSE serving as the dependent variable. Compared to older adult participants in a control condi tion, those exposed to negative stereotypes were expected to display lower levels of MSE and those exposed to positive stereotypes were expected to show higher levels of MSE. This effect was not anticipated to occur for younger adults. Hypothesis 2b As an extension of the first two hypotheses, it was also expected that MSE would mediate the stereotype performance relationship with both negative and positive med iation, the univariate relationships between stereotype condition and memory performance, and MSE and performance were each expected to be significant. However, once variance attributable to MSE was statistically controlled, stereotype
39 condition effects w ere predicted to be attenuated, such that their relationship to older adult memory performance would no longer be statistically significant. The association between MSE and performance, however, was expected to remain significant, irrespective of stereoty pe condition (i.e., positive, negative, control). Hypothesis 3 As cited earlier, in spite of its posited theoretical importance, empirical support for the relevance of stereotype related anxiety and its impact on performance has been mixed (e.g., Steele, and other self regulatory factors, including anxiety, are theorized to be one of the four sources from which self efficacy beliefs are derived (Bandura, 1993). These empirical data and theor etical claim led to the following predictions. First, stereotype effects were hypothesized to have a significant impact on older adult anxiety (MIA Anx) score, as evidenced by a 2 way interaction between age and stereotype condition on anxiety. Specifica lly, among older adults, those exposed to negative stereotype effects were expected to display increased anxiety. It was unknown if positive stereotypes would levels. Hypot hesis 4 To comprehensively test the interrelationships proposed in the above hypotheses, a structural equation model of the aforementioned variables was tested. Memory performance was expected to be predicted by age and MSE. MSE, itself, was also expected to be predicted by age, with lower MSE levels expected among older adults. Additionally, although stereotypes have been shown to impact performance, as illustrated by the current model, this is believed to occur via their impact on MSE. MSE
40 was therefor e hypothesized to mediate the link between stereotype condition and memory performance; that is, the addition of condition assignment and/or anxiety into the model as a predictor of performance was not expected to improve model fit because MSE was expected to be the critical predictor of performance. Anxiety was expected to occur in response to different conditions but was not expected to have a significant impact on memory performance. Figure 2 1. Model of h ypothesized interrelations among variables when positive or negative stereotypes are activated
41 CHAPTER 3 METHODS AND PROCEDUR ES Overall Design Initially, prospective participants were mailed a demographic questionnaire and a measure of self reported memory anxiety for completion at home El igible participants from both age groups w ere then randomly assigned to one of three groups; negative stereotype, positive stereotype, or control. All participants were informed they would need to participate in a follow up tele phone survey consisting of remembering and making judgments about words (see appendix D). During the telephone survey, participants were administered a second wave of demographic questions including the two exploratory items (domain identification and subjective age), and then comp leted the first memory recall task. Participants then completed the memory self efficacy measure and d epending on group assignment, age based stereotypes (positive or negative) were activated prior to the final recall performance Memory related anxiety o f all participants w as measured again after the second recall task and a manipulation check was employed to test whether stereotypes were induced. Participant Information A total of 196 participants (68 male, 128 female) completed the written survey administered as the initial part of the study. This sample was comprised of both young and older adult participants. The young adults were University of Florida undergraduate students ( N = 101; 75 female, 26 male) ranging in age from 18 25 years old. The older adults were community dwellers ( N = 94; 52 female, 42 male), mostly from the state of Florida, ranging in age from 60 80 years old. Both age ranges are consistent with prior literature. Younger adult participation partially satisfied academic
42 requirements at the University of Florida and older adults were compensated with a $10 gift card. All participants were recruited for a study on attitudes and performance so as to disguise the memorial component A pproximately 17% of participants did n ot complete the telephone interview. Th e results reported here are based on the remaining participants for whom we have complete survey and phone interview data ( N = 1 61 ) Traditional dem ographic information was captured including gender, race, education level, chronological age, and subjective health (1 = excellen t health, 10 = very poor health). Older adults constituted approximately 58% of the sample. Data were included from four older adults falling outside of the designated a ge range, ages 55, 57, 59, and 81 respectively. Overall, participants w ere predominantly female (64%, N = 10 4 ) and Caucasian (71%, N = 114) Among young adults ( N = 67), approximately 51% ( N = 34) were Caucasian, 28% ( N = 19) Black, 9% ( N = 6) Latino, 6% ( N = 4) Asian, and 6% ( N = 4) categorized N = 9 3 ) were predominantly Caucasian (87%, N = 80) and Black (10%, N =10). Sample characteristics are reported in T able 3 1. Cognitive Function Any older adult participant whose ability to complete the telephone survey seemed compromised in any way was administered t he Telephone Interview of Cognitive Status (Brandt et al., 1988; TICS ). The TICS is an 11 item, standardized testing instrument (maximum score = 41 points) ori ginally designed to assess cognitive functioning of in situations where in person cognitive screening is impractical or inefficient. The test covers domains including memory, attention, and language (e.g., Count backwards from 20 to 1 ; What do you call the prickly green plant that lives in the desert? ) and correlates highly ( r = .94) with the Mini Mental State Examination
43 ( MMS E ; Folstein et al., 1975). The TICS has been validated in various languages in both its original and modif ied versions, and has shown high sensitivity (94%) specificity (100%), and test retest reliability ( r = .965), disease (Brandt et al., 1998; Dal Forno et al., 2006). Published c utoff score criteria (<25; Desmon d et al., 1994) were applied to the current sample. O ne participant was administered the test and obtained a score of 30. Adhering to these criteria, t he participant was deemed competent and subsequently complete d the remainder of the telephone survey. No participant was excluded due to cognitive impairment. Memory Anxiety To assess baseline memory related anxiety, all participants completed a modified version of the Memory Anxiety (MIA Anx) subscale of the Metamemory in Adulthood (MIA) questionnaire (Dixon & Hultsch, 1983, 1984). The MIA is comprised of seven subscales, all o f which have shown good internal consistency reliabilit y ( = .74 to .93; Hultsch et al., 1988). The current modified version of the MIA anxiety subscale consist ed of 14 memory related items (e.g., I get upset when I cannot remember something ) from the original instrument, interspersed with 14 distractor items (e.g ., I worry when I have to climb stairs ; large group ) so as to de emphasize the memorial component. All items were rated on a 1 5 Likert s cale ( 1 = A g ree strongly 5 = D isagree strongly ) and show ed good internal consistency ( = .76 ). The average score across items was the dependent measure, with higher scores indicative of higher anxiety levels (range = 1 5). Results were examined across all three conditions to ensure there were no significant differences among baseline anxiety levels. A second administration of the MIA anxiety
44 subscale was conducted immediately following the final wave of memory recall. This consisted of the 14 memory related items from the original version only, and was used as the dependent measure of memory anxiety following testing As in prior studies, the instrument showed good reliabilit y ( = .77) with the current sample Memory Self E fficacy Immediately after the second stereotype manipulation, participants were administered the Memory Self Efficacy Questionnaire 4 (MSEQ 4; West et al., 2003). The MSEQ 4 is a 20 item self report measure of memory self efficacy for distinct memory tasks. These tasks include remembering names (e.g., If someone showed me the photographs of X people and told me their names once, I could identify Y persons by name if I saw the pictures again a few minutes later ), remembering items from a grocery list (e.g., of Y items, withou t taking any list with me to the store ), remembering main points from a story, and remembering the locations of household items. Individuals indicate their confidence level for performing each of these tasks at varying difficulty levels, responding on a scale from 0 ( I cannot do it ) to 10 ( 100% sure I can do it ). The dependent measure wa s self efficacy strength, calculated as the average confidence rating across all items, ranging from 0 100 in each case. The MSEQ 4 is a valid and reliable measure of self efficacy strength (Berry et al., 1989) and show ed excellent internal consistency reliability ( = .91 ) in this sample Stereotype Effects The stereotype literature indicates that two methodological approaches have been used for stereotype activatio n implicit and explicit. Most implicit activation (e.g., Levy
45 2003; Banaji, Hardin, & Rothman, 1993) has involved some form of exposure to positively (i.e., wisdom) or negatively (i.e., decrepit) valenced stereotype primes on a computer monitor at sub thr eshold speeds. This approach is posited to activate stereotypes quickly enough to occur below conscious awareness. Other researchers (e.g., Hess & Hinson, 2006; Desrichard & Kopetz, 2005) have also utilized more explicit approaches, such as manipulating task instructions to either emphasize or de emphasize age differences in performance and/or memorial components. Some feel this distinction between explicit and implicit stereotype activation is important, given that d unconscious attitudes in the same functional area may yield inconsistent results (e.g., Devine, 1989; Levy, 1996). Given the lack of evidence favoring one methodological approach over another, the current study employed both types of activation to ensur e maximal effect. Implicit Activation Initially, participants were asked to complete a short (20 word) recall task that required the m to memorize words that were read to them on the phone. Consistent with earlier work, the memorial component, by itself, was hypothesized to activate older addition, using an implicit priming format and depending on group assignment, stereotyped aging (positive or negative) or non aging rela ted (control) primes were embedded in these lists. This was an implicit activation as the semantic properties of the words were not alluded to in any way. Participants heard the words, then were given two minutes to recall as many as they could from me mory, in any order they could remember them. Words were categorized based on stereotypicality (e.g., age related or neutral) and valence (positive
46 or negative), and were adapted from prior studies related to aging, stereotypes, and memory (e.g., Hess et a l., 2004; Hess et al., 2003; Bargh et al., 1996; Howard, 1979). Items in the positive aging stereotype condition included dignified, insightful, kind, respected, wise (see appendix A). Items in the negative aging stereotype condition included brittle, co mplaining, confused, forgetful, senile (see appendix B). Items in the neutral condition included large, flat, green, costly, metallic (see appendix C). To reiterate, the primary purpose of this initial recall task was to implicitly activate age related stereotypes. Considering that prior studies have found mere engagement in recall tasks to be sufficient to activate such beliefs, particularly among older p articipants (e.g., Desrichard & Kopetz, 2005), inclusion of stereotypical terms on these lists was theorized to only further increase the likelihood that such activation would occur. Explicit Activation P articipants were also subjected to an explicit stereotype activation patterned after Hess and Emery (2008). Each person was assigned to the same condition that was assigned for the first activation task (positive, negative, control) Memory stereotypes were activated via a task instructions manipula tion, prior to the second memory recall task. Participants in the positive and negative stereotype conditions heard one of the following two explanatory statements below, while for those in the neutral/control condition, the phone survey continued uninter Positive Condition Shortly I am going to examine your ability to process verbal information. In an effort to reduce potential biases, we will be using a task that has been shown to be appropriate for individuals of all ages. Interestingly, older adults have been shown to s Negative Condition Shortly I am going to examine your memory ability using a test that has been used extensively by researchers to study aging effects on m emory. Younger adults typically do much bette r than older adults on this test
47 The MSEQ 4 was administered right after this manipulation. In l ight of the time that elapsed during MSEQ 4 administration, upon its completion, and immediately prior to the s econd wave of memory recall, participants in the positive and negative stereotype conditions were given an abridged, one instructions task on whic studies (e.g., Chasteen et al, 2005) and should have compensated for any time lag between stereotype activation and performance task. Neutral/control condition participants did not receive any reminder. Overall, this sequence of events allowed for maximum impact of stereotype effects on both MSE and performance, as a function of condition. Memory Performance Patterned after We st, Welch, & Thorn (2001), two versions of a 20 item categorized shopping list were constructed from a pool of over 1,000 items (see West et al., 2003). To maximize uniformity, each list was recorded on computer by an advanced doctoral student, reading the i t ems at 1 second intervals. Participants were randomly assigned to hear one of these two lists. After the word lists were playe d by the experimenter, p articipants were then given up to two minutes to recall as many items as they could from memory, i n any order. The number of items recalled correctly was used as the dependent measure. Because of potential variance due to differential sound quality of participant telephone connections, two separate scores were calculated; lenient and strict. The pri mary difference in scoring pertained to those words mentioned by participants that were phonologically similar to those on the recorded list (e.g., lips instead of clips). Lenient scoring included these as correct while strict scoring did not.
48 Recalled w ords identical to those on the lists, as well as singular and plural derivatives (e.g., beans for bean ) were accepted as correct by both scoring criteria. Exploratory Measures In addition to the hypotheses listed earlier, the roles of several other varia bles of interest were investigated in an exploratory fashion. These included domain identification and age (subjective and chronological). As alluded to earlier, stereotype threat theory has posited domain identification to be theoretically relevant (Ste ele & Aronson, 1995; Steele, 1997). Specifically, the theory argues that the more people identify with a particular domain, the more specific stereotypes. Empirical confirmation o f this concept, however, has been elusive (e.g., Hess et al., 2007 in press ). Given the increased importance older adults seem to ascribe to the memory domain (e.g., Dark Freudeman, West, & Viverito, 2006 ), this theoretical tenet seems worthy of further e xploration, especially in light of the mixed empirical results. To explore domain identification empirically, participants were asked to rate the domain of memory in terms of personal importance (e.g., On the following scale, please indicate how importan t memory performance is to you ). This item was part of the preliminary questionnaire completed at home prior to the phone survey, and responses were given on a 10 point Likert scale ( 1 = N ot at all important 10 = V ery important ) Related to the issue of domain identification is age identification. That is, do pe
49 as younger. Since self efficacy and stereotypes are first and foremost beliefs, it is s urprising that this distinction is rarely addressed in the respective literature, and adults with self stereotyping theory (e.g., Levy, 2003), has found that when partic ipants are How does this pertain to the current research? According to the various stereotype theories (e.g., L evy, 2003; Hess, 1997), people belonging to the stereotyped group will be susceptible to related stereotype effects. But if a 46 year old man effects ? Conversely, if a woman is 70 years old but considers herself much younger, is she any less susceptible to the impact of aging stereotypes? Further, does subjective age generally coincide with or deviate from chronological age? The current study addresse d this disti nction between chronological and subjective age, by examining and comparing them. Specifically, in addition to asking participants to indicate their chronological age, an additional questionnaire item ( When you think about how well you are doing overall how old do you feel? ) 10 point Likert scale ( 1 = young ; 10 = very old ). To ensure reliability, these exploratory subjective age and domain identification items were administered twice in the survey items completed at home before the telephone interview and a second time during the telephone interview. The dependent measure for each item was the average of responses at times 1 and 2, with higher
50 scores indicating older subjective age and greater memo ry ide ntification, respectively For both varia b l es, the responses at times 1 and 2 were significantly correlated with each other ( Domain Identification r = .45, p <.0 1 ; Subjective Age r = .48, p < .01), however test retest analyses revealed only moderate reliability as evidenced by a Guttman Split Half coefficient of .614 for Domain Identification and .652 for Subjective Age (see data for these measures in Table 3 2) This was likely influenced by the fact that on each occasion, only one item was used to measure the respective constructs. It should also be not ed that d ue to experimenter error, the se items were not included in all of the follow up phone interviews. M ost of the interviews that did include these measures were conducted with older adults T herefore the se data were captured disproportionately at time 1 ( N = 159) and of those participants completing both pre and post measures ( N = 87), most were older adults ( N = 79). Manipulation Check To confirm that stereotypes were indeed activated, a manipulation check was conducted after the second recall test. Four items measured agreement (1 = strongly agree ; 5 = strongly disagree ) with statements consistent with the task instructions. Two items coincided with the negative stereotype condition (e.g., It is likely that my age had an impact on my performance ; On the test you took earlier, people generally tend to perform differently depending on their age ) and two items coincided with the positive stereotype condition (e. g., On the memory test today, participants of all ages are supposed to perform at about the same level ; Your age did not play a strong role in your performance on the test today ). These items were summed with a possible cumulative score range of 4 20. As needed, items were reversed so that a high score indicated a strong belief that age was a factor leading to poor memory p erformance.
51 Table 3 1. Sample Characteristics Variable N M SD Chronological Age Younger Adults 67 19.25 1.13 Older Adults 92 67.10 5.23 Total 159 46.94 24.04 Years of Education Younger Adults 67 13.13 1.21 Older Adults 91 15.80 3.20 Total 158 14.70 2.90 Subjective Health Younger Adults 67 3.40 1.84 Older Adults 92 3.33 1.80 Total 159 3.40 1.80 Table 3 2 Exploratory Items Variable N M SD Domain ID Time I 87 9.10 1.20 Time II 87 8.93 1.43 Average 87 8.99 1.10 Subjective Age Time I 87 6.14 1.95 Time II 87 6.34 1.85 Average 87 6.24 1.64
52 CHAPTER 4 RESULTS Overview Preliminary analyses were condu cted to ensure that counterbalanced assignment to conditions resulted in an equitable distribution of key sample characteristics across the three experimental conditions. Preliminary analyses were also utilized to examine the manipulation check results. Primary h ypotheses were then tested with univariate analyses of variance (ANOVA) and structural equation analyses Due to potential variability in phone connections and su bsequent sound quality of the conversations and recorded telephone interviews, lenient and strict recall scores were assessed for each memory recall task Final scoring criteria for each list were developed and confirmed by the principal investigator to e nsure uniformity S eparate analyses were run for both the lenient and strict recall scores. These results were essentially the same as evidenced by the near perfect correlation (r = .98, p <.001) between lenient and strict recall scores; only strict reca ll data are reported in detail here. Preliminary Analyses Analyses of potential baseline condition differences were conducted using an a ge (younger, older) x c ondition (control, positive, negative) design, on the following variables: subjective age, year s of education, domain identification and anxiety. Al though participants were assigned to the three stereotype conditions via unbiased counterbalancing (control N =55, positive N =54, negative N =51), baseline univariate analys es of variance (ANOVA) reveale d significant condition differences in subjective age [ F (2, 15 4) = 3.9 0 p = .022 p 2 =.05 ], with those in the negative stere otype condition feeling oldest, those in the positive stereo type condition feeling youngest, and
53 control participants falling in bet ween (see T able 4 1 ) The C ondition x A ge interaction was not significant [ F (2, 154) =.50, p =. 61, p 2 =. 01]. Not surprisingly, significant baseline differences between younger and older adult participants were also found in education [ F (1, 153) = 41.32, p <.001, p 2 =.21] as shown in Table 3 1, and domain identification [ F (1, 154) = 8.50, p = .004, p 2 =.05] As evidenced in T able 4 2 across conditions, older adults rated memory as a more personally relevant domain than their younger counterparts. C ondition x A ge interactions were non significant for both education [ F (2, 153) = .13, p = .88, p 2 =.002] and domain identification [ F (2,154) =.30, p = .75, p 2 =.004] Perhaps the most important baseline difference was found in anxiety lev el, given its pertinence to several of the hypotheses. Across age, baseline anxiety differences were found as a function of stereotype condition [ F (2, 154) = 4.34, p = .02, p 2 =.05] and Bonferroni adjusted pairwise comparisons determined that the baseline anxiety difference between positive and negative conditions was significant ( p < .01) with higher baseline anxiety in the positive condition than the negative condition (see T able 4 3 below ). The C ondition x A ge interaction was not significant [ F (2, 154) = .23, p = .80, p 2 =.003]. In light of the observed condition differences in anxiety, hypotheses were tested in two waves of analyses. After all analyses were run initially, they were then rerun with baseline anxiety as a covariate. The manipulation check items we re examined in a 3 ( c ondition) x 2 ( a ge) analysis of variance. Results showed no significant differences based on condition [ F (2,150) = .45, p = .64, p 2 =.01] and approached significance for age [ F (1,150) = 2.87, p =.09, p 2 =.02], with older adults displaying stronger beliefs in a memory performance aging link
54 ( N = 92, M = 14.62, SD = 2.64) as compared to their younger counterparts ( N = 63, M = 13.84, SD = 2.78) The C ondition x A ge interaction was also non significant [ F (2, 150) = .30, p = .76, p 2 =.004]. These age differences were further explored in relation to two of the key outcome variables, self efficacy and memory performance, to ensure that aging effects on those measures were not due to the manipulation check resu lts. Variance due to the manipulation check was covaried out and a significant main effect for age emerged on MSE [ F (1,148) = 12.70, p =.001, p 2 = .08] and memory recall performance [ F (1,149) = 13.54, p <.001, p 2 =.08] This suggests that even when basic beliefs about memory aging are held constant, significant age differences still remain for self efficacy and memory performance Hypothesis 1 hypothesis predicted t hat stereotype condition would have an impact on older adult memory performance (either increase or decrease), and that the direction of this impact would vary by stereotype valence (positive, negative). It was also predicted that this would not occur for the younger adult s for whom memory is not particularly salient. Evidence for this would be a 2 way interaction between age group and stereotype condition, with memory performance as the dependent variable. To test this, univariate analyses of variance ( ANOVA) were conducted with m emory performance as the dependent variable and age group and stereotype condition as between subjects factors Original analyses showed best performance by participants in the control condition, followed by the negative and p ositive conditions respectivel y The main effect
55 of age was significant for recall performance [ F (1, 1 50 ) = 13.20, p < .001 p 2 =.08 ] Across conditions, o lder adult memory performance was significantly worse than th at of younger adults. However, c hypothesis, no significant main effect emerged for stereotype condition on memory recall [ F (2, 150) = .001, p = .99, p 2 =.000 ] The C ondition x A ge group interaction was also non significant [ F (2, 150) = 1.80, p =.17, p 2 =.02]. M emory performance data are reported in T able 5 5 Baseline anxiety was incorporated into the analysis as a covariate, due to baseline condition differences in anxiety. S tereotype condition remained non significant [ F (2, 149) = .05, p = .95, p 2 =.001] T he main effect of baseline anxiety was not significant The C on dition x A ge interaction was not signific ant [ F (2,149) = 1.74, p =.18, p 2 =.02 ]. Taken together, these findings show that differences in memory performance were affected by age but not ste reotype condition. Hypothesis 2a Hypothesis 2 a predicted stereotype condition would exert a similar impact on older adult MSE lowering it in the negative stereotype condition, boosting it in the positive stereotype condition. Statistically this would be supported by a 2 way interaction between age group and stereotype condition with MSE serving as the dependent variable. This hypothesis was tested in a similar fashion to hypothesis 1 A NOVA revealed a significant main effec t of age [ F (1, 150 ) = 13.50, p <.001, p 2 =.08] but not condition [ F (2, 150) = 1.08, p = .34, p 2 =.01]. The C ondition x A ge interaction was not significant [ F (2, 150) = .40, p = .67, p 2 =.01]. See T able 5 6 Similar to hypothesis 1 baseline anxiety was covaried out to determine what impact this might have in accounting for the memory self efficacy differences. The main
56 effect of age on MSE remained significant [ F (1, 149 ) = 15.15 p <.001, p 2 =.09], and m emory self efficacy scores increased slightly across conditions the biggest boost occurring for participants in the positive stereotype condition Stereotype condition remained non significant [ F (2, 149) = .40, p = .67, p 2 =.01] and in terms of MSE, t he order remained the same wi th control group participants showin g the highest level followed by negative and positive conditions respectively. T he stereotype C ondition x A ge interaction also remained non significant [ F (2, 149) = .36, p =.70, p 2 =.01]. Hypothesis 2b For this hypo thesis, the author predicted that self efficacy would mediate the impact of stereotypes, both positive and negative, on performance. Tests of mediation require significant correlations between the key dependent variable s in this case memory performance and MSE, and the predictor (stereotypes) as a basic assumption of a mediation analysis. The univariate relationships between stereotype condition and memory performance, and MSE and memory performance were each expected to be significant A nalyses revea led tha t only the latter was significant. Therefore, mediation could not be tested. Hypothesis 3 3 examined the impact of stereotype effects on older adult anxiety Compared to baseline levels, increased anxiety was expected among older adults exposed to negative stereotype s while i t was unknown if positive stereotypes would lead to a corresponding anxiety decrease Statistically, this would be evidenced by a 2 way interaction between a ge and stereotype condition on MIA anxiety subscale score s
57 To test this hypothesis a repeated measures ANOVA was conducted with age and stereotype condition as predictors, with the two anxiety assessments as a repeated factor Participant a nxiety leve ls were highest in the positive stereotype condition followed by control and negative stereotype conditions respectively (see T able 4 7), and the main effect of condition on anxiety approached significance [ F (2, 149) = 2.87, p = .06, p 2 =.04] due to the fact that anxiety overall was higher in the positive condition than negative condition ( p <.05) and somewhat higher in the positive condition than control con dition ( p = .09). This result was likely due to baseline differences in anxiety, as explained abo ve Follow up anxiety measured during the phone interview was significantly higher than baseline anxiety [ F (1,149) = 52.3, p < 0001, p 2 = .26], and time of assessment significantly interacted with age [ F (1,149) = 4.28, p < .05, p 2 = .03]. This was due to a larger change from baseline to follow up assessment for older adults than younger adults, although both groups showed a significant increase in anxiety ( p < .001). This change in anxiety across time did not interact with condition [ F (2,149) = 1.0, p =.37, p 2 =.013], nor was there a significant three way interaction between time, age, and condition [ F (2,149) = .12, p = .89 p 2 =.002] T aken together, these findings suggest that memory testing, itself, created greater anxiety in older adults than younger adults regardless of condition. Hypothesis 4 (Structural Equation Model) To examine the predicted relationships among the variables of focus in this paper, a structural equation model was analyzed using AMOS 16.0 Age was used as a dich otomous variable to reflect the categorical nature of the age groups in the current
58 sample According to Hu & Bentler (1999), the following criteria represent adequate fit: non significant Model X CFI .95, and RMSEA .06. A m easurement model was developed and included three latent factor s (see F igure 4 1) Four memory subdomain measures object location, story recall, name recall, and shopping list recall were used to construct a latent factor for memory self efficacy. Ea ch of these indicators significantly contributed to the latent factor (loadings were .67, .48, .61, and .80, respectively, all p <.0001). Baseline and follow up measures of memory anxiety were used to construct a latent factor for anxiety. Both indicator s significantly contributed to the latent factor (loadings for baseline and follow up anxiety were .78 and .92 respectively, all p < .0001). A final latent factor of memory performance was constructed with two memory recall tests These two factors signi ficantly contributed to the latent factor (loadings for the first and second test were .74 and .81 respectively, both p <.0001). Based on the conventions for model fit noted above, t he predicted model (see Figure 6 2) showed poor fit to the data, X (29)=55.51, p =.002, CFI=.900, RMSEA=.095, suggesting that the overall pattern of relationships represented in this model was not supported in this sample. The model did show significant individual paths from condition to anxiety ( =.22, p < .0 5 ), anxiety to MSE ( = .39, p < .0001), age to MSE ( = .26, p < .02 ), age to performance ( = .36, p < .00 5 ), and MSE to performance ( = .28, p < .05 ). Paths from age to anxiety and condition to MSE were both non significant ( p s = .26 and .58 respectively). These two paths were trimmed yet th is did not improve model fit overall X (31) = 56.78, p < .00 5 CFI=.903, RMSEA=.090 Alternative models were thus developed and will be discussed later (see exploratory analyses).
59 Exploratory Analyses Subjective Age and Domai n Identification Neither of the two exploratory items yielded significant results when analyzed as predictors of the primary outcome variables ( anxiety, age, self efficacy, performance) Regarding domain identification, there are different ways to measure this construct, and no agreed upon method in the literature. T at baseline and then repeat that item again during the phone interview seemed acceptable as the two were significan tly correlated with each other ( r = .45, p < .001). When the two items were averaged, domain identification was not significantly correlated with any of the relevant variables. Subjective age, however, was significantly correlated with health and both b aseline and follow up measures of memory related anxiety. Correlations with all other relevant variables, including chronological age, were not significant (See Table 5 8 for all correlations ) Structural Equation Model s As mentioned above the analyses of the predicted model revealed poor fit to the data. A series of alternative models were th en constructed in an exploratory fashion using the full sample to account for the predicted interrelationships among the key variables highlighted in this paper starting with the measurement model described above The model with the best fit for the full sample still did not have excellent fit to the data due to a significant chi square and low CFI, although the RMSEA value was acceptable X (96 )= 129.6 p =. 013 CFI =. 922 ., RMSEA = .047 Interestingly, this model fit was achieved with gender included as a predictor (see Figure 4 3 ) Though not a
60 theoretically and empirically established w ithin both Crandall, 2003) and memory self efficacy (West, Welch, & Knabb, 2002 ) literature s, respectively. Path estimates for old er and young er part icipants are provided in Table 4 9 As evidenced by the se data, relationships among variables were stronger for older adults. Given the relatively poor fit of the aforementioned models combining participants of all ages, and considering that all hypotheses in the current study predicted differential salience fo r older adults, these relationships were examined in a comparable model with older adults only. Using older participants only, a model (see F igure 4 4 ) wit h age (continuous variable) education, health, anxiety, and memory self efficac y as predictors s howed good fit to the data, X (38) = 50.085, p =.091, CFI=.946, RMSEA=.059 Significant predictive paths emerged from memory anxiety to MSE ( = .45, p <.01), MSE to memory performance ( = .30, p =.02), age to memory performance ( = .15, p > .10), education to memory performance ( = .33, p = .003), and health to memory anxiety ( = .32, p = .005). A Sobel test was conducted and also revealed a significant indirect effect for anxiety on performance via MSE (Sobel test statistic = 1.98, p <.05). Trimming the age variable (no significant prediction) slightly degraded model fit overall, X (31) = 45. 313 p =.047, C FI=.937, RMSEA=.071, though individual loadings were comparable and remained significant; health to anxiety ( = .32, p < .005), anxiety to MSE ( = .45, p < .0001), education to performance ( = .34, p < .00 5 ), and MSE to performance ( = .31, p < .0 2 ).
61 One way to sta ti sti cally identify is by conductin g a nested chi square test of two separate models one nest ed in the other This comparison statistically tests the null hypothesis of no significant difference in fit by evaluating whether the chi square difference between models is significant, for the given degrees of freedom and a chosen significance level. (Figures 4 3 & 4 4) were identical except for the exogenous gender variable in the combined age model (Figure 4 3) hereby satisfying the criterion of being nested one within the other. The test did reveal a sig nificant difference, X ( 58 ) = 79.51, p <.05 offering statistical (Figure 4 4) superiority. This model will be considered in the discussion in greater detail.
62 Table 4 1. Baseline Subjective Age as a Function of Condition and Age Group N M SD Control Younger 23 6.4 1.6 Older 32 5.9 1.8 Total 55 6.1 1.8 Positive Younger 22 5.6 1.7 Older 32 5.8 1.8 Total 54 5.7 1.7 Negative Younger 22 6.7 1.9 Older 29 6.7 2.2 Total 51 6.7 2.0 Total Younger 67 6.2 1.8 Older 93 6.1 1.9 Total 160 6.2 1.9 Table 4 2 Baseline Memory Domain Identification as a function of Condition and Age Group N M SD Control Younger 23 8.2 1.6 Older 32 8.9 1.9 Total 5 5 8.6 1.8 Positive Younger 22 8.2 1.6 Older 32 9.0 1.4 Total 54 8.7 1.5 Negative Younger 22 8.6 1.1 Older 29 9.1 0.9 Total 51 8.9 1.0 Total Younger 67 8.3 1.5 Older 93 9.0 1.4 Total 160 8.7 1.5
63 T able 4 3 Baseline Anxiety as a Function of Condition and Age Group N M SD Control Younger 23 2.7 .6 1 Older 32 2.6 .6 8 Total 55 2.7 .65 Positive Younger 22 2.9 .50 Older 32 3.0 .83 Total 54 2.9 .70 Negative Younger 22 2.6 .50 Older 29 2.5 .70 Total 51 2.5 .60 Total Younger 67 2 7 .54 Older 9 3 2.7 .80 T otal 160 2.7 .67
64 Table 4 4 Memory Performance as a Function of Condition and Age Group N M SD Control Younger 23 9. 9 3.6 Older 31 9.3 2.1 Total 54 9.6 2.8 Positive Younger 22 10.9 3.2 Older 32 8.2 3.5 Total 54 9.3 3.6 Negative Younger 19 10.7 3 Older 29 8.5 3.4 Total 48 9.4 3.2 Table 4 5 Self efficacy as a Function of Condition and Age Group N M SD Control Younger 23 77.1 9.7 Older 31 72.1 12.9 Total 54 74.2 11.8 Positive Younger 22 76.0 9.5 Older 31 66.9 15.1 Total 53 70.7 13.7 Negative Younger 20 78.3 11.2 Older 29 70.5 12.5 Total 49 73.7 12.5
65 Table 4 6 Anxiety as a Function of Condition and Age Baseline trial Follow up Phone Interview Group N M SD N M SD Control Younger 23 2.7 0.6 23 3 0 0.5 Older 31 2.6 0.7 31 3.0 0.6 Total 54 2.7 0.6 54 3 0 0.5 Positive Younger 22 2. 9 0.5 22 3 0 0.5 Older 3 2 3.0 0.8 32 3.3 0.7 Total 5 4 2.9 0.7 54 3.2 0.6 Negative Younger 18 2.7 0.4 18 2.9 0.6 Older 29 2.5 0.7 29 2.9 0.6 Total 47 2.6 0.6 47 2.9 0.6 Table 4 7 Manipulation Check as a Function of Condition and Age Group N M SD Control Younger 23 13.7 2.7 Older 31 14.8 2.6 Total 54 14.3 2.6 Positive Younger 22 13.5 3.0 Older 32 14.4 2.9 Total 54 14 1 2.9 Negative Younger 19 14.4 2.7 Older 29 14.7 2.6 Total 48 14.5 2.6 Total Younger 64 13.8 2.8 Older 92 14.6 2.6 Total 156 14.3 2.7
66 Table 4 8 Correlations Between Predictor and Memory Performance Variables V ariable 1 2 3 4 5 6 7 8 9 1. Memory Performance (Strict Recall) ** .98** .28** .12** .19* .28** .07 .08 .15 2. Memory Performance (Lenient Recall) ** .27** .13** .18* .25** .09 .08 .14 3. Memory Self Efficacy ** .23* .36** .28** .17 .08 .16 4. Memory Anxiety (Baseline) ** .68** .01 .43** 0.16 .27** 5. Memory Anxiety (Follow up) ** .10 .24* .06 .23 ** 6. Chronological Age ** .04 .11 .02 7. Subjective Age ** .03 .36** 8. Domain Identification ** .05 9. Health ** Note. ** p <.001, p <.05
67 Table 4 9 Path Coefficients as a Function of Age Group Path P Young er Adults H ealth to Anxiety .23 NS Gender to Anxiety .32 .033 Anxiety to MSE .11 NS Gender to MSE .32 .012 Age to Performance .09 NS MSE to Performance .26 NS Education to Performance .16 NS Older Adults Health to Anxiety .32 .005 Gender to Anxiety .16 NS Anxiety to MSE .53 <.0001 Gender to MSE .42 <.0001 Age to Performance .15 NS MSE to Performance .33 .010 Education to Performance .34 .003 NS = p >. 05
68 Figure 4 1. Measurement Model Figure 4 2. Predicted Model with Young er and Older Adults
69 Figure 4 3. Best Model with Young er and Older Adults Note: path coefficients are listed in Table 4 9. Fi gure 4 4. Final Model with Older Adults Only
70 Figure 4 5. Age Differences in Memory Anxiety Change After Baseline 1 1.5 2 2.5 3 3.5 4 4.5 5 Baseline Follow-up Memory related Anxiety Time of Assessment Younger Older
71 CHAPTER 5 DISCUSSION The current study addressed several interrelationships among memory self efficacy, memory performance, anxiety, age, and aging stereotypes. knowledge, this is the first empirical attempt to include memory self efficacy, as theorized by Bandura, when evaluating the link between aging stereotypes and memory performance across the lifespan Results for each of these variables will be considered in detail. Age It is clear that chronological age is a recurring, central theme in memory resea rch and a geist stereotypes, many of which are memorial in nature, are ubiquitous in many contemporary societies. Recently these stereotypes have become important in theoretical and empirical arenas (e.g., Steele, 1997; Levy & Leifheit Limson, 2009) inc luding memory (Hess et al., 2004) though experimental a ttempts to manipulate aging stereotypes have yielded mixed results in terms of effects on older adult memory performance (e.g., Hess & Hinson, 2006). stereotypi ng theory, the domain of memory becomes an increasingly sensitive area for people over time. Theoretically, as people age, they increasingly monitor their memory ability and interpret memory lapses as mounting evidence of age related memory decrement. St eele (1997) has also theorized that societal stereotypes can contribute to or reinforce such beliefs or schemas, inevitably hindering maximum performance in a functional domain, in this case memory. Although positive aging stereotypes do exist (e.g., wise kind), one need only glance at popular media or greeting cards to appreciate the great extent to which older adult
72 memory and cognitive function serve as negative stereotype targets. Elders are often mocked as senile, forgetful, or even demented. Band ura (1997) has highlighted the persuasive influence such messages can specific confidence (i.e., self efficacy) and ultimately, performance Self efficacy theory identifies four primary sources from which efficacy is derived, one of which Bandura terms verbal persuasion views, in general, as well as the beliefs of significant friends and family Often but not exclusively verbal in nature, negative messages about aging are often internalized and can sha pe how people view themselves (efficacious or inefficacious) in regard to specific areas of functioning Another theorized self performing in a domain, and utiliz e such information as a cue that can shape their perceived efficacy level. Reviews in this paper have highlighted overlap between theories of stereotype effects and self efficacy. Guided by the se theories, empirical work within both the memory self efficacy (e.g., West et al., 2003 ; West et al., 2005; West et al., 2007 ) and t he aging stereotype ( e.g., Hess et al. 200 3 ) literature ha s tested and established this belief performance link, and has also shown consistent age differences in performance favoring you th Not surprisingly, p reliminary ANOVA r esults from the current stu dy were no exception, as older adults were outperformed by their younger counterparts on the memory recall tasks possessed comparatively lower levels of MSE, and more frequently perceived age to have an impact on memory (manipulation check variable). Th e s e findings were consistent across stereotype conditions and memory tasks,
73 bolstering the notion that the domain of memory holds more sway as people age. Why is this the case? In regard to the above theories, it seems such beliefs may be partially rooted in actual performance differences. The trend of cognitive decline with increased age has been well established empirically. Age differences in memory performance occur on almo st all episodic memory tasks (e.g., Spaniol et al., 2006; Head et al., 2008) and on many o ther kinds of cognitive tasks (e.g., Colcombe & Kramer, 2003; Salthouse, 2009). Therefore, it is likely that pre existing beliefs in stereotypes, fadin g confidence, and actual performance differences all influence each other over time. T he current study utilized structural equation modeling, a powerful statistical technique for explor ing such complex, multivariate interrelationships. The study analyzed a single moment in time to understand the interrelationships. Informed by extant literature, an initial model (see Figure 4 2) hypothesized age would directly predict performance and exert an indirect effect on performance via MSE and anxiety. Stereotyp e condition, also a predictor in the model, will be discussed below (see Stereotype Condition). The predicted model (Figure 4 2) revealed results similar to those capture d by analyses of variance, namely, age was a significant predictor of MSE and me mory performance. Unexpectedly, however, these effects were attenuat ed in subsequent a lternative model s It appears that the predictive effect of age on performance in rather than chronological age, itself. In other words, age may have been a proxy for education, memory anxiety, and MSE, and it was those variables that showed the
74 strongest link to variance in performance when multiple factors were taken into account in subsequent models As shown throughout this paper, the case for the relevance of age seems a strong one. At the same time, there are ma ny different ways to assess age, and selection of specific age groups can affect study outcomes. First, c hronological age is commonly assumed to be an important and meaningful group ing variable Based on this default category, e xperimental aging research tends to utilize two kinds of samples ; combined young/old and older adult only Combined samples generally younger adults, mainly undergraduate college students, to adults, which often includes participant s ranging in age fro m 50 to 90 years old. Because people in their early 50s are likely quite different from those in their late 80s especially in terms of physical health and cognitive function (including memory), both commonly researched outcomes, some studies account for young old and oldest old This design is especially common when examining exclusively older adult samples. Though s till in its nascence, th distinction has been investigated empirically, in cluding research focusing on aging stereotypes and m emory (e.g., Hess et al., 2007). Th e current study did not include such a distinction, and analyses of the final model with older adults only (see Figur e 4 4) revealed a relatively unimportant role for age vis vis memory performance. Chronological age predicted l ess than 3% of recall score variance and this relationship was non older adults performed comparabl y whether at the younger (60 70) or older (>70) end of the late life spectrum. Although t he current results suggest little variation within the
75 the heterogeneity within this demographic remains understudied and theoretically important A ny arbitrary designation or may be at once common practice (socially and methodologically) yet not always empirically meaning ful. Inclusion criteria remai n s an unresolved empirical question. One alternative is to consider subjective age, rather than chronological age. Although this approach was used here, it was not as useful as expected (see exploratory analyses). Cumulatively, several pieces of evidence emphasize the importance of chronological age with the current sample: t he non significant correlations of subjective age mentioned a bove, ANOVA results showing age differences in MSE and memory performance, and the fact that the predictive paths were more frequently significant and loadings were stronger when modeling data from an exclusively older adult population However, it should be reiterated that the significant age effect revealed by ANOVA w as attenuated in exploratory structural equation models (see Figures 4 3 & 4 4) Th e fact that age differences constitute a consistent and robust outcome ac ross cognitive aging literature makes this particularly surprising. Again, i t could be the case that chronological age wa tion, self efficacy, anxiety, and overall health differences in other research, whereas with the multivariate models employed here, the re is a stronger respective influence More research should be devoted to the goal of unde rstanding factors that covary with age which may, in fact, be better predictors of performance than chronological age Stereotype Condition Stereotypes are a pervasive reality in many, if not all, societies. Psychologists have established a theoretical and empirical link between such beliefs and actual
76 behavior across diverse groups and performance domains, including older adults and memory, the primary focus here. Patterned after earlier work (Hess et al., 2004), the current study attempted to activate positive and negative aging stereotypes both implicitly and explicitly. Because members of all ages are exposed to aging stereotypes, and gi ven the theorize d differential salience for stereotyped group members (i.e., older adults) both younger and old er adults were assigned to various stereotype conditions. Based on the extant literature, a cross hypotheses aging stereotypes were expec ted to exert an impact on several outcome variables, namely, memory performance, memory self efficacy, and memory related anxiety for older adults only Contrary to hypotheses, d irect empirical tests of this interaction of age and condition proved non signific ant Because of the proposed multivariate nature of the stereotype performance link, a structural equation model was constructed to account for these multiple pathways. The predicted model identified age and stereotype condition as the primary predictors of performance. Given the proposed importance and theoretical mechanism of action of MSE, stereotype condition was expected to predict performance indirectly only, via MSE and memory related anxiety (a theoretical component of MSE). S tructural equation a nalyses of the predicted model (see Figure 4 2) showed poor fit to the data. The effect of condition on anxiety was significant but the effect on MSE was not. These mixed results are difficult to interpret. Reasoned from the perspective of self efficacy theory, s tereotype messages signify a type of verbal or social persuasion, one of four primary sources from which self effi c acy is derived according to
77 Bandura (1997) The null finding for MSE suggests the stereotype messages may not have be en persuasive enough to create a significant effect. However, another self efficacy source is theorized to be physio logical arousal such as anxiety or other affective states more broadly. In accordance with the theory, c ondition was found to exert a sign ifican t impact on memory related anxiety. Taken together, i t appears that condition may have been persuasive enough to tap th is arousal source, thus increasing anxiety, but the magnitude of this impact may have fallen short of that necessary to alter over all MSE level. Again, self efficacy is thought to stem from four unique sources. That said, i t is diffic ult to surmise precisely why condition did not significantly affect MSE al though several additional explanations seem plausible. First, al t hough par ticipants were assigned to condition by unbiased counterbalancing closer inspection revealed the majority of older adu lt males were placed in the positive stereotype condition ( N = 17) followed by negative ( N = 14) and control ( N = 10) conditions respectively M ost female older adults were placed in the control ( N = 22) condition with an equal number ( N = 15) in both negative and positive stereotype conditions respectively It is unclear wh ether participant gender has relevance for susceptibili ty to stereotypes though it seems feasible given the long history of women serving as target s of innumerable gender based stereotypes and collectively representing an oppressed social underclass Does such experience make them more sensitive or resilien t to stereotype effects? any gender stereotype relationship has not been empirically investigated with older adult memory performance. However, some stereotype literature ( e.g., Spencer et al., 1999 ) has demonstrated that women are vulnerable to stereotypes when performing tasks in
78 stereotypically male gendered domains (e.g., advanced math ematics ). Further, p rior self effic acy research has shown women report lower MSE levels than men, even when perf orming at comparable levels ( e.g., West, Welch, & Knabb 2002 ). Exploratory causal analyses were somewhat consistent with these findings. In the best fitting model, a significant path emerged from gender to MSE for participants of all ages. Gender was a ls o found to predict memory related anxiety, though this path was significant for younger adults only. Future investigation of the role of gender vis vis stereotype effects, within both stereotypically gendered and gender neutral domains, seems warrante d. Baseline levels of anxiety among older adults also may have been rele vant in in response to the stereotype activation. Again, in spite of counterbalanced assignment to condition, baseline anxiety was highest in the posi tive ster eotype group and lowest in the negati ve stereotype group. When reanalyzing the data with baseline anxiety as a covariate, the results for memory and self efficacy still were not consistent with planned hypotheses for these variables. Neve rtheless, i t is possible that these sample characteristics may h ave created a differential reaction to the condition manipulation as will be discussed later (see memory anxiety section). Th e ideas above have f ocus ed on sample characteristics the assumption being that some qualities inherent to participants may have impeded or even nullified an otherwise effective stereotype activati on. I n keeping with this idea it is interesting that limited empirical attention (e.g., Eibach et al., 2010) has focused on the divergent contrast and assimilation effects such s tereotype inductions can have that is, why
79 stereotype effects can influence some people but not others F or some, exposure to stereotypes about a group may elicit thoughts about how su ch beliefs are personally applicable (assimilation effect). Others, however, may decide such ideas are not personally relevant (contrast effect) even if they do belong to the stereotyped group. An example applied to the current context is illustrative. N egative aging stereotypes may lead some older adults to assimilate by identifying supporting evidence from their personal experience consistent with the stereotype. In the current study, when older adults heard a statement about age differences in memo ry performance (i.e., young are superior), this may have activated retrieval of personal experiences consistent with this idea. Conversely, the same stereotype may have le d other s to identify evidence that contrasts with the stereotype, rendering it perso nally inapplicable. Other research has begun to investigate the importance of stereotype content (e.g., positive vs. negative valence ; cognitive vs. physical ) in relation to differential effects on performance finding, for example, that when stereotype content matches outcome domain (e.g., cognitive, physical) its valence (e.g., positive vs. negative) has a greater impact on subsequent performance (e.g., Levy & Leifheit Limson, 2009). The current study investigated performance in a cognitive domain, me mory, and included some, but not exclusively, memory based stereotype content. It is unknown to what extent this could have weakened or diluted any stereotype condition effects. Broadly, t his raises questions about the impact of perceived self releva nce vis vis stereotype activation and threat Does belonging to a stereotyped group or attempting a task in a stereotyped domain suffice for threat to occur? What if one does
80 not identify with the group or is no t familiar with the particular stereotype ? What if the stereotyped domain does not hold value for the individual? For reasons explicated e arlier in the case of memory and the aged, this seems unlikely. However, even in the case of ageist stereotypes, stereotype manipulations are not always su ccessful ( Hess, 2008, personal communication ). Therefore, t he current study attempted to empirically assess factors that might affect susceptibility to the manipulation such as domain identification and subjective age, albeit in an exploratory fashion. T hese analyses will be discussed in more detail later (see exploratory analyses). Aside from per sonal attributes of the sample, the stereotype manipulation employed in this study should be considered as well. From a methodological perspective, it is unclear what impact the use of a telephone survey had on the study as a whole. This is particularly tru e of the stereotype activation, which has not been done via telephone in prior work. It is unclear if an in person interview might have been more effective at activating stereotype effects. I t seems intuitive that face to face communication might be p referable especially with an older adult sample for which hearing might be less acute. However, the telephone interview script was patterned closely after previously successful stereotype manipulations (e.g., Hess et al., 2004) published in the extant li terature. During p ilot trials interview length seemed acceptable as participants were able to follow instructions and perform tasks appropriat ely, and MSE and memory performance were highest in the positive condition and lowest in the negative condition, as predicted Similarly, older adults in the current study were able to follow directions and complete the required tasks. Anyone whose ability to participate seemed at all questionable was administered the TICS, which in the current sample,
81 occurred on ly once. Thus, the interview, itself, did not appear to be problematic in any way. It could still be the case, however, that stereotype manipulations are more effective in person, and future studies could address this issue. Memory Self Efficacy As emp hasized throughout this paper, extensive empirical literature has found older adults to have lower MSE than their younger counterparts. This is particularly important given that studies have also found that MSE predicts memory performance (e.g., West & Ya ssuda, 2004 ; Bensadon, 2007). With regard to performance, other factors, including other types of beliefs (e.g., stereotypes) have also been shown to play a predictive role. This begs the empirical question which is mo re important? The current study employed a unique methodology to test precisely that, examining MSE and stereotypes, simultaneously. No significant change in MSE level resulted from either positive or negative stereotype manipulations, in comparison to the control group. This lack of condition effects underscores the inherent difficulty of externally efficacy, which the positive stereotype condition intended (and was predicted) to do. Though Bandura identifies such external influence s (i.e., social persuasion) as a key source from which self efficacy is derived, over time, he does suggest it is more easily reduced than improved, as corroborated by results here and from memory intervention and goal setting studies showing that short te rm (beneficial) changes in self efficacy are empirically challeng ing (e.g., Floyd & Scogin, 1997; West & Thorn, 2001). Similarly, stereotype research has shown that negatively valenced stereotypes may be more easily internalized and quicker to exert their impact than those that are positively valenced, though both are possible (e.g Levy 2009).
82 In keeping with results from prior studies (e.g., West et al., 2005 ) and consonant with the author s prediction older adults in the current sample exhibited l ower levels of MSE than the younger adults These findings are consistent with the posited differential salience of th e memory domain as people age. In addition, bivariate correlations showed MSE and memory performance were significantly correlated MSE was also found to play a key role in several relationships among variables tested in exploratory structural equation models. As mentioned, the most parsimonious model included older adults only and showed MSE significantly predict ing performance acco unt ing for nearly 10% of the variance in memory recall scores. Considering that MSE is largely a Theoretically, Steele (1997) postulates that anxiety is the key variable influenced by stereotype activ ation, that subsequent ly threatens efficacy theory, however, states that anxiety or arousal is one of self precursors. In other words, anxiety, like other physiological arousal or affective states (i.e., depression) is a threat, but to self efficacy, which in turn influences subsequent performance. This proposed theoretical chain of events wa s precisely t hat elucidated by th e final model. erformance, t he path from anxiety to MSE was statistically significant and accounted for more than 20% of MSE variance. To reiterate, condition differences did not supp ort hypotheses, the testing environment seems to have had some impact on anxiety, particularly among older adults Both age groups showed comparable baseline anxiety levels, and a significant increase in anxiety at
83 follow up. This increase, however, was larger for older adults Like other arousal states, anxiety, as addressed by Bandura (1997), serves as a primary indicant of self efficacy. The indirect pathway from anxiety to MSE and then to performance was also confirmed and significant Again, as explicated by self efficacy theory, anxiety level is a source and therefore predictive of self efficacy level. H igher anxiety was associated with lower self efficacy. In fact, aside from relationships between latent variables and their predi ctors (e.g., measurement model), the path from anxiety to MSE constituted the strongest loading. In sum, the final mod el illustrated MSE multi faceted role with regard to o lder adult memory performance Memory Anxiety Thus far it is clear that aging s tereotypes and MSE were the primary predictors examined in the current study as both have garnered research support suggesting they directly influence memory performance. Anxiety was also included in the study, given its theoretical importance to both sel f efficacy and stereotype threat. As described earlier, anxiety and other types of affective states (e.g., depression) or physiological arousal were identified by Bandura (1997 ) as one of the four primary sources of self efficacy. The theory suggests tha t people are attentive to such states during task performance, which can increase or lower their perceived efficacy. If one feels calm while trying to remember something, efficacy may improve or at least remain at its current level. Conversely, if one st ruggles to remember something, this may increase anxiety and thus According to stereotype threat theory (Steele, 1997), stereotyped group members performing in a stereotyped domain may worry about confirming or phrased di fferently, feel pressure to disconfirm the prevailing stereotype Herein lies the threat which can
84 manifest as heightened anxiety when performing in the stereotyped domain. Based in part on this theory, the current study hypothesized that hearing ster eotype consistent messages about memory and aging would increase memory anxiety levels of older adults (for whom stereotypes about memory pervade). By and large analyses from the current study did not support these hypotheses. Stereotype condition and age were not significant predictors of memory anxiety level nor was their interaction Baseline anxiety and follow up memory anxiety measured during the phone interview were highly correlated. Initial analyses revealed younger and older participants alik e displayed comparable baseline anxiety levels. However, o nce baseline anxiety was covaried out, a significant age effect did emerge, with older adults showing higher levels of memory anx iety across conditions. In modeling these relationships, it was app arent that anxiety affected self efficacy but played no direct role in performance for either age group. In interpreting these findings, several ideas seem worthy of discussion. First, a s addressed earlier, in spite of randomization, baseline anxiety was not equally distributed across stereotype condition s Among older adults, highest baseline anxiety levels were found in the positive stereotype condition followed by the control and negative stereotype conditions. Though speculative, it stands to reason that such manipulation. Theoretically, the higher levels of anxiety in the positive condition may have created a suppression effect, nullifying any positive boost from positive ly valenced stere otypes and rendering their impact non significant. Conversely, the low anxiety levels in the negative stereotype condition may have served as a buffer to any
85 performance damaging impact that negat ively valenced stereotype content might ha ve otherwise exerted. Though this explanation is theoretically possible, condition remained non significant even after baseline anxiety was covaried out, making such an inter pretation speculative at best. R elated hypotheses were not supported in terms of stereotype valence, but closer inspection suggests the mere discussion of age and memory performance, in tandem, may have had some impact on anxiety Analyses revealed a significant increase in follow up anxiety measured as part of the telephone interview for all ages. However, this increase was stronger for older adults as evidenced by the 2 way interaction between time of a ssessment and age. Given the non significant results of condition, directly attributing the anxiety increase to stereotype effects, however, would be conjecture. Though c ondition findings d id not support ome authors have suggested a ny testing or evaluative environment, itself, can lead to increased anxiety ( Desrichard & Kopetz, 2005 ). It remains unclear to what extent the experimental condition itself may have increased anxiety, or whether the mere mention of memory was sufficient. Regardless, it seems safe to conclude from the current findings that memory is a particularly sensitive functional domain for older adults and that anxiety is affected by testing situations for both age groups. Exploratory Analyses Given their theoretical importance, domain i dentification and subjective age were measured in a single item pre post format. Stereotype threat theory explicit ly mention s domain identification as directly r elevant to performance. Those f o r whom the domain holds most importance are theoretically most vulnerable to threat when performing in that domain. Backed by an extensive literature review, t he author identified memor y as
86 a domain of particular salien ce to older adults. Explicit t ests of this with the current sample, however, generally did not support this notion. Domain identification was not a significant predictor of any of the outcome va riables in this study. Further, it was not that high domain identifi cation corresponds with high susceptibility to stereotype threat. Possible explanations for the single items have been used in prior literature, ensuring reliability can be challenging. The current study attempted to account for thi s by capturing these data twice. Although the two items were significantly correlated more extensive assessment of domain identification may be needed. Similarly, subjective age was measured in an exploratory fashion. The assumption throughout this paper, and cognitive aging literature as a whole, is that older and can be grouped accordingly yet direct study of this assumption is rare Because many of the hypotheses here we re focused on chronological age, the current study explored the importance of subjective age. This was measured in the same manner as domain identification, and yielded several non significant results. Perhaps most interesting among these was a non significant correlation with chronological age, thus providing empirical suppo rt that age as a meaningful performance predictor and demographic grouping variable may be more than just a chronological number. The fact that s ubjective age was not a significant predictor of the primary outcome variables wit h the current sample may r elate more to the single item
87 measurement structure. This methodology may have been too limited to detect any predictive relationship with memory performance or MSE. Given the state like nature of perceived age which may be heavily influenced by mood or other seemingly unstable affective factors, accurate measurement seems problematic. Methodology notwithstanding, subjective age was significantly correlated with both baseline and follow up memory anxiety, and health. Though beyond the scope of this paper, these relationships might be worthy of further exploration in future cognitive aging work. Structural equation analyses revealed two additional variables to be relevant for older adults subjective health an d education level. Not surprisingly, t he causal path from health to anxiety was found to be significant for ol der adult s only (figure 4 4 ) Although anxiety items were specific to memory, it could be that individuals experiencing health difficulties are more vigilant about memory and concerned that their health problems will lead to memory problems as well. Again, in terms of methodology, a snapshot of perceived health was assessed with only one item, while anxiety was assessed with multiple items all m emory specific in content and administered b oth at baseline and during the phone interview. Therefore, it seems safe to conclude that beliefs about memory were tapped more deeply than subjective health in the current study. However, while a single item health measure should be interpreted with caution, s tudies across various disciplines have demonstrated the adequacy and predictive power of this approach finding particularly robust associations between single self perceived health measures and mortality (e.g., Idler & Benyamini, 1997; DeSalvo et al., 2006).
88 The finding that formal education was predictive of superior performance in a cognitive domain such as memory is theoretically intuitive and empirically consistent with some memory aging literature (e. g., Huppert et al., 2000). Conversely, in terms of stereotype threat some literature has revealed the opposite negative impact on memory performance most powerful among those highest in education (e.g., Hess et al., 2007 ) Th e current result it seems, can be explained at least in part through a self efficacy lens. More formal education increases opportunities for enactive mastery experience, in this case, memory related cognitive tasks. In fact, w hile definitions and types of education may vary, it stands to reason that virtually any type of new learning requires some degree of memory function. Congruent with self efficacy theory, empirically, those higher in education should display higher self efficacy, which as ex plicated t hroughout this paper, would be closely tied to better performance. Results demonstrating an education performance link in the current study would seem to favor the older adult s since the sample data reveal ed on aggregate, older participants have amassed mo re education than the younger adults. However, confounding this conclusion is the fact that older adults, including those in the current sample tend to display lower MSE levels than their young er counterparts. Perhaps th ese seemingly conflicting ideas are more palatable when considering at once the potential impact of stereotype threat the uniqueness and differential salience of the memory domain within the broader cognitive arena and educational status of the current s sample.
89 On one hand, by definition, elders have had more years of life experience, which in th is sample included years of education, during which they have theoretically needed to utilize their memory more often and in more varied ways th an their young er counterparts. However, the younger adult sample in the current study was comprised exclusively of undergraduate university students. Though exceptions may exist, it seems safe to conclude that college in the United States embodies a deve lopmental process where young people increasingly assert their independence and discover personal strength s and weakness es Integral to this process is the identification from within and via external sources (e.g., social persuasion) of domains in which they perceive themselves as more or less efficacious. Over time s uch perceptions take hold and increase in permanence ultimately shaping the young person adult identity. Beyond th is temporal component i n practical terms, college represents a period where memory reliance peaks, with students consistently required to absorb, retain, and later retrieve information (e.g., exams, written research papers, oral presentations, notetaking ) Framed in self efficacy terminology younger adults currently enrol led in college are likely presented with and engaged in enactive mastery opportunities more than ever. In spite of this theoretical rationale structural equation mod els with the full sample found a significant education perfo r mance relationship with olde r adults only suggesting this link is not as strong with younger adults However, t his was likely due to limited variability in young er adult education level, since nearly two thirds ( 62% ) of younger adults in the current sample were underclassmen (e.g., 12 or 13 years of educational experience). It remains to be seen whether such a relationship would prove
90 significant with a more heterogeneous sample of younger people (e.g., high school freshmen vs. advanced doctoral students). Study Limitations Thoug h informed by prior empirical designs the use of telephone interviews is rare in the stereotype threat literature. The extent to which this was a limitation is unclear, though the non significant findings for stereotype condition suggest this may have bee n relevant. It should be noted however, that those conducting the ph one interviews participated in several weeks of intensive training and subsequent evaluation of role played interviews prior to contacting study participants. In addition, the telephone interview script was patterned after previous ly successful stereotype manipulations (e.g., Hess et al ., 2004 ) published in the extant literature. Though not focused on stereotype effects, other cognitive aging studies examining memory have found comparable validity between in person and telephone interview methodologies (West et al., 2008). Thus, it is not clear whether the condition manipulation would have been more effective using in person interviews. Interview method aside, condition assign ment may have been a limitation As mentio ned earlier, participant gender and anxiety levels were not evenly distributed across the three conditions. It is unclear to what extent this affected current results. As mentioned earlier, while the role of gen der has been investigated in stereotype literature, it has been considered less frequently within cognitive aging. Regarding anxiety, though baseline levels varied across condition, they were comparable for younger and olde r adults. Another methodologica l limitation pertained to assessment of subjective age and domain identification. As mentioned earlier, these variables were measured with single
91 items, which, generally speaking, can be limited in terms of reliability. In light of this, each item was ut ilized in an exploratory fashion and assessed twice, both before and during the telephone survey. Furthermore, although both seem central to theoretical tenets of stereotyping and self efficacy literature, empirically, neither concept has been examined in a consistent or comprehensive manner. A similar lack of consensus exists when determining how to measure anxiety. The current study relied on the MIA anxiety subscale, a broadly accepted and well established tool for tapping memory specific anxiety. Comparisons of baseline and follow up anxiety were made based on slightly different assessments. As mentioned earlier, participants were initially administered the MIA anxiety subscale plus distractor items created by the author as a means of disguising the memorial component of the stu dy. This instrument was completed by participants in person on paper. Follow up anxiety was measured with the MIA anxiety subscale in its original form without any distractor items. Further, these questions were adminis tered by student interviewers via telephone. The actual items were the same, both sets of items were significantly correlated with each other, and both assessments were found to have good reliability with the current sample. But i t is unclear what impact the subtle differences in administration may have had on these results A final potential limitation concerns the content of the stereotype adjectives used. Though memory specific stereotypes and stereotypic messages were included in the current study, most stereotype prim es related to age more broadly (i.e., wise, respected, confused brittle ). T his does not appear problematic since the primary stereotype activation in keeping with published designs consis ted of messages explicitly linking
92 memory pe rformance and ag e In fact, t study has been guided by the theoretical importance of content match (e.g., L evy et al., 2009) These authors grouped aging stereotype content into physical and cognitive domains and s imilar to the current approach th ey included material related to cognitive health but not necessarily memory specific ( i.e., sage, alert, dementia, confused). Further, when employing adjectival primes, it is unclear to what extent a pos itive stereot ype condition with exclusively memory specific content is possible Positively valenced stereotypes about older adult memory are rare to non existent Thus, for memory, it would be very challenging, if not impossible, to create testing conditions using b oth negatively valenced and positively valenced adjectives that were all tied directly to the domain of memory.
93 CHAPTER 7 CONCLUSION Clinical Implications As age increases, so does the likelihood of some form of decline, which may or may not be amenable to clinical treatment. Beliefs, however, central to both stereotype effects and self efficacy, lie well within the purview of psychological intervention. As centra l concern to individuals as they age Supporting evidence for this was found in the current investigation which showed lower MSE levels among older participants across all stereotype conditions. Reinforcing this concern, unfortunately, are myriad negativ e and ageist societal messages, often rooted in stereotypes, which can discredit, discount, devalue, or even insult older adults (e.g., Kite et al., 2005). The deleterious consequences of aging stereotypes have been well documented, and these external mes sages can be internalized as self beliefs. This can result in psychological damage, performance, independent of objective ability or capacity. Similarly, self effica cy can b e influenced by external sources via social persuasion (Bandura, 1997 ), and can subsequently impact performance. The more efficacious people fee l, the likelier they are to perform successfully, or persist until they achieve a successful outcome (i.e., mastery). Conversely, those feeling inefficacious may shy away from experiences they perceive as threatening thus limiting opportunities for improvement and enhanced efficacy (e.g., West et al., 2003 ). Clearly, beliefs occupy a central role in de how the interplay of competing beliefs (i.e., self efficacy & stereotypes) can impact
94 performance. By manipulating aging stereotype activation and incorporating measures of memory self effica cy, the current study carried out a novel appr oach to understanding just that and yielded some results worthy of cautious optimism. Evidence from the current study underscores the importance of beliefs in determining performance outcome. Here, MSE, and a account ed for percentages of variance comparable to that of chronological age alone. In light of this finding, r ather than accept a reality of inevitable decline, it seems elders and the clinicians who serve/treat them alike, may f ocus on a more active approach to self efficacy At the risk of stating the obvious, this need not be limited to memory or aging. Self efficacy theory, predicated on a bel ief performance link, is applicable to any functional domain. The preponderance of data from the current study showed non significant effects of stereotype condition. Nonetheless, as illustrated by the extensive literature review summarized earlier, stere otypes can exert an impact, often negative, on performance, and there is little reason to assume such beliefs will somehow disappear from society In light of this reality, the current results underscore the importance of believing in self, and suggest cl inical attempts to empower older adults by incrementally enhancing their personal efficacy may provide them with an adequate buffer to any negative impact Regarding stereotype beliefs more specifically, interventions may be directly informed by self efficacy theory Clinicians may foster non threatening opportunities for enactive mastery (e.g., identifying realistically attainable memory goals), while simultaneously using verbal persuasion to counter negative stereotypical messages while increas ing the frequency, salience of and exposure to positive ones The data presented here
95 argue for a focus on efficacy enhancement a result of which might include reduction of uncomfortable affective states such as anxiety. U ltimate ly, the goal of such int ervention should be ensuring virtually anyone will be able to achieve peak memory performance in spite of stereotyped beliefs asserting the contrary Theoretical Implications The stereotype literature reviewed in this paper identifies areas of uncertain ty regarding the mechanisms underlying the impact that stereotypes have on people, particularly with regard to their performance in a given domain. Theories of stereotype threat (Steele, 1997) and self stereotyping (Levy & Langer, 1994; Levy, 1996) have g arnered the most attention and have informed the majority of research designs carried out to date, particularly in the domain of memory performance. As the current literature review indicates, these two approaches share some overlap yet are also distinct explications, and both have been bolstered by some empirical support. However, these theories, alone, do not sufficiently explain the entire stereotype effects performance picture. Researchers have therefore sought to identify variables that may moderate or mediate the stereotype performance relationship. Empirically, these efforts have yielded mixed results. Largely absent among these has been self efficacy. Considering its direct theoretical relevance, this absence is particularly curious. T he current study focuse d on the impact of both stereotype effects and self efficacy on performance, in an attempt to provide the empirical basis on which to further extend theories of stereotype effects to incorporate the self efficacy construct. The non significant findings regarding condition differences however, make any conclusions about the impact of stereotypes speculative at best. As alluded to earlier, not everyone is equally influenced by stereotypes. Why do some stereotypes seem to have an
96 imp act on performance of some individuals but not others? Are all stereotypes created equal? Current theory attempting to elucidate the stereotype performance link does not sufficiently explain this Are individual differences (e.g., personality traits/ cha racteristics ) most important? Are certain types of stereotypes more powerful than others (e.g., negative vs. positive) ? Do certain stereotyped contexts or domains leave individuals more vulnerable? C alls to address these nuances have led to some interes ting recent research, most notably that of Levy & coll e agues (2009) who have discovered matching stereotype content with outcome domain (e.g., cognitive or physical) increases the strength of stereotype performance associations. Given the abundance of existing stereotypes about people and their performance in various functional domains, and self across human performance, future work may aim to extend both theories to more directly answer such questions an d further explore their respective relevance to each other and resulting performance. If stereotypes and their effects differ from other forms of soci al persuasion, this should be explicated in self efficacy theory. If stereotype effects exert their impa ct on performance by altering self efficacy, this should be included in theoretical models of stereotyping. These are merely two examples but t o date, neither of these ideas has been investigated systematically. Current findings provide some evidence for the relevance of a nxiety F urther theoretical expansion may in stereotype effects, self efficacy, and performance, both within memory and beyond. Research Implications The fact that s tereotypes abound for virtually every categorizable group and functional domain is not a novel phenomenon Regardless of whether one views
97 stereotyp ing as inevitable, there is little reason to believe th at such categorization will disappear in the near future. Psychologists h ave focused exte nsively on the potential links between stereotyped individuals and their performance in a stereotyped domain. Initial studies focused on gender and racial differences in academic achievement (Steele & Aronson, 1995; Spencer et al., 1997). More recently, perhaps fueled by the increase in human longevity, cognitive aging researchers have applied stereotype 2006; Hess & Hinson, 2006; Hess et al., 2007). In addit ion, some have theorized that age based (e.g., ageist) stereotypes have unique features, such as holding no self relevance until late in life, not being proscribed by a culture of political correctness, and the simple fact that unlike other prejudicial are as (e.g., race, gender, sexual orientation), with ageism, everyone, assuming they live long enough, eventually belongs to the group being stereotyped. It has been argued that this uniqueness may intensify the impact of such stereotypes (e.g., Levy, 2009). Though results have varied at times, the link between stereotype effects and performance has been empirically established. Mechanisms through which the stereotypes exert their impact, however, remain an empirical mystery. Various me diators and moderators have been proposed and tested, yielding mixed findings Hess et al., 2007). The current study attempted to build on this research by including memo ry self efficacy, consistently shown to be directly related to memory performance (e.g., Bensadon, 2007; West et al., 2007; West et al., 2008), but conspicuously absent from
98 the stereotype literature Although stereotype effects on both performance and se lf efficacy were not statistically significant in the current study, this may have been related in part to sample characteristics as opposed to the stereotype activation itself. Ideally, participants would have shown equivalent baseline anxiety levels acr oss conditions. Counterbalancing ensured this to the extent possib le, though differences remained Such challenges not withstanding, future empirical work, including similar designs or replications conducted in person might shed further light on how b est to activate stereotypes as consistently and efficiently as possible. Methodological consensus might allow for greater focus on the impact such effects may exert on memory anxiety, self efficacy, and ultimately, performance In addition, further research is needed to test theory asserting the uniqueness of age based stereotypes, and the importance of stereotype content matching the performance domain (e.g., Levy, 2009; Levy & Leifheit Limson). Though memory was included in th e current study, part of the stereotype manipulation relied on stereotypes related to age more broadly (i.e., not memory specific). Beyond stereotype condition, t he exploratory models reported here did offer further empirical the relationship between anxiety and memory performance. Given the theoretical significance of anxiety to both self efficacy and stereotype threat, future research should replicate the current design by elucidating a model of performance that include s both s tere otype effects and self efficacy, each shown to inf luence performance individually. Although stereotype effects did no t have a significant impact with the current sample, i deally, th is project may provide the buildi ng blocks for future empirical attemp ts both to disentangl e the relationships among anxiety,
99 stereotyping and self efficacy, and to examine their multivariate effects on performance Results from the current study provide a strong case for anxiety and self efficacy to play a central role in future efforts.
100 APPENDIX A RECALL LIST (POSITIV E) buttery dignified idea aluminum bud kind lift wise clear bush silky can blossom respected toast calf nail blue insightful sound
101 APPENDIX B RECALL LIST (NEGATIV E) sky brittle red damp confused hot tooth forgetful curly leg complaining button lock screen senile detergent circling scrubs snap pan
102 APPENDIX C RECALL LIST (AGING NEUTRAL) high beat costly foil breeze green ink southern large paper flat merger drum wavy shoe carton glass boat yelling metallic
103 APPENDIX D PHONE SURVEY SCRIPT I am calling because you are scheduled to complete a brief telephone interview today. This interview should take 20 30 minutes. Is this a good time to do the interview? During this phone call, I will ask you a few questions and ask you to do some exercises that involve remembering and making judgments about words. Your participation is completely voluntary. If you prefer not to answer any question, just let me know and we will move on to the next questi on. The information you give me will be confidential. All of your answers will be identified by a computer code. Your name and identifying information will not be attached to your responses. This phone call will take about 20 minutes. We will be recor ding the interview today so that we can review your responses later, if need be. Feel free to stop me if you have any questions about the directions. Do I have your permission to go ahead with the interview?
104 LIST OF REFERENCES Baltes, P.B. (1997). On the incomplete architecture of human ontogeny: Selection, optimization, and compensation as foun dation of developmental theory. American Psychologist, 52(4) 366 380. Bandura, A. (1993). Perceived self efficacy in cogniti ve d evelop ment and functioning. E ducational Psychologist, 28(2), 117 148. Bandura, A. (1997). Self efficacy: The exercise of control. New York: Freeman. Bargh, J.A., Chen, M., & Burroughs, L. (1996). Automatic ity of social behavior: Direct effects of trait cons truct and stereotype activation on action. Journal of Personality and Social Psychology, 71, 230 244. Baron, R. M., & Kenny, D. A. (1986). The moderator mediator va riable distinction in social psychological research: Concept ual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51 1173 1182. Bensadon, B.A. (2007). Self efficacy and memory aging University of Florida. Bensadon, B.A., West, R.L., & Bagwell, D.K. (2007, August). The impact of domain specific self efficacy on memory in middle aged and older adults. Poster presented at the annual convention of the American Psychological Association, San Francisco, CA. Blair, I.V. (2002). The malleability of automatic stereotypes and prejudice. Personality and Social Psychology Review, 6(3), 242 261. Brandt, J., Spencer, M., Folstein, M. (1988). The telephone interview for cognitive status. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, 1, 111 117. Butler R.N. (1969). Ageism: another form of bigotry. Gerontologist, 9, 243 24 6. Cavanaugh, J.C., & Poon, L.W. (1989). Met amemorial predictors of memory performance in young and older adults. Psychology and Aging, 4(3), 365 368. Chasteen, A.L., Bhattacharyya, S ., Hohota, M., Tam R., & Hasher, L. (2005). How Experimental Aging Research, 31, 235 260. Chasteen, A.L., Schwarz, N., & Park, D.C. (2002). The activation of aging stereotypes in yo unger and older adults. Journals of Gerontology: Psychological Sciences, 57B(6), P540 P547. Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1), 155 159.
105 Colcombe, S., & Kramer, A.F. (2003). Fitness effects on the cognitive function of older a dults: A meta analytic study. Psychological Science, 14(2), 125 130. Dal Forno, G., Chiovenda, P., Bressi, F., Ferreri, F., Grossi, E., Brandt, J., Rossini, P.M., & Pasqualetti, P. (2006). Use of an Italian version of the telephone interview for cognitiv Inte rnational Journal of Geriatric Psychiatry, 21, 126 133. Dark Freudeman, A., West, R. L., & Viverito, K. M. (2 006). Future selves and aging: Educational Gerontology, 32 85 109. DeSalvo, K.B., Bloser, N., Reynolds, K., He, J., & Muntner, P. (200 6) Mortality prediction with a single general self rated health question: A meta analysis. Journal of General Internal Medicine, 21 (3), 267 275. Desmond, D.W., Tatemichi, T.K., & Hanzawa, L. (1994). The telephone interview for cognitive status (tics): Reliability and validity in a stroke sample International Journal of Geriatric Psychiatry, 9(10), 803 807. Desrichard, O., & Kopetz, C. (2005). A threat in the elder: The impact of task instructions, s elf efficacy and performance expectations on memory performance in the elderly. European Journal of Social Psychology, 35 537 552. Devine, P.G. (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal of Personality and Social Psychology, 56, 680 690. Dixon, R.A., & Hultsch, D.F. (1983). Structure and development of metamemory in adulthood. Journals of Gerontology, 38, 682 689. Dixon, R. A., & Hultsch, D. F. (1984). The metamemory in adulthood (MIA) instrument. Psychological Documents, 14, 3. (Ms. No. 2605 ). Eibach, R.P., Mock, S.E., & Courtney E.A. (2010). Having a senior moment: Induced aging phenomenology, subjective age, and susceptibility to ageist stereotypes. Journal of Experimental Social Psychology, 46 643 649. Fisk, S.T., & Taylor, S.E. (1991). Social Cognition, McGraw Hill: New York. Floyd, M., & Scogin, F. (1997). Effects of memory training on the subjective memory functioning and mental health of older adults: A meta analysis. Psychology and Aging 12(1), 150 161. Freund, A. M., & Ebner, N. C. (2005). The aging self: Shifting from promoting gains to balancing losses. In W. Greve, K. Rothermund, & D. Wentura (Eds.), The adaptive self: Personal continuity and intentional self development (pp. 185 202). New York: Hogrefe.
106 Guo, X., Erber, J.T., & Szuchman, L.T. (1999). Age and forgetfulness: Can st ereotypes be modified? Educational Gerontology, 25, 457 466. Head, D., Rodrigue, K.M., Kennedy, K.M., & Raz, N. (2008). Neuroanatomical and cognitive mediators of age relate differences in episodic memory. Neuropsychology, 22(4), 491 507. Hertzog, C., Dixon, R.A., & Hultsch, D.F. (1990). Relationships between metamemory, memory predictions, and memory ta sk performance in adults. Psychology and Aging, 5(2), 215 227. Hess, T.M., Auman, C., Colcombe, S.J., & Rah hal, T.A. (2003). The impact of s tereotype threat on age differences in memory performance. Journals of Gerontology: Psychological Sciences, 58B (1) P3 P11. Hess, T.M., & Hinson, J.T. (2006). Age related variation in the influences of aging stereotypes on memory in adulthood Psychology and Aging, 21(3), 621 625. Hess, T.M., Hinson, J.T., & Hodges, E.A. (200 7 in press). Moderators of and mechanisms und performance. Experimental Aging Research. Hess, T.M., Hinson, J.T., & Statham, J.A. (2004). Explicit and implicit stereotype activation effects on memory: Do age and awareness moderate the impact of priming? Psychology and Aging, 19(3), 495 505. Hu, L., & Bentler, P.M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multid i sciplinary Journal, 6(1), 1 55. Hultsch, Hertzog, C., Dixon, R.A., & Davidson, H. (1988). Memory self knowledge and self efficacy in the aged. In M. L. Howe & C. J. Brainerd (Eds.), Cognitive development in adulthood: Progress in cognitive development res earch (pp. 65 92). New York: Springer Verlag. Hummert, M.L. (1990). Multiple stereotypes of elderly and young adults: A comparison of structure and evaluations. Psychology and Aging, 5(2) 182 193. Hummert, M.L., Garstka, T.A., Shaner, J.L., & Strahm, S. ( 1995). Judgments about stereotypes of the elderly: Attitudes, age associations, and typicality ratings of young, middle aged, and elderly adults. Research on Aging, 17(2), 168 189. Huppert, F.A., Johnson, T., & Nickson, J. (2000). High prevalence of pros pective memory impairment in the elderly and in early stage dementia: Findings from a population based study. Idler, E. & Benyamini, Y. ( 1997 ). Self rated health and mortality : A review of twenty seven community studies. Journal of Health and Social Behavior 38 2 1 37
107 Jost, J., & Banaji, M. (1994). The role of stereotyping in system justification and the production of false consciousness. British Journal of Social Psychology, 33, 1 27. Kite, M.E., & Johnson, B.T. (1988). Attitudes toward older and y ounger adults: A meta analysis. Psychology and Aging, 3, 233 244. Kite, M.E., Stockdale, G.D., Whitley Jr., B.E., & Johnson, B.T. (2005). Attitudes toward younger and older adults: An updated meta analytic review. Journal of Social Issues, 61(2), 241 266 Levy, B.R. (1996). Improving memory in old age through implicit self stereotyping. Journal of Personality and Social Psychology, 71(6) 1092 1107. Levy, B. R. (2000). Handwriting as a reflection of aging self stereotypes. Journal of Geriatric Psychiatry, 3 3, 81 94. Levy, B. R. (2003). Mind matters: Cognitive and physical effects of aging self stereotypes. Journal of Gerontology: Psychological Sciences, 58B P203 P216. Levy, B.R., Hausdorff, J.M., Hencke, R., & Wei, J.Y. (2000). Reducing cardiovascular stre ss with positive self stereotypes of aging. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 55 (4), P205 P213. Levy, B.R., & Langer, E. (1994). Aging free from negative stereotypes: Successful memory in china and among the american deaf. Journal of Personality and Social Psychology, 66(6), 9 89 997. Levy, B.R., Slade, M.D., Kunkel, S.R., Kasl, S.V. (2002). Longevity increased by positive self perceptions of aging. Journal of Personality and Social Psychology, 83(2), 261 270. Levy, B.R. (2009). Stereotype embodiment: A psychosocial approach to aging. Current Directions in Psychological Science, 18(6), 332 336. Nelson, T. D. (2002). Preface. In T. D. Nelson (Ed.), Ageism: Stereotyping and prejudice against older adults Cambridge, MA: MIT Press. aged adults: Contrasting self stereotyping and stereotype threat accounts of assimilation to age stereotypes. Social Cognition, 24 338 358. Personality and Social Psychology Bulletin 29(6), 782 789. Poon, L. W. (1985). Differences in human memory with aging: Nature, causes, and clinic al implications. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the psychology of aging ( 2nd ed., pp. 427 462). New York: Van Nostrand Reinhold.
108 Rapp, S., Brenes, G., & Marsh, A.P. (2002). Memory enhancement training for older adults with mild cognitiv e impairment: A preliminary study. Aging and Mental Health, 6(1), 5 11. Salthouse, T.A. (2009). When does age related cognitive decline begin? Neurobiology of Aging, 30(4), 507 514. Seibt, B. & Frster, J. (2004). Stereotype threat and performance: How se lf stereotypes influence processing by inducing regulatory foci. Journal of Personality and Social Psychology, 87(1), 38 56. Shapiro, J.R., & Neuberg, S.L. (2007). From stereotype threat to stereotype threat: Implications of a multi threat framework for causes, moderators, mediators, consequences, and interventions. Personality and Social Psychology Review, 11 107 131. Spaniol, J., Madden, D.J., & Voss, A. (2006). A diffusion model analysis of adult age differences in episodic and semantic long term memory retrieval. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32(1), 101 117. Spencer, S.J., Steele, C.M., & Quinn, D.M. (1 math performance. Journal of Experimental Social Psychology, 35(1), 4 28. Steele, C.M. (1997). A threat in the air: How stereotypes shape intellectual identity and performance. American Psychologist, 52(6), 613 629. St ein, R., Blanchard Fields, F., & Hertzog, C. (2002). The effects of age stereotype priming on the memory performance of older adults. Experimental Aging Research, 28, 169 181. Wentura, D., & Brandtstdter, J. (2003). Age stereotypes in younger and older wo men: Analyses of accommodative shifts with a sentence priming task. Experimental Psychology 50 16 26. West, R.L., Boatwright, L.K., & Schleser, R. (1984). The link between memory performance, self assessment, and affective status. Experimental Aging Rese arch, 101(4), 197 200. Welch, D.C., & West, R.L. (1995). Self efficacy and mastery: Its application to issues of environmental control, cognition, and aging. Developmental Review, 15, 150 171. West, R.L., Crook, T.H, & Barron, K.L. (1992). Everyday memory performance across the life span: Effects of age and noncognitive individual differences. Psychology and Aging, 7(1), 72 82. West, R.L., Dennehy Basile, D., Norris, M.P. (1996). Memory self evaluation: The effects of age and experience. Aging, Neuropsychol ogy, and Cognition, 3(1), 67 83.
109 West, R. L., Welch, D. C., & Thorn, R. M. (2001). Effects of goal setting and feedback on memory performance and beliefs among older and younger adults. Psycholo gy and Aging,16, 240 250. West, R.L., Welch, D.C., & Knabb, P.D. (2002). Gender and aging: Spatial self efficacy and location recall. Basic and Applied Social Psychology, 24(1), 71 80. West, R.L., Thorn, R.M., & Bagwell, D.K. (2003). Memory performance and beliefs as a function of goal setting and aging. Ps ychology and Aging, 18(1), 111 125. West, R.L., & Yassuda, M.S. (2004). Aging and memory control beliefs: Performance in relation to goal setting and memory self evaluation. Journals of Geronto logy: Psychological Sciences 59B(2), 56 65. West, R.L., Bagwell, D.K., & Dark Freudeman, A. (2005). Memory and goal setting: The response of older and younger adults to positive and objective feedback. P sychology and Aging, 20(2), 195 201. West, R. L., Dark Freudeman A., & Bagwell, D. K. (2007). Goals feedback conditi ons and memory: Mechanisms for memory gains in older and younger adults. Under review West, R.L., Bagwell, D.K., & Dark Freudeman, A. (2008). Self efficacy and memory aging: The impact of a memory intervention based on self efficacy. Aging, Neuropsychology, and Cognition, 15(3 ), 302 329. West, R.L., Hastings, E., & Dark Freudeman, A. (2008). Methodology: Reliability, validity, and alternate forms of the Brandeis Test of Adult Cognition by Telephone BTACT. Paper presented at the Cognitive Aging Conference, Atlanta. Wheeler, S.C., & Petty, R.E. (2001). The effects of stereotype activation on behavior: A review of possible mechanisms. Psychological Bulletin, 127, 797 826. Yoon, C., Hasher, L., Feinberg, F., & Rahhal, T.A., & Winocur, G. (2000). Cross cultural differences in memory: The role of culture based stereoty pes about aging. Psychology and Aging, 15(4), 694 704.
110 BIOGRAPHICAL SKETCH Benjamin Andr Bensadon was born on April 10, 1974, in New York City. An only has s haped him as a person and continues to guide his quest to help others. He earned his Bachelor of Arts in psychology and his Master of Education from Boston University in 1996 and 2000, respectively. He earned his Master of Science from the University of Florida in 2007, and has supplemented his graduate studies in psychology with coursework in epidemiology, social policy, and health & social behavior while an employee at the Harvard School of Public Health. Benjamin has provided individual and group p sychotherapeutic and assessment services to children and adults, in college counseling and primary healthcare settings. Benjamin has also spent substantial time traveling studying, and living in Europe, and is fluent in Italian and conversant in Spanish. Upon completion of his Ph.D. program, Benjamin will utilize his linguistic, research and clinical skills to maximize his ability to ameliorate human suffering, particularly that of older adults, as broadly as possible.