Self-efficacy, self-regulation, and complex decision-making in younger and older adults

C O M P L E X D E C IS IO N -M A K IN G IN Y O U N G E R A N D O L D E R A D U L T S by

L eslie C a ro l M cD onald-M iszczak M .A ., U niversity o f V ictoria, 1991

B.A ., U niversity o f A lb e rta 1987

A T h esis S u b m itte d in P a rtia l F u lfillm en t o f th e R e q u ire m e n ts fo r th e D e g re e of

Ph.D .

in th e D e p a r tm e n t o f Psychology

W e a c c e p t this thesis as co nform ing to th e re q u ire d sta n d a rd

D r. D .F . H u ltsc h , ^W pervisor (D e p a rtm e n t o f Psychology)

Dr. R .A . Dix0fi, D e p a rtm e n ta l M e m b e r (D e p a rtm e n t o f Psychology)

-f —* „ 1 ^ 1 ^ - ■ ... ... D r /N .L . G a la m b o s, D e p a rtm e n ta l M em b er (D e p a rtm e n t of Psychology)

D r. E. G a lla g h e r, O u trid e M e m b e r (School o f N ursing)

D r. K. H o o k e r, E x tern al E x a m in e r, O re g o n S ta te U niversity

© L E S L IE C A R O L M C D O N A L D -M IS Z C Z A K 1995 U niversity o f V ictoria

All rights reserv ed . T his thesis m ay n o t be rep ro d u c e d in w hole o r in p a rt, by p h o to co p y or o th e r m eans, w ith o u t th e p erm issio n o f th e a u th o r.

ii

Supervisor: Dr. David F. Hultsch

ABSTRACT

A computer simulation of a managerial decision making task (Wood & Bailey, 1985) was used to investigate the role that self-efficacy and self-regulatory factors play in younger and older adults' cognitive processes. Sixty-eight younger (aged 18 to 30 years) and 69 older (aged 65 to 77 years) adults were included in this study.

Participants were randomly assigned to receive either progressively positive, stable, or progressively negative social comparison feedback information as they performed the task across three blocks of trials (Bandura & Jourden, 1991 ). Mean-level results indicated that performance, decision making time, strategy use, goal setting, performance satisfaction, and self-efficacy decreased across the course of the experiment. Younger adults performed the task more efficiently, took less time to make their decisions, and reported less positive mood than older adults. Moreover, path analyses indicated that the role self-efficacy and self-regulatory factors play in complex cognitive processes differed depending on the participants' age. In earlier stages of the experiment, both younger and older adults' efficacy influenced self-regulatory factors and performance in the expected fashion. However, goal setting played a more direct role in older adults' performance, whereas the role of

performance satisfaction was more apparent in younger adults' performance. The most visible age differences occurred in the latter part of the experi'ment. Younger adults' B!ock 3 performance was primarily determined

by

their earlier task performance, whereas older adults' Block 3 performance was primarily determined by their earlier

·••.·

addition, t.he higher older adults' self-efficacy scores, the poorer their Block

3

performance. Collateral results addressed the relation between (a) task-specific

measures

of

self-efficacy and questionnaire measures of efficacy, and (b)

task-specific self-efficacy and self-reported interest to participate in future studies.

varying in similarity to the present one.

Examiners:

Dr. D.F. Hultsch, Supervisor (Department of Psychology)

Dr.

R.~~ Dixon:?.~artmental

Member (Department of Psychology)

Dr.

N.L.

Galambos, Departmental Member (Department

of

Psychology)

Dr. E.

Gallagher,

Outside Member (School of

\fursing)

- - - -

IV T A B L E O F C O N T E N T S

Chapter Page

I. Introduction ...1

II. L iterature R eview ... 6

The Self-C oncept... 7

Stereotypes...7

Social Com parison... 10

M etacognition and Aging... 12

Age and Self-Efficacy...15

Task-Specific Self-Efficacy...19

Theoretical Fram ew ork... 21

The Role o f Self-Efficacy and Self-Regulatory Factors on B ehaviour...22

Self-Efficacy... 23

A ffect... 24

M otivation... 25

Strategy U se...28

The Se'f-Regulatory Process...29

Advantages of A dopting a Process View into C ognitive/D evelopm ental R esearch 34 The Present Study... 36

The Experimental Task... 37

Prim ary Research Questions... 40

1. Age D ifferences in Self -Efficacy, Self-Regulation, and Perform ance... 40

2. Age D ifferences in the Self-Regulatory Process... 44

Collateral Q uestions... 46

1. D o G eneral Efficacy Beliefs Correlate More Strongly with Initial 'fask-Specific Self-Regulatory M easures than with M easures Obtained After Task Experience?... 46

2. Does Self-Efficacy Influence Self-Reported Behaviour?...47

III. M ethod... 49

D esign... 49

Subjects...49

M easures I. Questionnaire Battery... 51

M easures II. Com puter Display Self-Regulatory Factors... 54

M easures III. Performance M easures... 56

M easures IV. Post-Experim ental M easure... 57

Procedure... 57

IV. R esults... 61

M anagerial Experience... ...61

Prim ary A nalyses... .62

Perform ance and Self-Regulation with Task Experience...62

Path M odels... 67

vi

G eneral Beliefs and Self-Regulatory Factors... ..71

Correlations Between General Beliefs and Perform ance M easures...72

C orrelations Between General Beliefs and Self-Regulatory M easures... 74

Post-Experim ental Questionnaire... 76

V. D iscussion...79

Perform ance and Self-Regulation... ...79

Perform ance... 79

Self-Regulation... 81

The Role o f Self-Regulatory Factors in Cognitive Processes... 84

T he Use o f Self-Efficacy as a M otivator o f Perform ance... f 5 T he Role o f Previous Perform ance... ... 88

Perceptions o f the Experim ental T ask... 89

Collateral Findings... 91

General Efficacy Beliefs...91

Post-Experim ental Q uestionnaire... 93

Conclusion... ... 95

References...99

A ppendix... 112

Table 1: T able 2: T able 3: Table 4: Table 5: T able 6: Table 7: LIST OF TABLES

M ean sample characteristics as a function o f participants’ age group. Y ounger and older adults’ mean scores on perform ance and self-regulatory variables across all three blocks o f trials.

M ean initial scores for questionnaire and self-regulatory m easures as a function o f age group.

Correlations between older adults’ PIC questionnaire scale scores, decision-tim e scores for each block, and change in decision-tim e across blocks 1 and 3.

Correlations between younger adults’ questionnaire scores and self- regulatory variables,

Correlations between older adults’ questionnaire scores and self-regulatory variables.

C orrelations between Block 3 self-efficacy and self-reported interest to participate in future psychology studies.

Figure 1: Figure 2: Figure 3: Figure 4: Figure 5: Figure 6: Figure 7: Figure 8; Figure 9: Figure 10:

viii

LIST OF FIGURES

Decision m aking perform ance as a function o f age group and trial block. Decision m aking tim e as a function o f age group and trial block.

Strategy use as a function o f age group and trial block. G oal setting as a function o f age group and trial block.

Perform ance satisfaction as a function o f social com parison inform ation condition and trial block.

Self-efficacy levels as a function o f social com parison inform ation condition and trial block.

Bandura and Jo u rd en ’s (1991) path model for the younger adults’ data. B andura and Jo u rd en ’s (1991) path m odel for the older ad ults’ data. The younger a d u lts’ path m odel indicating the relationships between self- efficacy, self-regulatory factors, and perform ance on a com plex decision m aking task.

The old er ad ults’ path m odel indicating the relationships between self- efficacy, self-regulatory factors, and perform ance on a com plex decision m aking task.

A C K N O W L E D G E M E N T S

This project involved num erous individuals w ithout whose help it w ould not have been com pleted. I ow e m any thanks to m any people.

First, I w ould like to thank my m entors. I am eternally grateful to my Supervisor and m entor, Dr. D avid F. Hultsch, w ho supported my ideas and alw ays asked the right questions to strengthen them. I owe so much to his encouragem ent and advice over the y ja rs, but m ostly I am indebted to him for providing an exam ple o f a consum m ate professional -- i thank you. I would also like to thank Dr. R oger A. D ixon w ho was alw ays available when I was seeking advice on research and professional issues — your tim e w as greatly appreciated. In addition, I am deeply indebted to Dr. D ixon for the opportunities that he afforded me through our work on the Canadian Aging

Research N etw ork (CARNET), and feel that these experiences have shaped this w ork and will continue to influence my ideas.

I w ould also like to thank my com m ittee m em bers: Dr. N ancy G alam bos, for her thoughtful com m ents and her support for getting this thesis com pleted; and Dr. Elaine G allagher, for her encouragem ent and help throughout this entire project. In addition, I w ish to acknow ledge Dr. M ichael H unter, w ho devoted m uch tim e and effort to this project at earlie.' stages - - 1 appreciate your efforts.

C om pletion of this project would not have been possible w ithout the consent and support o f Dr. R obert W ood who created the com puter sim ulation o f m anagerial decision m aking used in this study. Dr. W ood went well beyond the call-of-duty by providing m e with these m aterials. I am deeply indebted to his generosity.

X T o collect my data, I required the cooperation and help o f several individuals. First, I w ould like to acknow ledge M rs. M aureen Pugh, w ho helped with the data collection. She spent m any hours w orking on this project, and I w ould not have com pleted it w ithout her help. In addition, I wish to thank the consultants at the Cornett C om puter Facility at the U niversity o f Victoria. I appreciate their

consideration o f m y data collection and all the hours they spent to m ake this project happen.

I w ould also like to acknow ledge m y colleagues at L akehead U niversity. T heir support was greatly appreciated during the final stage o f this project. In addition, I would like to thank the Natural Sciences and Engineering Research Council for its generous support.

Finally, I thank my fam ily and friends. I give special acknow ledgem ent to my husband, Tom M iszczak, who has alw ays supported me and m y work. W ords really cannot describe how grateful 1 am to have you in my life; I am blessed. T o my parents, M artha and Hugh M cD onald, thank you for believing in m e and always encouraging me to do my best -- you are my rock. To my friends, W endy, Roland, H eather, Phil & D iane, W ally, and N orV ic Toastm asters — I thank you for alw ays providing encouragem ent and support, I have learned so m uch from know ing all o f you,

INTRODUCTION

Researcher, .terested in behavioural developm ent have begun to focus on psychosocial factors that m ay play an im portant roie in successful aging in later life. A dults’ beliefs influence their choices to become involved in certain situations which m ay, in turn, shape their overall developm ent. In particular, it has been suggested that positive self-efficacy is necessary for im proving perform ance (i.e., cognitive) levels and m oving towards optim al developm ent (M aeiel, Heckhausen, & Baltes, ’.992). A ccording to B andura (1986), "self-efficacy is defined as people’s judgm ents o f their capabilities to organi7_e and execute courses of action required to attain designated types o f perform ances. It is not concerned with the skills one has but with judgm ents o f what one can do with w hatever skills one possesses" (p. 391). It is im portant to note that the self-efficacy concept is not a passive judgm ent about o n e ’s capability; rather, it is a call to action (Bandura, 1986). Thus, individuals w ho have the same potential fo r developm ent m ay behave and develop quite differently because one person has very positive beliefs and chooses to act in a way that prom otes optim al developm ent, whereas another person has very negative beliefs and acts in a way that does not (M aciel et al., 1992).

G row ing older in our society is often associated with negative im ages (e.g., Richm an, 1977) and stereotypes (H endrick, G ekoski, & Knox, 1991; H endrick, Knox, G ekoski, & Dyne, 1988; Rubin & Brown, 1975; Schwab & H enem an, 1978), This negative social context m ay shape older adults’ beliefs about them selves. In particular, it has been suggested that beliefs about cognitive declines m ay be

2

internalized by individuals as they grow older (e.g., Langer, 1989), o r used as

heuristics to estim ate change (e.g., Ross, 1989). Such beliefs m ay influence not only how older adults view their abilities, but also affect their cognitive abilities directly.

Som e researchers, interested in the role that older adults’ beliefs play in cognitive perform ance, have developed questionnaire instrum ents to m easure self-efficacy within specific cognitive dom ains such as m em ory (Berry, W est, & D ennehey, 1989; Dixon, H ultsch, H ertzog, 1988) and intelligence (Lachm an, Baltes, N esselroade, & W illis, 1982). G enerally, these researchers have found that older adults have m ore negative self-reported efficacy beliefs than young*, adults. For exam ple, older adults report that they feel less capable in a variety o f everyday m em ory situations than younger adults (e.g., Hultsch, H ertzog, & Dixon, 1987). A lthough studies have shown that the questionnaire m easures are quite reliable and valid (e.g., B erry et al., 1989; H ertzog, Hultsch, & Dixon, 1989; Hultsch, H ertzog, Dixon, & D avidson, 1988; l achman et al., 1982), the relation betw een a d u lts’ self-reported beliefs and their cognitive

perform ance is quite m odest (see G ilew ski & Zelinski, 1986; H errm ann, 1982). The m odest relation betw een questionnaire m easures o f self-efficacy beliefs and cognitive perform ance has prom pted som e researchers to m ake the distinction between general self-efficacy beliefs and task specific self-efficacy beliefs. H ertzog, Dixon, and Hultsch (1990a) suggested that questionnaires m easure self-efficacy in a general m anner by sam pling a variety o f cognitive situations (albeit within a particular domain o f cognitive functioning), w hereas cognitive perform ance is often m easured quite specifically (e.g., word-list recall). This distinction suggests that different levels o f

self-efficacy (i.e., general vs. specific) m ay operate at different tim es during cognitive processing. It follow s then, that the .elation between general self-efficacy and

cognitive perform ance m ay be m uch w eaker than the relation betw een task specific self-efficacy and perform ance.

Som e researchers have used participants’ global predictions to exam ine age differences in the relation between task-specific self-efficacy and cognitive

perform ance. A global prediction m ight consist of participants predicting the num ber o f words they wiii correctly recall from a 30 word-list. A lthough the findings with regard to age differences in prediction accuracy are m ixed (e.g. Lovelace & M arsh,

1985), older adults typically predict that they will perform at a low er level than younger adults (e.g., H ertzog, Dixon, & H ultsch, 1990b; H ertzog, Saylor, Fleece, & D ixon, 1994; M cD onald-M iszczak, H unter, & H ultsch, 1994).

B andura (1989) has argued that self-efficacy scales that vary the level o f task difficulty, provide a m ore sensitive m easure to variations in perceived self-efficacy. B andura (1986) suggested that if o n e ’s interest is in the relation between self-efficacy and cognitive perform ance, then self-efficacy should be m easured with respect to particular levels o f perform ance on a task in conjunction with other self-regulatory m easures (i.e., goal setting, strategy use, affect). Typically, B andura and his colleagues have m easured self-efficacy as the num ber o f tim es a person agreed that s/he could perform at different levels on the sam e task (B andura, 1986; Bandura & C ervone 1983, 1986). For exam ple, on a m em ory task involving 30 words,

4

rem em ber) given the choices o f 5, 10, 15, 20, 2.5, and 30 w ords, respectively.

Follow ing each increm ent in perform ance, subjects are also asked to indicate how confident they feel about each response on a scale ranging from 100 percent (very sure) to 10 percent (very unsure). In this way, B andura suggested that both Self- Efficacy Level (the total num ber of "Yes" responses) and Self-Efficacy Strength (the total level o f confidence) can be m easured. In addition, m ore recent studies have exam ined the relation between self-efficacy and other self-regulatory factors (i.e., goal setting and strategy use) in younger ad ults’ cognitive perform ance. Self-efficacy m easures were found to work in conjunction with other aspects o f self-regulatory behaviour to influence overall perform ance (Bandura & Jourden, 1991; B andura & W ood, 1989; W ood & Bandura, 1989).

W ithin cognitive aging research, Berry et al., (1989) developed a reliable and valid questionnaire (the M entor, Self-Efficacy Questionnaire; M SEQ ) to m easure adult age differences in m em ory self-efficacy based on B andura’s m ethod for m easuring self- efficacy level and self-efficacy strength. For exam ple, one study included both laboratory and everyday m easures o f m em ory and show ed that self-efficacy scores were significantly correlated with everyday m em ory m easures but not with laboratory m easures (W est, Berry, & Pow lishta, 1986). How ever, although the M SEQ m easures self-efficacy for different levels o f perform ance using a questionnaire form at, it still treats self-efficacy as a passive (i.e., isolated) judgm ent rather than as an elem ent o f a process (e.g., a call to use strategies and m otivation) as B andura intended it (Bandura, 1986).

A ccording to B andura’s (1986) perspective, self-efficacy is an integral part o f cognitive processes. Thus, in order to exam ine age differences in cognitive abilities, researchers have begun to focus on the role that self-efficacy m ay play in such differences. H ow ever, the relation betw een self-efficacy and other self-regulatory factors has been largely overlooked. Perhaps by refram ing the relation between self- efficacy and cognitive perform ance as an active one involving other self-regulatory factors, researchers m ight better understand the social com ponents of cognitive processing in younger and older adults.

The purpose o f the present investigation was to exam ine the role that self- regulatory factors play in younger and older adults’ decision-m aking performance. The self-regulatory process outlined by Bandura and his colleagues (e.g., Bandura & Jourden, 1991; B andura & W ood, 1989; W ood & Bandura, 1989) was used to study potential age differences in self-regulation and the relation o f such age differences to perform ance of a com plex task. Specifically, the present study included m easures of self-efficacy, goal setting, m ood, strategy use, perform ance satisfaction, and decision­ m aking perform ance. In order to obtain these m easures, subjects perform ed a complex decision-m aking task over m ultiple trials and answ ered questions regarding their self- regulatory behaviours at regular intervals.

6

Chapter Two

LITERATURE REVIEW

Beliefs about the self are dynam ically interconnected with each other and with behavioural experiences (M arkus & W urf, 1987). As individuals age, their self­ referent thoughts are shaped by the developm ental process and m ay be based on past, present, o r future situations. Self-efficacy is one m easure o f self-referent thought, believed to be related not only to behaviours that one perform s (e.g., Bandura, 1986), but also to other types o f self-referent thought such as o n e ’s self-concept (M arkus & W urf, 1987).

W hy are older adults’ beliefs about their cognitive capabilities hypothesized to be different from younger adults’ beliefs? In order to understand the potential role o f self-efficacy in cognitive developm ent across the lifespan, it is im portant to place such beliefs within the context o f other (i.e., m ore general) self-referent beliefs. In

particular, the early sections o f this review will focus on self-referent beliefs that m ay be m ore readily understood in terms of possible age-related differences (i.e., age- related stereotypes), than task-specific beliefs (i.e., self-efficacy for perform ing a particular cognitive task). this way, I hope to dem onstrate that such general beliefs m ake it necessary to examine task specific beliefs, especially when the purpose o f this line o f research is to exam ine the relation between beliefs and behaviour.

I will first review literature addressing (a) self-concept and stereotypes, and how such beliefs may, in turn, shape older adults’ behaviour (i.e., social com parison), (b) age differences in general self-efficacy beliefs within the general dom ain o f cognition

(i.e., m etacognitive questionnaire assessm ents), and (c) age differences in task specific self-efficacy within the context o f particular cognitive tasks (i.e., global predictions on a particular m em ory task). Next, I will review the social psychological perspective on self-efficacy and other self-regulatory variables as an active process in behaviour. Finally, I will discuss the potential usefulness o f the social psychological view in studies o f cognitive ability in later life.

The Self-Concept

D ifferent types o f self-referent thought may represent different levels o f generality involving self-referent beliefs and knowledge. The more dom ains o f functioning a self-referent concept encom passes, the m ore general it is (M arkus & W urf, 1987). It has been suggested that everyone has a concept o f the self w hich influences their behaviour. A lthough researchers have disagreed about the structural content o f the self-concept, m ost agree that the self-concept is a m ultifaceted and dynam ic structure that is system atically implicated in all aspects o f social inform ation processing (M arkus & W urf, 1987; M ueller, Johnson, D andoy, & K eller, 1992). Given that the w ay in w hich we view ourselves (i.e., ou r self-concept) influences our self-referent thoughts and behaviours, one issue pertinent to this discussion is w hether aspects o f the self-concept (i.e., self-im age) can becom e system atically distorted or changed over the life span.

Stereotypes

A ge-related stereotypes have been the focus o f m uch research (K ite & Johnson, 1988; K ogan, 1979; Lutsky, 1981; W alsh & Connor, 1979), It has been suggested that

8 socially held perceptions m ay interact with age-related events to foster a different self- image in elderly individuals than in younger adults (Langer, 1989; M ueller et al., 1992). R esearch on attitudes tow ard the elderly varies quite extensively in its

findings with respect to whether negative or positive attitudes are held by m em bers of society (K ogan, 1979; Lutsky, 1981). Obviously, when studying an entire age group, opinions and attitudes will differ depending on what aspect o f the group is highlighted. For exam ple, some research has shown that there are m ultiple stereotypes o f older individuals that include both positive and negative subcategories (H um m ert, 1990; H um m ert, G arstka, T.A., Shaner, J.L., & Strahm, S, 1994; Schm idt & Boland, 1986). In one study, H um m ert (1990) found that there is little overlap o f stereotypes between younger and older age groups. H um m ert’s (1990) findings indicated that although the elderly are not necessarily evaluated in a negative light, som e elderly people are (e.g., perceptions o f severe mental and physical im pairm ent increased with target age), and the probability o f negative evaluation increases with the targ et’s age (e.g., age 75 and over). Research by Heckhausen, D ixon, and Baltes (1989) also found both positive and negative characteristics assigned to individuals o f various ages. In addition, they discovered there was considerable agreem ent between younger and older subjects’ ratings (H eckhausen et al., 1989).

A lthough research on age-related stereotypes has generally shown m ixed results, one line o f research that has routinely found negative views tow ard the elderly focuses on attitudes toward intellectual and cognitive com petence with age. In a m eta-analysis o f research on attitudes toward the elderly, Kite and Johnson (1988) found that older

people are consistently rated m ore negatively than younger people in term s o f their com petence. Specific studies have shown that the elderly are perceived to be less com petent and have low er potential as em ployees (Rosen & Jerdee, 1976), are view ed as less intellectually com petent than both younger and m iddle-aged adults (Rubin & Brown, 1975), and have received negative ratings with respect to m em ory ability (H endrick, Knox, G ekow ski, & D yne, 1988; Ryan, 1992) and psychom otor speed variables (H endrick et al., 1988). These findings suggest that intellectual and cognitive abilities are perceived to decline in later adulthood. Thus, there m ay be a negative societal context in which adults develop cognitively.

A lthough research on younger and m iddle-aged adults’ attitudes tow ard the elderly has indicated som e negative perceptions, it is a separate issue w hether older

individuals view them selves in the same light. Perhaps belonging to a specific age group influences o n e ’s view. G roup m em bership may influence perceptions in a num ber o f ways. F o r exam ple, older individuals may view their age group m ore positively than younger m em bers o f society because m em bership creates a personal stake w hich may lead to positive perceptions o f o n e’s self. A lternatively, older adults m ay endorse negative perceptions o f their group and of them selves. T he elderly may also m ake a distinction betw een them selves and other older adults w hich m ay lead to m ore positive (i.e., above the m ean) views o f the self in com parison to their peers. Thus, a separate issue for exam ination is older adults’ self-referent beliefs about their ow n cognitive behaviour.

1 0

Social Comparison

One im portant w ay in which perceptions o f an age group influence individuals is through social com parison. Tw o o f the m ost basic propositions o f social com parison theory are that (a) individuals do indeed com pare them selves to o th er people and (b) to do this, they com pare them selves with people w ho are sim ilar to them selves

(Festinger, 1954). It is useful to know how we perform certain activities. K now ing if one is good or bad at some activity can aid o n e’s choice o f behaviours so that one’s tim e and effort are used m ost efficiently. Such choices ultim ately influence

individuals’ developm ent across the lifespan (M aciel et al., 1992).

Because physical and objective m easures are rarely available, social com parison is often used to gauge how good individuals are at some activity. C om parison often gives us a standard by which we are able to judge ability (G oethals & D arley, 1987). F or exam ple, a person m ight feel good because he or she is able to com plete a cross word puzzle in ten m inutes. However, the act is given m ore m eaning when one learns that very few people can com plete sim ilar puzzles within that tim e frame.

Another im portant aspect o f social com parison theory is that individuals prefer to com pare them selves with other people who are sim ilar to them selves (G oethals & Darley, 1987). Sim ilarity with the com parison group gives the com parison m eaning. If the com parison group is very sim ilar to the individual, then the feedback

inform ation readily applies to that individual’s perform ance. This aspect o f social com parison theory m ight explain why age-related attitudes and beliefs are potentially im portant factors in older a d u lts’ self-referent beliefs.

One im portant am endm ent to Festinger’s original theory o f social com parison, suggested by G oethals and D arley (1987), is that not only do people have a need to evaluate their abilities and their perform ance, but such a need is rooted in their need for self-esteem (i.e., self-w orth or self-im age, com m only noted as the evaluative and affective dim ension o f the self-concept). For exam ple, person A m ight seek out com parative inform ation regarding his or her mem ory rH lities, If person A is an elderly individual, part o f the reason for com paring o n e ’s self to others is to gain know ledge about o n e ’s abilities. A nother reason for com parison is validation that one is doing pretty well for o n e ’s age and thereby gain positive inform ation to support self-esteem .

O n e ’s choice o f com parison group will obviously influence the inform ation one receives with respect to their standing in the group (i.e., above or below the group mean) (see Baltes & Baltes, 1993). In particular, changing o n e ’s com parison group can be an im portant m echanism in m aintaining o n e’s self-esteem across the lifespan. In support o f this suggestion, some researchers have found that older individuals do not differ from younger adults on m any self-reports related to the self (Baltes & Baltes, 1986; Butt & Beiser, 1987; Lachm an, 1986b), Thus, the m aintenance o f self­ esteem across the lifespan cannot be overlooked. How ever, one m ight expect that within som e dom ains o f functioning, older adults might not exhibit such m aintenance.

W ith respect to cognition, it has been suggested that because there are so many generalized perceptions regarding changes in older adults’ cognitive abilities with age, it is very difficult for then) to accurately assess changes in their own abilities (Ross,

12

1989; M cFarland, Ross, & G iltrow , 1992). Such difficulty m ay not only stem from generalized societal stereotypes, but also from internalizing age-related stereotypes thereby view ing personal cognitive abilities in an increasingly negative light (Langer, 1989). Thus, it m ay be very difficult for older individuals to use social com parison inform ation in an optim al fashion. M oreover, because m any com m only held beliefs about cognitive aging are negative, these beliefs m ay work to dim inish self-esteem , self-efficacy, and even cognitive abilities.

Metacognition and Aiiiniz

M etacognitive research has addressed the relation betw een a d u lts’ perceptions o f their abilities and their actual abilities (e.g., D ixon & H ertzog, 1988; D ixon, H ertzog, & H ultsch, 1986; Gilev/ski, Zelinski, & Schaie, 1990; H ultsch, H U tzog, & Dixon, 1987; L achm an, 1986a). Obviously, individuals’ beliefs about their cognitive abilities will depend on the cognitive dom ain o f interest. For exam ple, if asked about o n e ’s perceived cognitive abilities within the dom ain o f practical advice giving versus the dom ain o f m em ory, an individual m ight give very different responses.

Several questionnaires have been developed to exam ine ad ults’ perceptions o f their m em ory (e.g., Dixon & Hultsch, 1983; D ixon et al., 1988; G ilew ski, Zelinski, & Schaie, 1990) and intellectual abilities (Lachm an et al., 1982). T ypically, these questionnaires assess different dim ensions o f m etacognition within a particular dom ain o f cognitive functioning (e.g., m em ory or intelligence). For exam ple, the M etam em ory In A dulthood (M IA) questionnaire has seven different dim ensions tapping perceived m em ory (Capacity, Change, Strategy Use, Anxiety, A chievem ent m otivation, Task

know ledge, and Locus o f Control) (see D ixon et al., 1988). In addition, the

Personality In C ontext (PIC) questionnaire has six different dim ensions o f intellectual functioning (Internal Locus o f Control, External Locus o f Control, Pow erful Other, A chievem ent, A nxiety, and M orale) (Lachm an et al., 1982). Although such questionnaires address a particular cognitive dom ain, these m easures are still

considered to be relatively general because the questions cover several different types o f everyday tasks (e.g., perceived ability to rem em ber telephone num bers, nam es, and im portant dates) (Hertzog et al., 1988).

T ypically, cross-sectional studies using such questionnaire instrum ents have shown that older adults have more negative perceptions o f their cognitive abilities than do younger adults (G ilew ski, et al., 1990; Hultsch et al., 1987). In addition, Lachm an (1986b) show ed that dom ain specific m easures of cognitive (i.e., intelligence) are more likely to result in age differences in self-reported ability than m easures that encom pass several dom ains o f cognition or do not specify a particular dom ain (i.e., L evenson’s (1974) Locus o f Control Scale). Thus, specifying a particular cognitive dom ain in questionnaire assessm ents should lead to a higher relation betw een self-reported abilities and actual cognitive abilities.

O riginally, m any o f these questionnaires were indeed designed to tap perceptions o f cognition that m ight prove to be valid indicators o f cognitive perform ance (Herrm ann,

1982). F o r exam ple, an individual’s self-rated ability to rem em ber phone num bers should reflect their actual m em ory ability. G enerally, it was expected that the correlation betw een questionnaire responses and m easures o f ability w ould be very

14

high. Indeed som e research has shown that certain m etacognitive dim ensions are m ore likely to correlate with ability than other dim ensions (see D ixon & H ertzog, 1988). In particular, some research has found significant correlations betw een the know ledge dim ension o f the M IA (e.g., Strategy U se and Task K now ledge) and m easures o f cognitive ability (Dixon et al., 1986). H ow ever, correlations betw een self-reported abilities (i.e., scale relating to the efficacy dim ension on the M IA ) and actual abilities are typically m odest (Hertzog et al., 1990a). Various reasons for such m odest

correlations have been suggested (Rabbitt & Absen, 1990), including the validity of the questionnaires them selves (Herrm ann, 1982).

Dixon and H ertzog (1988) and H ertzog et al. (1990a) suggest that such

questionnaires are reliable and valid m easures o f self-reported ability, but that such self-reports reflect beliefs about ability that m ay be quite inaccurate. Psychom etric research on m etacognitive questionnaires has indeed resulted in evidence for a belief laden factor (e.g., M emory Self-Efficacy in the M IA and M FQ and Intellectual Self- Efficacy in the PIC). Consistent with the view that the m odest relation betw een self- reports and ability stem s from inaccurate beliefs, H ertzog and his colleagues (H ertzog et. al., 1989; Hultsch et al., 1988) report (a) high internal consistency reliability estim ates (C ronbach’s alpha) from ^v e independent samples for the MIA

questionnaire, (b) good discrim inant validity between the self-efficacy factor on the M IA and related concepts (i,e„ subscales in the Jackson Personality Inventory), and (c) good convergent validity o f the mem ory self-cfficacy factor in both the M IA and M FQ instrum ents. In addition, the PIC questionnaire has also shown psychom etric evidence

for (a) gooo reliability and validity in both younger and older sam ples (Lachm an, 1978) and factor analytic evidence fo r an Intellectual Self-Efficacy factor (Lachm an, 1983). In particular, the Internal Locus o f Control Scale and the A nxiety Scale from the PIC were found to load highly and significantly on the Intellectual Self-Efficacy factor (Lachm an, 1983). Thus, there appears to be strong evidence, at least for certain m etacognitive m easures, that self-efficacy is an im portant aspect o f ad u lts’ perceptions o f their cognitive abilities. Several researchers have suggested that self-reported abilities are influenced by perceptions o f personal efficacy (Berry et al., 1989; Cavanaugh & G reen, 1990; Dixon & H ertzog, 1988; Hultsch et al., 1988). Age and Self-Efficacy

It has been suggested that adults o f all ages form their self-efficacy beliefs in the same m anner (e.g., direct and vicarious experiences) (Bandura, 1986; Sehulster, 1981a, 1981b, 1982), but the types o f experiences or information attended to during such experiences m ight be very different for younger and older adults. F o r exam ple, it is a com m on belief that m em ory declines with age (e.g., Ryan, 1992). If both younger and older adults m isplace their car keys, the experience might be interpreted quite

differently. The younger person m ight not even think o f the event as a m em ory slip, whereas the older person m ay be more likely to attend to it and m ay even interpret it as evidence o f a m em ory problem since memory is often a grow ing concern with age (see G ilew ski & Zelinski, 1986 & H errm ann, 1982 for review s), T he reason is

probably quite sim ple; m em ory problem s are com m only associated with advancing age (e.g., Ryan, 1992). Therefore, a m em ory slip would not typically be interpreted as a

16

sign o f a problem to a younger person who is not expected to have m em ory problem s, but m ight be interpreted as such by an older person (Ham m er, 1994). Such an

interpretation becom es problem atic, however, when an older individual is not

cognitively im paired and such beliefs negatively influence their perform ance (e.g., they avoid o r expend less effort in mem ory situations).

Cross-sectional studies have indicated that one o f the m ajor challenges facing researchers who are interested in the interface between social and cognitive factors is dissociating the age-related beliefs from the age group differences on task variables. F or exam ple, older adults m ay simply find certain laboratory tasks quite difficult because they are placed in an unfam iliar cognitive situation. In short, older ad ults’ cognitive perform ance on laboratory tasks m ay suffer due to their lack o f fam iliarity in the testing situation. Specifically, task factors such as task fam iliarity (lab vs everyday tasks) and training have been exam ined (Dittm an-Kohli, Lachm an, Kliegl, & Baltes, 1991; Rebok & Balcerak, 1989; W est, et al., 1986). G enerally, researchers have found that there are age differences in self-efficacy levels in favour o f younger adults, and that task factors do correlate with self-efficacy levels (e.g., higher task fam iliarity correlates with higher self-efficacy levels). For exam ple, W est et al., (1986) found that younger adults had higher self-efficacy scores than older adults, people w ho did not com plain about their m em ory ability had higher self-efficacy scores than people who did com plain, and individuals who perform ed fam iliar tasks rated their self- efficacy higher than those who perform ed unfam iliar tasks.

W ith regard to task training, Rebok and Balcerak (1989) found that older adults who received m em ory training had higher perform ance expectations than older adults in the other conditions, but these results were not statistically significant. H ow ever, another study conducted by D ittm ann-K ohli et al. (1989) found that ratings o f self- efficacy im proved after subjects received training on cognitive tasks, even though perform ance on the tasks did not im prove any m ore than the perform ance o f subjects who w ere sim ply re-tested. M oreover, D ittm ann-K ohli et al., (1989) found enhanced self-efficacy on tasks for which subjects received training, but not on other tasks. Thus, cross-sectional studies have shown that self-efficacy differs between younger and older adults, and also between different aspects of the task.

An im portant focus of developm ental research is the relationship between sclf- efficacy and perform ance change across time. A study by Cornelius and Caspi (1986) exam ined the relationship between perceptions o f intelligence and intellectual

perform ance in an elaborate cross-sectional study spanning m iddle-aged and older adults. They found that internal and external control beliefs were quite stable during m iddle-age, but participants in their 60s or 70s perceived iess internal control and m ore external control than m iddle-aged participants. M oreover, adults w ho scored high on external control perceived a decline in their intellectual ability, experienced more anxiety, generally endorsed an external locus o f control, and had low er fluid and crystallized intelligence scores. Changes in o n e ’s perceived intellectual efficacy and concern with intellectual behaviour m ay be related to actual changes in intellectual ability and changes in generalized perceptions o f control.

18

Longitudinal studies have been conducted to exam ine the link betw een older adults’ efficacy beliefs and cognitive perform ance (G rover & H ertzog, 1991; Lachm an, 1983; Lachm an & Leff, 1989). Specifically, these studies focused on the relationship between self-reported personality variables (i.e., Intellectual Self-Efficacy) and

m easures o f intellectual ability (i.e., crystallized and fluid intelligence) over

longitudinal periods. G enerally, personality variables (e.g., Intellectual Self-Efficacy) did not affect changes in m em ory perform ance (e.g., m em ory span) even though there were significant changes in cognitive ability across time. Self-efficacy was not associated with changes in intellectual perform ance across the longitudinal period, but intellectual perform ance did predict changes in self-efficacy. Thus, when generalized m easures are used, the direction o f influence is from perform ance to self-efficacy perceptions, but not the other way around.

One recent study conducted by M cD onald-M iszczak, H ertzog, and Hultsch (in press) did not support these longitudinal findings. Using a two- and a six-year longitudinal period with different samples, M cD onald-M iszczak et al. (in press) found that adults’ perceptions o f changes in their m em ory abilities w ere significantly

correlated with their m em ory self-efficacy beliefs, but not correlated with actual changes in m em ory abilities. This study m ore strongly supported the notion that perceptions o f changes in m em ory abilities stem from an age-related im plicit theory o f age-related changes in cognitive abilities (i.e., age-related stereotypes about cognitive declines), rather than a realistic assessm ent o f memory changes across time.

O ne problem that arises when exam ining the relation betw een generalized assessm ents o f m em ory/intellectual self-efficacy and cognitive abilities, is that such beliefs are m easured at a general level on the questionnaires (e.g., several questions assessing perceptions o f different everyday m em ory/intelligence situations), whereas perform ance is often m easured specifically (e.g., m em ory for particular lists o f nouns). The relation betw een perceived ability and actual ability m ay be very complex

depending on the specific ability that is m easured (Dixon & H ertzog, 1988). Because o f potential task differences, task specific m easures of self-efficacy beliefs and

cognitive abilities are reviewed next. Task Specific Self-Efficacv

Several researchers have used perform ance predictions to m easure m etam em ory (Bruce, Coyne, Botw inick, 1982; C onnor & H ertzog, 1993; H ertzog et al., 1990b; H ertzog et al., 1994; Lachm an, Lachm an, & Thronesbury, 1979; Lovelace & M arsh,

1985; M cD onald-M iszczak et al., 1994; M urphy, Sanders, G abriesheski, & Schmitt, 1981; Perlm utter, 1987; Rabinowitz, Ackerm an, Craik, & H inchley, 1982). Predictions are typically either global or item -by-item in nature. A global prediction is a general task prediction w hich requires the individual to estim ate how he or she will perform on an entire m em ory task. For exam ple, study participants m ight be asked to predict how m any words they will correctly recall on a 30-w ord list. Item -by-item predictions require the individual to estim ate the likelihood that a specific test item will or will not be recalled. F o r instance, an individual m ight be presented with m em ory test item s one at a tim e and asked to rate the likelihood o f rem em bering each item on a

5-20

point L ikert scale ranging from 1 (will certainly not rem em ber) to 5 (will certainly rem em ber). It has been suggested that these tw o different prediction m easures tap different aspects o f m etacognition (Nelson & Narens, 1990). G lobal predictions have been hypothesized to tap task specific self-efficacy because individuals are required to estim ate their personal perform ance on a task prior to task experience (H ertzog ct al.,

1990b). A lternatively, item -by-item predictions are generally regarded as a judgem ent o f learning (i.e., the degree to which an item has been learned or encoded) because individuals have had a chance to study the particular item p rio r to m aking a judgm ent (Dunlosky & Nelson, 1992; 1994). Because it has been suggested that global

predictions are a m easure of task specific self-efficacy (H ertzog et al., 1990b), research involving global predictions will be discussed in m ore detail below.

The relation betw een global predictions and perform ance is typically inaccurate (e.g., Lovelace & M arch, 1985; R abinow itz et al., 1982). D ue to such inaccuracy, H ertzog et al. (1990b) suggested that global predictions m ight represent task-specific self-efficacy. If global predictions are belief (i.e., self-efficacy) laden like

m etacognitive questionnaires, the relation between global predictions and perform ance will not necessarily be accurate. One would expect, however, that predictions w ould be significantly correlated with general self-efficacy beliefs (e.g., the C apacity subscale from the M IA questionnaire). H ertzog et al. (1990b) suggested that the prediction process involves (a) an estim ation o f o n e ’s general mem ory self-efficacy, (b) an estim ation o f the difficulty o f the task, and (c) m apping o n e ’s self-efficacy onto the task distribution (H ertzog et al., 1990b).

Support for this suggestion comes from studies that have incorporated both a questionnaire m easure o f general m em ory self-efficacy (e.g., the C apacity scale from the M IA ), a task-specific m easure o f m em ory self-efficacy (i.e., global perform ance prediction), and m ultiple m em ory test trials. Using this paradigm , researchers have found that initial global predictions are m ore highly correlated with questionnaire m easures o f general m em ory self-efficacy than with mem ory perform ance (Hertzog et al., 1990b, H ertzog, et al., 1994; M cDonald-M iszczak et al., 1994). How ever,

subsequent global predictions (i.e., on the second and third trials) are increasingly correlated with task perform ance and decreasingly correlated with general mem ory self-efficacy (H ertzog et al., 1990b, H ertzog, et al., 1994; M cD onald-M iszczak, et al., 1994). Thus, there is evidence to suggest that prior to direct task experience,

individuals’ global predictions reflect general mem ory self-efficacy, w hereas follow ing experience, reflect task-specific self-efficacy.

Theoretical Framework

Q uestionnaire and prediction studies have provided valuable inform ation regarding the m ultidim ensionality o f m etacognition and age differences in m etacognitive

functioning. M etam em ory and m etacognitive research has already addressed som e aspects (i.e., task factors, task training/experience, implicit theories o f m em ory change) concerning the interplay betw een social and cognitive factors in age-related differences in and changes in cognitive perform ance. Cavanaugh has outlined such aspects in a fram ew ork pertaining to m em ory perform ance (see Cavanaugh, 1989). This model includes general concepts (i.e., biological factors, personality, experience, social

22

context), general cognitive constructs (i.e., executive abilities and processing, on-line awareness), elem ents o f the task itself (i.e., task factors, task evaluation), self- regulatory constructs (i.e., beliefs, efficacy judgm ents, outcom e expectations, effort, strategy selection, perform ance evaluation), and perform ance. Prim arily, the m odel encourages researchers to think about the connection between social and cognitive factors as an active process. H ow ever, because the m odel is a theoretical

representation of a com plex process, it does not overtly suggest a m ethod for em pirically exam ining particular subprocesses or subm odels. O ther areas o f psychological research have exam ined the connection betw een cognitive and social factors related to the present discussion. Although such research has not highlighted developm ental issues per se, the m ethods m ight be used with younger and older adults to address age-related research questions.

The Role of Self-Efficacy and Self-Regulatory Factors on Behaviour

B andura (1986) suggested that the role o f self-efficacy in perform ance be exam ined using a inicro-analytic technique. This technique involves taking several different m easures o f self-efficacy and self-regulatory factors when observing perform ance on a specific task. The m easures o f self-efficacy and self-regulatory behaviour are assum ed to tap different processes involved in cognition (i.e., m otivation and affect). Thus, this m ethod allows for proxim al exam ination o f the factors hypothesized to play an im portant role in behaviour.

Self-Efficacy

In social psychological studies o f self-efficacy, the construct has typically been m easured in two different ways. First, subjects are asked to provide their level o f self- efficacy. The exact form the prediction takes will depend on the task, but generally it is an estim ate o f how an individual will perform . F or exam ple, subjects will be given several different perform ance levels (e.g., 2 out o f 10, 4 out o f 10, etc.) and will be asked to respond "yes" or "no" as to w hether they will be able to perform at each level. T he num ber o f "yes" responses is used as the m easure o f Self-Efficacy Level (SEL). A person w ho feels highly efficacious about a task will have a higher SEL than a person w ho does not feel efficacious about a task. Second, subjects are typically asked to provide a confidence rating o f their perform ance at each level. For exam ple, subjects m ight be highly confident (100%) about their perform ance at the low est levels (e.g., 2 & 4), but less confident about their perform ance at higher levels, All confidence estim ates are summed and the overall m ean rating is used as the m easures o f Self-Efficacy Strength (SEST). These two m easures usually com prise an individual’s self-efficacy m easures in social psychology studies conducted with younger adults.

Several studies have supported the suggestion that task specific self-efficacy does influence perform ance (e.g., Bandura & C ervone, 1983; W einberg, G ould, & Jackson, 1979). T he level o f self-efficacy (high vs low) is usually m anipulated by the

experim enter (e.g., by com petition with another individual or by false perform ance feedback). G enerally, high self-efficacy leads to better perform ance than low self­

24

efficacy (Bandura & Cervone, 1983; W einberg, et al., 1979). O ther im portant factors, however, may play a very im portant role in the effect o f self-efficacy on perform ance. Self-regulatory factors regulate the relationship between self-efficacy and perform ance on a specific task. Bandura (1986; 1989) has specifically suggested (a) affect, (b) goal setting, (c) m otivation/effort/task persistence, and (d) strategy use as im portant self- regulatory factors. These factors are discussed below.

Affect

According to B andura (1989) individuals’ em otional states interact with their feelings o f efficacy. Em otional reactions can alter the course o f thinking and thereby influence perform ance (Bandura, 1989). Perform ance anxiety is often m entioned in experim ental studies, and researchers often endeavour to put their subjects at ease when they arrive at the laboratory since such anxiety m ay influence their results (e.g., lower perform ance). Thoughts about o n e ’s inefficacy m ay indeed im pair perform ance, and such im pairm ent may, in turn, encourage m ore negative thinking. Thoughts o f inefficacy are often accom panied by subjective distress and autonom ic arousal (Bandura, Reese, & A ' ams, 1982). In turn, subjective distress m ay be interpreted as inefficacy to perform certain behaviours which m ay lead to avoidance and/or

depression. Bandura (1989) suggests that perceived efficacy in control o f o n e ’s thoughts is a prim ary factor in the regulation o f cognitively generated arousal. Besides anxiety, B andura (1989) also specifically links self-efficacy and depression. Since negative efficacy involves negative thoughts about o n e ’s self, the individual m ay become depressed by such ruminations.

The role o f affect in self-efficacy and perform ance can be illustrated by a num ber o f studies exam ining anxiety and depression. Bandura and his colleagues (Bandura & Adam s, 1977; Bandura, A dam s, H ardy, & H ow ells, 1980) conducted studies focusing on self-efficacy and anxiety related to phobias. These studies indicated that self- efficacy was a very accurate predictor o f desensitizing behaviour (e.g., decreasing anxiety so that a participant could approach a snake) and also future behaviours not practised during the desensitizing procedure. Studies o f the relationship between self- efficacy judgm ents and depression in younger adults indicates that depression is related to low er self-efficacy (K anfer & Zeiss, 1983) and that gender differences may influence certain m anipulation o f affect (Davis and Yates, 1982). Thus, these studies indicate that there is an im portant relationship between self-efficacy, affect, and perform ance. Em otional disruptions m ay dem and individuals’ attention and energy thereby taking im portant resources away from perform ing the task at hand. Such disruptions lead to avoidance behaviour (e.g., phobias), or m ay lead to debilitated perform ance through depression or learned helplessness. Research findings suggest that self-efficacy is influenced by (K anfer & Zeiss, 1983), and exerts an influence on perform ance (Bandura & A dam s, 1977; Bandura et al., 1980).

Motivation

It has been suggested that intentions regarding future perform ance are im portant aspects o f perform ance (Locke, 1968), and are also thought to be highly related to self-efficacy (Bandura, 1986). A ccording to Bandura (1986) an intention may be defined as "the determ ination to perform certain activities or to bring about a certain

26

future state o f affairs" (p. 467). Locke (1968) developed a theory about goals which also states that goals play an im portant role in perform ance m otivation. Locke (1968) suggested that the relationship between goals and perform ance is linear as long as the perform ance goal is perceived to be attainable. Thus, the m ore challenging

individuals’ perform ance goals are, the m ore m otivated they are to achieve their goals and this higher m otivation should lead to higher perform ance attainm ents.

Goal setting is view ed as an extrem ely im portant elem ent o f perform ance in B andura’s (1986) m odel o f task perform ance because goals are thought to m easure individual’s perform ance intentions, which, in turn, are view ed as highly related to self-efficacy levels. Typically, subjects are asked to set their own perform ance goal or a goal is provided by the experim enter that is either difficult to achieve or very easy to achieve. Self-efficacy is thought to influence the goals adults set for them selves and also influence how they react to goals set by the experim enter. For exam ple, som eone with high self-efficacy is m ore likely to set higher goals and will probably react positively to a challenging goal, w hereas a person with low self-efficacy is m ore likely to set low er goals and will likely be discouraged by a high goal.

Studies on goal setting have shown that previous goal assignm ent affects subsequent goal choice and task perform ance (Latham & Baldes, 1975; Locke, Frederick, Buckner, and Bobko, 1984). F or exam ple, subjects w ho were previously assigned a difficult goal were m ore likely to choose more difficult goals for

them selves, w hereas those assigned easier goals were m ore likely to choose easier goals on subsequent trials (Locke et al., 1984). G enerally, goals do affect

perform ance, but this relationship depends on the perceived attainability (Locke et al., 1984) and com m itm ent to the goals (M cCaul, H inz, & M cCaul, 1987). M oreover, in the absence o f other task information, a goal is treated m uch like a perform ance norm reflecting the difficulty o f the task (M eyer & G ellatly, 1988). It seem s that one way in w hich assigned perform ance goals influence perform ance is by providing an

estim ate o f the difficulty o f the task and im plicit inform ation about w hat is acceptable perform ance. T hus, assigned goals m ay provide a perform ance standard for assigning one’s ow n goals and the level o f satisfaction one should feel with their performance.

W ith respect to the present discussion, goals are hypothesized to be an im portant self-regulatory elem ent determ ining perform ance (Bandura, 1986). Goals provide im portant inform ation about the task that helps adults determ ine w hat level of perform ance they should be satisfied with. Individuals with higher self-efficacy set higher perform ance goals, and are more com m itted to such goals than subjects with low er self-efficacy (Bandura, 1986; 1989). In addition, people with high self-efficacy are typically less satisfied with perform ances that fall short o f a goal than subjects with low self-efficacy because such shortcom ings contradict the perso n ’s theory about him self o r herself (Bandura 1986; 1989). Research on effort and task persistence supports these findings. Bandura (1986; 1989) suggested that individuals with higher self-efficacy will persist longer in the face o f adversity than those with lower self- efficacy since they feel m ore capable o f com pleting the task. Researchers found that subjects with high m easures o f self-efficacy persisted longer on a task (Jacobs, Prentice-Dunri, & Rogers, 1984) and showed a greater increase in effort than subjects

2 8 with low self-efficacy (Bandura & Cervone, 1986; Cervone & Peake, 1986).

M oreover, subjects w ho reported being m ost dissatisfied with their perform ance also heightened their effort (Bandura & Cervone, 1986). M otivation to perform , then, stem med from the m atch between subjects’ self-efficacy and their perform ance standards (i.e., goals and goal com m itm ent). Therefore, perform ance m otivation should be considered a m ultifaceted aspect o f self-regulation w hich requires a variety o f m easures (i.e., goals, goal com m itm ent, perform ance satisfaction and task

persistence) to understand the relation betw een self-efficacy, m otivation, and performance.

Strategy Use

People who report being highly efficacious about their abilities rem ain more focused in their analytic thinking than those w ho are significantly less efficacious (Bandura, 1989). O ne way to m easure how focused individuals are in their analytic skills is to exam ine the strategies that they use to perform a task. One w ould assum e that the more optim al the strategies used to com plete a task, the better o n e ’s

performance.

Research has shown that perceptions o f the task and o n e ’s self-efficacy

significantly affect strategic behaviour. W ood and Bandura (1989) found that the type o f analytical thinking that fostered m anagerial decisions differed depending on

subjects’ im plicit theories o f their ability. Those within the skill condition (i.e., those with the concept that skill on the experim ental task could be continually enhanced by acquiring know ledge) were system atic in their use o f analytic strategies and became

even m ore system atic with practice, whereas entity theory holders (i.e., adhering to the concept that ability is a fixed entity) became less efficient. M oreover, Bandura and W ood (1989) found that the less controllable participants’ thought the task was and the higher the pre-set perform ance standard, the less strategic were subjects’ decisions. A nother study by Locke, Frederick, Lee and Bobko (1984) exam ined the effects of self-efficacy, goals, and task strategies on task perform ance. U sing path analysis, Locke et al. (1984) found that ability is directly related to strategies used along with other self-regulatory behaviours (i.e., self-efficacy m easures and goals). Thus, perceptions o f o n e ’s ability to perform the task significantly affected subjects’ use o f analytic strategies and their overall task perform ance.

The Self-Regulatory Process

W ith all the self-regulatory factors influercing each other and, in turn, influencing cognitive perform ance, perhaps the m ost interesting and direct test o f the process suggested by B andura (1986) are studies that used path analytic techniques. In such studies, it is extrem ely important to choose a task that is com plex enough to

incorporate all the factors researchers wish to m easure (e.g., goal setting, task persistence, strategy use). A decision m aking task developed by W ood and Bailey (1985) has been used in several studies. The task is a com puter sim ulation o f a m anufacturing business which the subjects m ust run appropriately by filling different m anufacturing orders within the budgeted num ber of hours specified by the program. The com puter program m ay be changed and m anipulated to vary different aspects o f the decision-m aking process under exam ination, Subjects are asked to (a) assign

30

em ployees to appropriate jobs, (b) set appropriate goals for them selves and for the em ployees, (c) motivate the workers using appropriate incentives, (d) m onitor and make changes over multiple trials (e.g., change m anagerial strategies, goals, and incentives) in order to m eet or beat the budget, and (e) m onitor their own perform ance and the em ployees’ perform ance over m ultiple trials. Subjects’ perceptions o f their own capability to perform the task (i.e., self-efficacy level and self-efficacy strength) are also m easured at regular intervals during the sim ulation. The com plexity o f this task perm its exam ination of decision-m aking as an active process that involves several self-regulatory factors (B andura & Jourden, 1991; Bandura & W ood, 1989; Cervone, Jiw ani, & W ood, 1991; W ood & Bandura, 1989).

The ordering o f variables is crucial to the statistical integrity o f the path model. Studies o f Scif-regulatory m odels (e.g., Bandura & W ood, 1989; W ood & Bandura, 1989) usually order the variables (a) past perform ance (on previous trials) (b) self- efficacy, (c) goal-setting, (d) analytic strategies, and (e) present perform ance. This ordering o f variables assum es that previous task experience positively affects o n e ’s self-efficacy. Self-efficacy is then thought to positively affect o n e ’s task m otivation (goal setting) which, in turn, is thought to positively affect the strategies one uses to perform the task. It should be noted that studies exam ining causal relationships o f self-regulatory variables do not necessarily assum e that the effects of these variables are indirect. In addition to an indirect positive effect o f self-efficacy through goal setting on perform ance, the researcher m ay also predict a positive direct effect o f self- efficacy on performance. Using this technique, Bandura and his colleagues exam ined