Goal orientation and self-efficacy in relation to memory in adulthood

Abstract

The achievement goal framework (Dweck, 1986) has been well-established in children and college-students, but has rarely been examined empirically with older adults. The current study, including younger and older adults, examined the effects of memory self-efficacy, learning goals (focusing on skill mastery over time) and performance goals (focusing on performance outcome evaluations) on memory performance. Questionnaires measured memory self-efficacy and general orientation toward learning and performance goals; free and cued recall was assessed in a subsequent telephone interview. As expected, age was negatively related and education was positively related to memory self-efficacy, and memory self-efficacy was positively related to memory, in a structural equation model. Age was also negatively related to memory performance. Results supported the positive impact of learning goals and the negative impact of performance goals on memory self-efficacy. There was no significant direct effect of learning or performance goals on memory performance; their impact occurred via their effect on memory self-efficacy. The present study supports past research suggesting that learning goals are beneficial, and performance goals are maladaptive, for self-efficacy and learning, and validates the achievement goal framework in a sample including older adults.

In recent years, there has been an explosion of research directed at understanding how social cognitive factors, such as control beliefs and self-efficacy, may influence older adult performance on memory (e.g., West, Dark-Freudeman, & Bagwell, 2009), problem-solving (e.g., Artistico, Cervone, & Pezzuti, 2003), reading (e.g., Stine-Morrow, Miller, & Hertzog, 2006), and other cognitive measures (e.g., Jopp & Hertzog, 2007). One theoretical framework for examining the role of beliefs in cognition, which seems to have potential for understanding information processing in late adulthood, is the achievement goal framework. Although aging researchers have discussed Dweck’s (1986) achievement goal framework in relation to older learners, by describing the potential impact that implicit theories of ability could have on older adult performance (Cavanaugh, 1996; Elliott & Lachman, 1989; Hertzog, Lineweaver, & McGuire, 1999; Kanfer, 1990), to our knowledge, this framework has not been empirically validated with an older sample. That is the purpose of this research.

The Achievement Goal Framework

The achievement goal framework proposes that learners approach tasks with one of two goal orientations: learning goals or performance goals (Dweck, 1986; Elliott & Dweck, 1988; Kaplan & Maehr, 2007). Learning goals are concerned with the development of mastery and improvement of existing skills. Individuals who focus on learning accept failure as a necessary and important part of the learning process. Performance goals, in contrast, are focused on test results; from this perspective, failures are evidence that the individual lacks inherent ability. In general, learning goals are associated with positive outcomes, including higher levels of information processing (Grant & Dweck, 2003), better self-efficacy and subjective achievement (Schunk & Ertmer, 1999), more sophisticated strategy use (Elliott & Dweck, 1988), higher intrinsic motivation (Rawsthorne & Elliot, 1999), and increased pursuit of challenges (Smiley & Dweck, 1994). Performance goals are generally associated with negative outcomes, including withdrawal from challenging activities, negative affect, poorer strategy use, ineffective problem-solving, and worse academic achievement (Elliott & Dweck, 1988; Smiley & Dweck, 1994).

Researchers have suggested that performance goals may be especially detrimental when perceived ability is low (Elliott & Dweck, 1988), which is likely to be the case for older adults (e.g., Berry, 1999). When learners endorse low perceived ability, and they are concerned with results (performance goals), they are thought to adopt an “avoidant” style. These types of learners steer clear of challenging tasks in order to prevent negative judgments of ability. In contrast, those with higher perceived ability and performance goals may have an “approach” style, seeking out challenges to demonstrate their competence to others (Elliot & Harackiewicz, 1996; Elliott & Dweck, 1988; Elliot & McGregor, 2001; Kaplan & Maehr, 2007).

Whether a learner adopts an overall learning or performance goal orientation is thought to depend partly on implicit theories of ability (Button, Mathieu, & Zajac, 1996; Dweck & Leggett, 1988), known as entity and skill theories. People who hold entity theories believe that ability is fixed and cannot be modified through effort. Conversely, those who hold skill (incremental) theories believe that ability is malleable and can be improved through practice and hard work (Blackwell, Trzesniewski, & Dweck 2007). People who hold an entity theory are more likely to adopt performance goals (to prove themselves or avoid judgments of their fixed ability), whereas people who hold a skill theory are more likely to adopt learning goals (to improve skills which they perceive as changeable through effort). People with dispositional learning goals and skill theories tend to have better learning outcomes than those with dispositional performance goals or entity theories (Blackwell et al., 2007; Button et. al., 1996; Smiley & Dweck, 1994).

Much of the achievement goal literature has conceptualized goal orientation as an either/or distinction, but Button and colleagues (1996) have suggested that one may “simultaneously strive to improve one’s skills and to perform well relative to others.” In fact, a learning/performance goal combination may produce more desirable outcomes than one goal alone, at least for college students (Harackiewicz, Barron, Tauer, Carter, & Elliot, 2000; Harackiewicz, Barron, Pintrich, Elliot, & Thrash, 2002), because academic achievement is best predicted by performance-approach goals, and interest is best predicted by learning goals (Harackiewicz et al, 2002). Clearly both interest and performance are important factors in the learning process, and for this reason, holding both types of goals simultaneously may be beneficial.

Past research has suggested that goal orientation may impact learning via its influence on self-efficacy (Boon, 2007; Coutinho & Neuman, 2008; Phan, 2009a; Phillips & Gully, 1997). The potential association between dispositional goal orientation and self-efficacy can be understood by examining the theoretical sources of self-efficacy – mastery, vicarious experience, verbal persuasion, and physiological states (Bandura, 1997). Individuals who generally approach tasks through the lens of a learning goal are likely to have had more mastery experiences because they have persisted more in challenging activities, and if they had experienced failure, they would be unlikely to view it as a sign of lack of ability (Phillips & Gully, 1997). They probably perceived past experiences as a process of building or achieving mastery, and thus should have higher self-efficacy. Goal orientation is probably also linked to physiological states. Consistently approaching cognitive tasks with concern about demonstrating ability (performance goals) is likely to create anxiety. For those with performance goals, any failure is seen as an indication of low ability, leading to more failure-related anxiety than for those who embrace learning goals and see failure as a natural part of learning (Cury, Elliot, Sarrazin, DaFonseca, & Rufo, 2002). In line with this, Brodish and Devine (2009) found that performance avoidance goals positively predicted worry, which then negatively predicted performance. Individuals would then report less confidence for success due to the anxiety provoked by their performance goal orientation.

Goal Orientation Applied to Aging

The achievement goal framework has most often been studied with problem solving task using samples from elementary school through college (Hulleman, Schrager, Bodmann, & Harackiewicz, 2010; Utman, 1997). With college students, it has largely been examined in relation to course grades (Button et al., 1996; Grant & Dweck, 2003; Harackiewicz et al., 2000; Hulleman et al., 2010). Never to our knowledge has this framework been examined in the context of adults and memory performance. It is valuable to examine the achievement goal framework in relation to memory aging because the memory domain is highly salient for older adults (e.g. Cutler & Grams, 1988; Dark-Freudeman, West, & Viverito, 2006; Ryan, 1992). Adults of all ages know that age-related memory decline exists (Hertzog et al., 1999), and older adults may be especially attuned to memory issues because they are experiencing memory changes firsthand, especially in episodic memory (Backman, Small, & Wahlin, 2001; Craik, 2000; Nilsson, 2003; Zacks & Hasher, 2006). This evidence for decline in scores and changes in beliefs underscores the importance of considering self-regulatory factors that may influence memory performance, such as goal orientation or memory self-efficacy (Stine-Morrow et al., 2006).

Although few aging researchers have considered goal orientation per se, extensive aging research has suggested that older adults engage in a constellation of self-limiting beliefs that may impact memory performance, and this belief system is closely related to constructs in the achievement goal framework. Aging is associated with lower perceived memory ability and reductions in memory self-efficacy (Berry, 1999; Jopp & Hertzog, 2007). Older adults believe more strongly than younger adults that their performance is due to external uncontrollable causes (Miller & Lachman, 1999), and older adults tend to endorse fixed “entity” views of their ability (Elliott & Lachman, 1989). Attributing memory difficulties to uncontrollable causes is maladaptive because it suggests that nothing can be done to maintain or improve performance. Given these significant age changes in beliefs, it is important to take an integrative approach to beliefs and performance when considering age declines.

The achievement goal framework allows for such an integrative approach because it links goal orientation with the self-limiting beliefs often held by older adults. For the first time, the current study examines dispositional orientations toward learning and/or performance goals in younger and older adults, in relation to memory. Using a related construct, namely process versus outcome orientation, Freund, Hennecke, and Riediger (2010) found that older adults were more likely to focus on the actual process of pursuing a life goal (process orientation), whereas younger adults tended to focus on the end result (outcome orientation). This finding suggested that older adults may be more oriented toward learning goals than performance goals, because learning goals incorporate the process of building mastery and performance goals are more concerned with outcomes. However, goals related to cognitive skill and goals related to the pursuit of life tasks may not show the same age patterns. Due to an increased tendency to attribute memory performance to uncontrollable causes (similar to an entity theory), older learners may be more likely to endorse performance goals than younger learners (Elliott & Lachman, 1989). Furthermore, with the low self-perceptions of ability prevalent in aging (Berry, 1999; West, Thorn, & Bagwell, 2003), older adults who have adopted performance goals may be at risk for negative learning outcomes associated with a performance-avoidant style, such as poor use of strategies and low memory performance (Elliott & Lachman, 1989). This research will address these important issues.

Although the link between memory self-efficacy and achievement goal orientation has not been examined in older samples, the relationship between memory self-efficacy and memory performance has been investigated as a function of aging¹. In the short term, research has supported a positive relationship between memory self-efficacy and memory performance, such that those who report feeling more capable of success on memory tasks perform better on memory tasks (Berry, 1999; West et al., 2009), with a stronger relationship for everyday tasks in contrast to laboratory memory tests (West, Dennehy-Basile, & Norris, 1996), and stronger efficacy-performance links for older adults than younger adults (West et al., 2009). Memory self-efficacy also seems important for future memory performance in older samples. Valentijn and colleagues found that memory self-efficacy predicted actual memory performance six years later (Valentijn et al., 2006). The long-term benefits of higher memory self-efficacy may occur because people high in domain-specific self-efficacy are likely to engage in more challenges, persist longer, and work harder in that domain (Bandura, 1997; West et al., 2009). For instance, adults higher in memory self-efficacy tend to set higher goals for their own memory performance (Berry & West, 1993; West & Thorn, 2001) and tend to be more motivated by memory goals (West, Bagwell, & Dark-Freudeman, 2005), and self-efficacy contributes to responsiveness to different memory goals (Stine-Morrow, Shake, Miles, & Noh, 2006). For these reasons, memory self-efficacy is an important social-cognitive factor to examine in late life. Because efficacy has been proposed to exacerbate the negative effects of performance goals, it is also critical to examine memory self-efficacy within the achievement goal framework using an older sample, as the current study sets out to do.

Aims and Hypotheses

To begin to understand these important relationships in adults, we tested an exploratory model (Figure 1) proposing links between achievement goal orientation, self-efficacy, and memory. A multiple group analysis was also conducted to examine whether the model held true for both older and younger groups, as expected. Based on past research, age and education were predicted to affect self-efficacy and performance. Specifically, younger participants should perform better and endorse higher memory self-efficacy than older participants. Memory self-efficacy should predict memory performance. Additionally, those with more years of education should perform better and due to the cumulative impact of better performance, should also endorse higher memory self-efficacy than those with fewer years of education (Lachman, Agrigoroaei, Murphy, & Tun, 2010). It is unknown whether increased age will be related to learning or performance goals. Past aging research suggests that older adults may focus on the process of goal pursuit (similar to learning goals; Freund et al., 2010). However, scholars have also suggested that older adults may be more likely to hold an entity theory of cognitive ability, implying a tendency toward performance goals (Elliott & Lachman, 1989).

Figure 1.

Predicted model of dispositional goal orientation, memory self-efficacy, and memory performance (dashed lines represent paths that were removed to create the final model, presented in Figure 2).

The achievement goal framework has not been widely applied to adults and memory. Therefore, the predictions for learning goal orientation and performance goal orientation are based on past research – mainly conducted with children in relation to grades and problem-solving skill. Based on that literature, learning goals should have a positive impact and performance goals should have a negative impact on memory performance. Further, based on self-efficacy theory, it was predicted that learning goals would have a positive impact on memory self-efficacy and performance goals would have a negative impact.

The current study also examined subjective performance. Theoretically, individuals with learning goals should appraise their performance more positively than those with performance goals (after controlling for actual performance), because those with learning goals tend to focus less on perceived task failure (Elliott & Dweck, 1988; Phillips & Gully, 1997). Finally, it is expected that subjective performance will be positively affected by actual performance. Relationships between subjective memory ratings and objective performance are often modest when such ratings are assessed before a memory test, but subjective reports are often fairly accurate after testing for both old and young (Hertzog & Dixon, 1994). Thus we expect that those who performed better on the memory tasks should report a higher subjective rating of performance. See Figure 1 for an illustration of these predicted relationships.

Methods

Participants

College students, aged 18–22, and older adults, aged 60–84, were recruited in classes or through referrals from other participants (i.e., snowball sampling), respectively. Younger adults were given course credit for participation and older adults were compensated with $10 gift cards. Five participants were eliminated for evidence of cognitive problems (use of anticholinergic medications, dementia diagnosis, stroke, or difficulty following instructions), or for conditions that significantly interfered with memory assessment via telephone (i.e., language problems or major hearing loss). Data were taken from a survey packet and a follow-up phone call typically completed about two weeks later; six participants who did not complete the two parts of the study within 30 days of each other were also excluded.

The final sample included 119 undergraduates and 97 older adults. Sample characteristics are reported in Table 1, which shows that the participants were healthy (self-rated health around seven or eight on a Likert scale where 1=very poor and 10=excellent health) and well-educated (mainly high school graduates). The majority of participants were Causasian (67.1%), followed by Hispanic/Latin (11.1%), African American (12.5%), and mixed/other categories (9.3%). Sample size (N=216) was adequate for estimation of the models, according to Bentler and Chou (1987), who recommends 5–10 participants per parameter estimated (33 distinct parameters were estimated in the predicted model, and 29 in the final model). Kline (2005) also suggests that samples over 200 can be considered “large” for structural equation models.

Table 1.

Sample characteristics

Variable	Mean			Standard deviation
	Younger	Older	Total	Younger	Older	Total
Age (years)	18.50	71.28	42. 07*	.86	6.58	26.67
Education (years)	12.74	15.10	13.80*	.93	2.82	2.33
Self-rated health	8.27	7.40	7.88*	1.21	2.11	1.73
Learning goals (GOQ)	41.56	43.18	42.28	10.55	9.44	10.08
Performance goals (GOQ)	41.57	42.90	42.16	10.92	8.81	10.02
Memory self-efficacy (MSEQ-4)	73.84	64.53	69.66*	14.62	20.03	17.83
Free recall (words recalled)	4.92	3.98	4.49*	2.66	3.46	3.08
Cued recall (words recalled)	25.05	21.41	23.41*	2.92	4.40	4.08
Subjective recall	5.99	5.21	5.64*	1.65	1.94	1.82

^*Significant mean difference (t-test) between age groups, p<.05

Materials and Procedure

During the first part of the study, participants completed an informed consent form, measures of goal orientation, memory self-efficacy, self-rated health, and general demographic information (in that order) in a take-home survey packet. Most young participants filled out the packet in a laboratory office, or took it home and returned it within four days. All referral participants (mostly older adults) were mailed the survey packet along with a postage-paid return envelope. Table 1 presents mean scores for each of the survey packet measures, for each age group and the pooled sample, and indicates for which variables a significant age difference was found. An average of two weeks later (M=13.48 days, SD=4.13), participants were contacted via telephone to complete an incidental learning task, followed by free and cued recall. The incidental learning task was administered under conditions which were designed to emphasize 1) learning goals or 2) performance goals, or 3) neither (control). These conditions did not have any significant effect on recall (p’s > .25) and will not be considered further. On average, the entire phone call lasted 20–30 minutes.

Goal orientation

The goal orientation questionnaire (GOQ; Button et al., 1996) was used to assess general predisposition toward learning and/or performance goals. The GOQ includes eight learning goal items (examples: “I prefer to work on tasks that force me to learn new things” and “The opportunity to do challenging work is important to me”) and eight performance goal items (examples: “I like to be fairly confident that I can successfully perform a task before I attempt it” and “The opinions others have about how well I can do certain things are important to me”). Each question was answered on a 7-point Likert scale (1= strongly agree and 7 =strongly disagree). Responses were reverse-coded such that a higher score indicated a stronger goal orientation. The relevant items were summed separately for learning (α=.93) or performance goals (α=.90), resulting in a possible score range of 8–56 for each scale. Convergent and divergent validity for these measures has been established (Button et al., 1996).

Memory self-efficacy

The Memory Self-Efficacy Questionnaire (MSEQ-4; West et al., 2005) was used to obtain a measure of confidence for success on memory tasks, to be used as an indication of perceived ability. The MSEQ-4 consists of questions about remembering names, stories, a shopping list, and object locations, with five items for each task. For the first item, participants rated their confidence (0–100%) that they could complete the task to a high criterion (shopping list example: “If I went to the store the same day, I could remember 18 items from a friend’s shopping list of 18 items, without using a list.”). Each remaining item asked participants to rate their confidence (0–100%) on a lower criterion (for example, question two asks about remembering 14 of the 18 items, and question three asks about remembering 10 items, etc.). Comparable progressive changes in task requirements were included on each of the four subscales. Responses were averaged across subscales to obtain an overall measure of self-efficacy strength, creating a possible score range of 0–100 (α=.95). The MSEQ-4 has shown high validity and reliability in past research (Berry, West, & Dennehy, 1989; West et al., 2003).

Memory

Once participants returned the informed consent form, basic demographic information, and the questionnaires described above, they were contacted by telephone to complete an incidental memory task, which was referred to as a “word game.” The paradigm was adapted from one of the few goal orientation studies focused on memory performance (Graham & Golan, 1991) – the task was scripted for telephone administration, and more statement-noun combinations were added to make the task more challenging for an adult sample. The word game included two practice trials and 40 scored trials. Each trial consisted of a question (e.g., “Is this word a type of fruit?”) followed by a two-second pause. Then, a target noun, referred to as the “game word” was presented (e.g., “cherry”). Participants were asked to indicate yes or no as to whether the target word fit the question (in the example above, the correct response would be yes). There were 40 questions, representing two types, with 20 items for each type: sentence (e.g., “Does this word fit into the sentence: She spilled the _______?”) and category (e.g., “Is this word an ingredient for cooking?”). Two different question orders were assigned randomly to participants. After completing the 40 trials, participants were given a surprise free recall task, in which they were asked to verbally list all of the target words they could remember. Then, a surprise cued recall task was administered, in which each question served as a cue (the experimenter re-read each question, and the participant was to state the target word that was associated with each question). Cue words were counterbalanced, using two new word orders. During the word game, 25% (10 out of 40) of the questions required a negative response (Graham & Golan, 1991) to provide response variety. Following the methodological recommendations of Craik and Tulving (1975), these 10 negative response items were not counted. Therefore, for free and cued recall, scores represented the number of target nouns correctly recalled out of 30 (possible score range = 0 to 30).

Other measures

The Short Form Health Survey (SF-36; Ware & Sherbourne, 1992) is a brief health questionnaire with good reliability and validity, including eight subscales. This measure was administered in the survey packet to screen participants for eligibility based on three mental health subscales related to depression: Vitality, Emotional Limitations, and Mental Health. Responses were summed within each individual subscale following the procedures recommended by Ware and Sherbourne (1992).

After the free and cued recall tasks, participants were asked to provide a simple subjective rating of their overall memory performance on those two tasks on a scale from 1–10, where 1= excellent and 10=very poor. Responses were reverse-coded such that higher scores indicated a higher subjective rating of performance.

For any older participant who had difficulty following instructions during the telephone survey, we administered the Telephone Interview of Cognitive Status (TICS; Brandt, Spencer, & Folstein, 1988) just prior to debriefing. When used, it was presented to the participant as a final activity that was a standard part of the phone interview. The TICS is an 11-item, standardized testing instrument that covers 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 scores on the Mini-Mental State Examination (MMSE; Folstein et al., 1975). The TICS has high sensitivity (94%), specificity (100%), and test-retest reliability (r = .96), with regard to the diagnosis of Alzheimer’s disease (Brandt et al., 1998; Dal Forno et al., 2006). Published cutoff score criteria (<25; Desmond, Tatemichi, & Hanzawa, 1994) were applied to the current sample. The TICS was administered three times, and each time the participant scored above the cutoff of 25 (Desmond et al., 1994). Thus, in no case did the TICS result in the elimination of a participant.

To be sure that participants were not expecting a memory test following the word game, two yes/no questions were asked for a manipulation check at the end of the phone call: “During the word game, were you trying to memorize the words?” and “During the word game, did you realize that you would be asked to recall the game words later?” Responses to both questions were analyzed with regard to age group, education, memory self-efficacy, and the two goal orientation variables (learning goals and performance goals) to examine the distribution of responses across the independent variables in the models.

Results

The manipulation check suggested that participants were adequately surprised by the recall tasks. Only 8.3% of participants reported trying to memorize the words, and only 5.1% of participants said they thought they would be asked to recall the words. There was no association between the second manipulation check question and any of the variables in the model. However, t-tests revealed that participants who responded yes to the first manipulation check question (indicating that they were trying to remember the words) were significantly more likely to be part of the older age group, t(20.77)= −2.59, p<.05, and were less likely to endorse dispositional learning goals, t(212)= 2.55, p<.05. In no instance did allowing this first manipulation check question to correlate with learning goals and age group in the model impact any parameters. Therefore, it was not included in the model.

Participants who scored more than two standard deviations below the group mean (bottom 5%) on any of the self-report mental health subscales from the SF-36 (N= 17; 5 younger adults and 12 older adults) were eliminated for preliminary model runs. These preliminary model runs were compared to the final model reported here, and it was found that all relationship patterns were maintained and model fit statistics were very similar. Therefore, to maximize statistical power, these participants were included in the dataset to create the models discussed in the remainder of this paper.

The model was analyzed using Amos 17.0, to examine the predicted relationships between demographic variables (age group and education), beliefs (dispositional goal orientation, memory self-efficacy, and subjective performance), and memory (free and cued recall). Age was used as a dichotomous variable, such that 0= younger adult and 1=older adult, to reflect the categorical nature of the age groups in this sample. Age and education, education and learning goals, and learning and performance goals were allowed to covary, because preliminary analyses suggested that these were the only significantly correlated exogenous variables in the model (see Table 2). As shown in Table 2, age was not related to holding either learning or performance goals. Following Kline (2005), these criteria were used to indicate good model fit: non-significant Chi-Square, CFI and TLI ≥ .90, and RMSEA ≤ .05.

Table 2.

Inter-correlations between model variables for the pooled sample

Variable	1	2	3	4	5	6	7	8
1. Age group	1.00
2. Education	.51**	1.00
3. Learning goals	.08	.17*	1.00
4. Performance goals	.07	.05	.66**	1.00
5. Memory self-efficacy	−.26**	.06	.14*	−.06	1.00
6. Free recall	−.15*	−.03	−.08	−.09	.19**	1.00
7. Cued recall	−.45**	−.16*	−.06	−.02	.24**	.41**	1.00
8. Subjective performance	−.22**	.06	.09	.03	.24**	.29**	.51**	1.00

^*p<.05,

^**p<.01

Predicted Model (Figure 1)

The predicted model (Figure 1) showed excellent fit to the data, χ²(11)=13.0, p=.29, CFI=.995, TLI=.982, RMSEA=.03, suggesting that the overall pattern of predictions was supported in this sample. In order to construct the more parsimonious final model (Figure 2), four nonsignificant paths were trimmed (including paths from learning goals and performance goals to memory, education to recall, and performance goals to subjective performance). Removing these paths did not significantly degrade model fit, according to a nested-model Chi-Square test, Δχ²(4)=4.22, p=.38. Therefore, the remainder of this paper will focus on the results obtained in this more parsimonious model.

Figure 2.

Final model of dispositional goal orientation, memory self-efficacy, and memory performance.

Final Model (Figure 2)

Free and cued recall measures were used to construct a latent factor for memory. Each of these indicators significantly contributed to the latent factor (loadings for free and cued recall on memory were .45 and .92, respectively, both p<.001). The final model showed excellent fit to the data, χ²(15)=17.22, p=.31, CFI=.994, TLI=.986, RMSEA=.03.

Age and education were significantly positively correlated (r=.50, p<.01), with the older group reporting more years of education than the younger group. Both age and education predicted memory self-efficacy. Age group had a negative impact on memory self-efficacy (β= −.38, p<.01), with younger participants reporting higher memory self-efficacy. Education had a positive impact on memory self-efficacy (β=.22, p<.01), such that more educated individuals reported higher memory self-efficacy. Age group negatively predicted memory performance (β= −.43, p<.01), such that younger participants had better performance than older participants.

Dispositional learning goals and performance goals were highly correlated (r=.66, p<.01), suggesting that such goals are not mutually exclusive. Learning goals were also significantly related to education (r=.12, p<.05). Learning goals had a positive effect on memory self-efficacy (β= .28, p<.01), in that participants higher in general dispositional learning goals reported higher memory self-efficacy. Performance goals had a negative effect on memory self-efficacy (β= −.22, p<.01), such that participants higher in general dispositional performance goals reported lower memory self-efficacy. These two variables (learning and performance goals), along with the demographic variables (age and education) explained 16.9% of the variance in memory self-efficacy in the final model.

Subjective performance was positively predicted by actual memory performance (β= .56, p<.01), with participants fairly accurately assessing their general level of performance. Learning goals also had a significant effect on subjective performance independent of actual performance (β= .12, p<.05). The final model explained 33.6% of the variance in subjective performance.

Memory self-efficacy had a significant positive impact on memory performance (β= .16, p<.05), with higher memory self-efficacy leading to better performance. Sobel tests were conducted to learn whether learning and/or performance goals had significant indirect effects on memory performance, via memory self-efficacy, because it was discovered that their direct effects were non-significant (and were therefore removed from the final model). Results suggested that the indirect effect of learning goals on memory performance via memory self-efficacy was significant (Sobel test statistic=1.97, p<.05) and the indirect effect of performance goals on memory performance via self-efficacy was marginally significant (Sobel test statistic= −1.81, p=.07) (Preacher & Leonardetti, 2006). Thus, learning goals seemed to have a positive effect on memory performance, and performance goals seemed to have a negative effect on memory performance, but both effects occurred indirectly through the impact of both goals on memory self-efficacy. About 24.8% of the total variance in memory performance was explained by the final model.

To further examine directionality, an exploratory model was run that was identical to the final model presented in this paper, however, with the two causal arrows between learning and performance goals to memory self-efficacy reversed.² Although model fit was comparable, paths from memory self-efficacy to learning goals and performance goals were non-significant, suggesting that the causal direction represented in our model is valid.

Multiple Group Analysis

A multiple group analysis was run using Amos 17.0, to ensure that the pattern of results was replicated in both age groups (see Kline, 2005). In this analysis, the age variable was treated as a continuous variable so that continuous age effects could be examined within each age group of participants. Results indicated that the general pattern of results, including regression path weights, held true for both age groups. When measurement weights and intercepts, as well as structural weights and intercepts were constrained to be equal in the younger and older sample, model fit was still very good: χ²(43)=47.83, p=.28, CFI=.986, TLI=.977, RMSEA=.02. Therefore, to maximize statistical power, this report focused on the two age groups combined. The multiple group analysis illuminated only two differences between the older and younger participants: the correlation between learning goals and education was significant only for older participants, and the relationship between age and education was significant only in the younger group. These results are not surprising, given that the younger participants were primarily undergraduate students, whose ages generally correspond with their years of education, and both of these variables (age and years of education) show a relatively narrow range in the younger group.

Discussion

The results of the present study demonstrate that learning and performance goals, outlined in Dweck’s (1986) achievement goal framework, may have effects in older adult learners that are similar to those established in younger students. It has been shown that older adults are more likely than younger adults to report that their memory performance is due to external uncontrollable causes (similar to an entity theory of intelligence; Elliott & Lachman, 1989; Lachman, 2006; Miller & Lachman, 1999), but to our knowledge, the present study represents the first empirical examination of aging and learning and performance goals, per se. The validation of this framework in young and older adults is important to the understanding of the influence of social-cognitive factors on memory performance. Research has often shown that older adults report relatively low memory self-efficacy (Berry, 1999; West et al., 2003), and that memory self-efficacy is related to performance (Berry, 1999; Valentijn et al., 2006; West et al., 2009). The current study builds on these findings by showing how goal orientation is related to both efficacy and performance.

The expected outcomes of achievement goals were generally supported, however, the models suggested that these outcomes were mediated by their impact on other social cognitive factors (specifically, memory self-efficacy). This indirect effect of learning and performance goals on performance via memory self-efficacy has been found using adolescent (Boon, 2007) and college-aged samples (Coutinho & Neuman, 2008; Phillips & Gully, 1997; Phan, 2009a), but not with older adults. In keeping with this past literature, our model conceptualized memory self-efficacy as a causal effect of dispositional goal orientation; we found that learning goals had a significant positive effect on memory self-efficacy, and that performance goals had a significant negative effect on self-efficacy. In turn, memory self-efficacy positively predicted memory performance, leading to a significant indirect effect for learning goals on performance and a marginally significant indirect effect for performance goals. Therefore, the model presented here supported the importance of dispositional goal orientation for memory self-efficacy, and the importance of memory self-efficacy for memory performance across adulthood.

The impact of achievement goals on memory self-efficacy is an important finding because it indicates that dispositional learning goals are adaptive, and performance goals may be maladaptive, in the context of adults and memory. These findings are especially interesting in light of self-efficacy theory (Bandura, 1997). Individuals who consistently approach tasks with the goal of challenging themselves and building mastery have likely had more past successes (mastery experiences), and approach cognitive tasks with less anxiety (relaxed physiological state). Both of these factors are proposed to enhance self-efficacy.

It is also interesting that those higher in learning goals were more likely to positively appraise their memory performance, after controlling for actual performance, because it suggests that these participants were less attuned to memory failures and were therefore biased to overrate their performance. This finding fits with the achievement goal framework’s supposition that people with learning goals are less affected by task failure, because they do not view failure as a mark against fixed ability (Smiley & Dweck, 1994; Phillips & Gully, 1997). A focus on successes rather than failures is likely helpful for intrinsic motivation and pursuit of challenges, and may help further explain the positive relationship between learning goals and memory self-efficacy. If people view cognitive experiences as more successful, they interpret more of their experiences as masterful, and mastery experience is proposed to be the most powerful source of self-efficacy (Bandura, 1997).

Other predictions set forth at the beginning of the study were supported as well. Age had the expected effect: younger participants reported higher memory self-efficacy and performed better on the memory tasks. Also, those with more years of education endorsed higher memory self-efficacy than those with fewer years of education. Not surprisingly, those who performed better also reported having better performance on the post-test subjective rating.

Learning goals and performance goals were positively associated. Interestingly, aging scholars have suggested that adults’ conceptions of ability may not be purely entity theories (fixed) or skill theories (incremental) (Hertzog et al., 1999; Kanfer, 1990). That is, adults may simultaneously believe that biological aging processes lead to memory decline, but still think that behavior (e.g., mental exercise) may ameliorate that decline. Cognitive activity can be seen as a way to exert some compensatory control (skill theory) over a somewhat uncontrollable biological aging process (entity theory) (Hertzog et al., 1999). In this context, the correlation between learning and performance goals makes sense in the aging literature, and as indicated earlier, these two types of goal orientation are often correlated in younger adults as well (Harackiewicz et al, 2002).

A related issue is the fact that the Goal Orientation Questionnaire used in the present study did not distinguish the approach and avoidant styles of performance goals (Elliot & Murayama, 2008; Hulleman et al., 2010). The correlation between learning and performance goals could possibly represent an overall “approach” dimension that is common between those with learning goals and those with performance-approach goals (Elliot & Murayama, 2008). However, it is important to note that for all relationships estimated in the model, the relationship between learning and performance goals was controlled for statistically, by estimating their correlation in the model. Thus, it may be reasonable to assume that the performance goal variable in the model represents the “avoidant” elements of performance goal orientation (what is left after controlling for any relationship with learning goals) and the learning goal variable in the model represents a learning orientation independent of performance-approach goals. Thus, all relationships between learning or performance goals and any other model variable can be interpreted as the unique effects of each type of achievement goal.

The findings presented here are especially interesting in the context of recent cognitive aging research. Stine-Morrow and colleagues’ self-regulated language processing model (SRLP; Stine-Morrow et al., 2006) considered the impact that motivational factors may have on self-regulated learning. The authors suggested that perceptions of control over performance (how changeable it is), as well as confidence in the ability to succeed (self-efficacy), together may affect allocation of cognitive resources. If the cognitive resources invested in a task are not sufficient to compensate for normative age-related changes, this failure of self-regulated learning can negatively impact performance. Thus, one possibility is that the negative outcomes associated with performance goals and memory self-efficacy in the current study may be partially due to their role in self-regulation (Stine-Morrow et al., 2006). Our model supports the SRLP model by suggesting that adult learners who adopt performance goals have a tendency to be lower in memory self-efficacy and to perform at a reduced level (as would be expected if fewer cognitive resources were being allocated to the task). If adult learners engage in this maladaptive learning pattern continuously over time in their everyday lives, the cumulative effect of these failures may further undermine motivation and future performance (West, Welch, & Yassuda, 2000). Thus, the present study yields an important and original contribution to the aging literature by showing how the achievement goal framework could link key self-regulatory factors that have been widely supported as crucial to aging and memory: cognitive resource allocation, control beliefs, implicit theories, and memory self-efficacy.

While providing some of the first evidence that the achievement goal framework operates as expected in an older adult sample, the current study also has several limitations that should be addressed in future research. First, this study was cross-sectional, and longitudinal investigations would be needed to examine the more dynamic, and possibly reciprocal, relationships among memory, goal orientation, and self-efficacy over time. Second, every effort was made to eliminate participants with obvious cognitive impairment, but a formal assessment of mild cognitive impairment was not conducted (Petersen et al., 1999). Thus, it is possible that some participants had subclinical or undiagnosed cognitive impairment which may have limited the effect to which dispositional learning or performance goals could impact performance (by capping the maximum potential of some participants). Third, the current study used a memory task that had shown some significant performance effects in past goal orientation research (Graham & Golan, 1991), but it is possible that a more challenging memory task might yield different results. Given the strong support for the benefits of learning goals on cognitive performance found in past research (e.g., Utman, 1997) it was expected that participants high in learning goals would be more engaged in the incidental word game, and thus perform better when surprised with recall tasks. However, because participants were not aware they would be tested, and the task only required a simple yes or no response, it is possible that participants high in learning goals did not really view the task as an opportunity to build mastery (Utman, 1997). Likewise, participants high in performance goals may not have been significantly threatened by the potential for poor performance on such a task, and thus their goal orientation did not impair information processing. Additionally, participants were likely not using cognitive strategies during the incidental task, and the impact of goal orientation on performance is often related to strategic differences (Elliott & Dweck, 1988; Phan, 2009a; Phan, 2009b).

The present study did not examine the role of perceptions of memory controllability. Thus, it is not known whether implicit theories of ability impact achievement goal adoption in the same way they do in children (Elliott & Dweck, 1988). Future research should examine factors that lead to learning or performance goal adoption in adulthood, to learn whether perceptions of memory controllability influence achievement goal adoption as has been shown in children (Blackwell et al., 2007; Dweck & Leggett, 1988). If the achievement goal framework can be further validated in an adult sample, an important next step would be to examine modifiability. It is possible that, as people age and have many cognitive experiences, goal orientation becomes more dispositional and ingrained. However, memory training studies have been successful in modifying cognitive beliefs, such as memory self-efficacy and locus of control (Floyd & Scogin, 1997; Lachman, Weaver, Bandura, Elliott, & Lewkowitz, 1992; Valentijn et al, 2005; West, Bagwell, & Dark-Freudeman, 2008), providing some evidence that such beliefs can be modified even into older age.

In conclusion, scholars should continue to explore mechanisms for the impact of goal orientation on performance in aging. The present study established a link between goal orientation and memory-self efficacy, but there are other potential mechanisms suggested in research using younger samples, including level of cognitive processing (Coutinho & Neuman, 2008; Phan, 2009a; Phan, 2009b; Sins, van Joolingen, Savelsbergh, & van Hout-Wolters, 2008), goal-directed behavior (Seijts, Latham, Tasa, & Latham, 2004), critical thinking (Phan, 2009a), and metacognition (Coutinho & Neuman, 2008). The role of self-regulation, including allocation of cognitive resources, should also be examined in the context of achievement goals, self-efficacy, and memory performance, as suggested by Stine-Morrow and colleagues (Stine-Morrow et al., 2006). It is possible that goal orientation has a broad impact on how adults approach tasks in multiple domains in everyday life: how they interpret failure when it occurs, their confidence for success, and ultimately, how well they perform. Learning more about this interplay between goal orientation, social cognitive factors (including self-efficacy), and performance could be important for understanding cognitive functioning and engagement in adulthood.