*موسیقی درمانی *

موسیقی درمانی رشته های موسیقی ودرمان را باهم تلفیق و ترکیب می کند تا درمان و وسیله ای خلاق را فراهم کند این روش الگوه های موسیقی رابا رویکردهای انسانی ، روان پویایی، رفتاری وبیولوژیکی – طبی تلفیق می کند تا کمک کند که درمان جویان به اهداف درمانی که معمولاً ماهیتی روانی، جسمانی ، عاطفی ، اجتماعی ویا معنوی دارند،  نائل شوند . مشکلات یا نیازها از طریق رابطه درمانی میان درمانجو وموسیقی درمانگر وهمچنین مستقیماً از طریق خود موسیقی مورد بررسی قرار می گیرند . موسیقی یکی از اجتماعی ترین انواع هنر است که به شیوه های بسیار مختلفی بین انسانها ارتباط برقرار می کند. بررسی های متعددی درخصوص تغییرات فیزیولوژیکی مفیدی که توسط موسیقی برروی خلق وخو ، فشار خون ، تنفس ، ضربان قلب ، تعرق ، بازده قلبی ، نبض ، کشش عضلات ، درد وتسکین درد به وجود می آید ، انجام گرفته است . موسیقی به لحاظ درمانی به عملکرد جسمانی ، روان شناسی ، شناختی ، رفتاری ، ویا اجتماعی می پردازد .

متخصصان موسیقی درمانی درشرایط گوناگونی باافرادی که دچاراختلال عاطفی هستند،

با معلولین ذهنی باافرادی که دچار ناتوانی یادگیری هستند،وباافرادی که دچار مشکلات

جسمانی هستند ، کار می کنند . متخصصین موسیقی درمانی همچنین با درمان جویان دچار اختلالات روانی ، افراد دارای مشکلات ناشی از مصرف الکل ومواد مخدر ، وافراد مبتلا به اختلالات نورولوژیکی وافرادی که بطور علاج ناپذیری بیمار هستند ، کار می کنند . متخصصین موسیقی درمانی خدمات خود را در مراکز مراقبتی ماهرانه ومتوسطه ، درمراکز مراقبتی وپرورش بزرگسالان ، دربیمارستان های توانبخشی ، در تسهیلات مراقبتی منازل مسکونی ، دربیمارستان ها ، درمراکز مراقبت های روزانه بزرگسالان ، درمراکز بازنشستگی ،در مراکز وخانه های سالمندان ، دربرنامه های ارزیابی سالمندان ، درمراکز درمان بیماریهای روانی ، ودرسایر مراکز ارائه می دهند . متخصصین موسیقی درمانی همچنین برای موسسات تامین کننده مراقبت های درمانی درمنازل نیز کار می کنند . ازموسیقی درمانی همچنین درمورد افراد سالم وتن درست برای کمک به کاهش استرس ، درزایمان کردن خانم ها ، ودر بازخورد بیولوژیکی استفاده می شود . فرصت های پیش رفته برای تحصیل ومشاغل خصوصی نیز دراین زمینه امکان پذیر است .

موسیقی درمانی برای خانواده ها وکودکان

·       زمینه ای برای تشریک تجربیات ولذت وخوشی مشترک به عنوان یک زوج یا خانواده را فراهم می کند .

·       به شیوه ای مثبت وخلاق به گذران وقت وزمانی که با یکدیگر صرف می شود معنا می بخشد .

·       ازطریق تعامل درمانی کلامی وغیر کلامی برای خانواده ها اتحاد وصمیمیت را حاصل می نماید .

·   تمامی حواس را تحریک می کند وکودک را درسطوح مختلف شامل می گردد . این رویکرد "چندحالتی" مهارتهای رشد وپیشرفت را ترویج می کند.

·       امکان وقوع طبیعی و مکرر نمایش را فراهم می کند.

·   تضادهارا برطرف می کند ، وبه روابط قویتر خانوادگی وگروه همسالان وهمچنین تغییرات مثبت در خلق وخو وحالات احساسی وعاطفی منتهی می گردد .

·       می تواند به کودک دراداره موقعیت استرس زا کمک کند .

·   اجتماعی شدن ، بیان حال ، برقراری ارتباط وپیشرفت وتکامل حرکتی را تشویق می کند وهمچنین به کشف احساسات شخصی ومسائل درمانی از قبیل عزت نفس یا بینش شخصی کمک می نماید .

·   آزمون واقعیت ، حل مسائل ، تمرکز حواس ، حیطه توجه را بهبود می بخشد وهمچنین استقلال ومهارتهای تصمیم گیری را ایجاد می کند .

·   چون مغز، موسیقی را درهردو نیمکره مغز پردازش می کند ، بنابراین موسیقی می تواند عملکرد شناختی را تحریک کند ومی توان از موسیقی برای ترمیم بعضی از مهارتهای کلامی / یا زبانی استفاده کرد . (بارتلت ، کافمن ، اسملت کوپ ،1993 ؛ بولت ،1996) .

موسیقی درمانی می تواند به عنوان دریچه مثبت برای تعامل بکار رود ، وفعالیتهای تفریحی را فراهم سازد که والدین ،فرزندان وخانواده های گسترده را شامل گردد . اغلب اوقات موسیقی درمانی این امکان را به وجود می آورد که خانواده ای با تظاهر یافتن نقاط قوت فرزند خود درمحیط  موسیقی درمانی با دید دیگری به فرزند خود نگاه کنند . موسیقی درمانی می تواند درمانی التقاطی باشد که مفاهیم موجود در روان درمانی "راجرزی" (انسانی) رفتارگرایی شناختی (یعنی گفتار شخصی مثبت ) وروان درمانی "یونگ" ( آنالیتیک یا روان تحلیلی ) رادربرمی گیرد تابه درمان جویان برای نیل به اهداف درمانی آنها کمک کند .

متخصصین موسیقی درمانی درشرایط بسیار متنوعی درکار باافراد دچار آشفتگی عاطفی، افراد دچار ناتوانی یادگیری، افراد عقب مانده ذهنی ، وافراد دارای مشکلات جسمانی موفق بوده اند .

متخصصین موسیقی درمانی همچنین با درمان جویان دچار اختلالهای روانی ، افراد دارای مشکلات استعمال مشروبات الکلی ومواد مخدر ، افراد دچار اختلال های نورولوژیکی ( عصب شناختی وافرادی که دارای بیماریهای علاج ناپذیر هستند ، کار می کنند . از موسیقی درمانی همچنین درمورد افراد سالم استفاده می شود تابه آنها درکاهش استرس ، در زایمان ، ودر بازخورد بیولوژیکی کمک کند (بارتلت، کافمن ، اسملت کوپ،1993 ، بولت ،1996

Social cognitive predictors of college students’ academic performance and persistence: A meta-analytic path analysis

Steven D. Brown^{a, ,} , Selena Tramayne^a, Denada Hoxha^a, Kyle Telander^a, Xiaoyan Fan^a and Robert W. Lent^b
aLoyola University Chicago, School of Education, 820 N. Michigan Avenue, Chicago, IL 60091, USA
^bUniversity of Maryland, Department of Counseling and Personnel Services, 3214 Benjamin Building, College Park, MD 20742, USA
Received 10 July 2007.  Available online 3 December 2007.

Abstract

This study tested Social Cognitive Career Theory’s (SCCT) academic performance model using a two-stage approach that combined meta-analytic and structural equation modeling methodologies. Unbiased correlations obtained from a previously published meta-analysis [Robbins, S. B., Lauver, K., Le, H., Davis, D., & Langley, R. (2004). Do psychosocial and study skill factors predict college outcomes A meta-analysis. Psychological Bulletin, 130, 261–288.] were used to create the input correlation matrices for subsequent path-analytic tests of the model, using both college GPA and retention as performance criteria. Results suggested that SCCT does an adequate to excellent job of modeling academic performance and persistence, but that model fit was better when general cognitive ability versus high school GPA was used to operationalize the ability/past performance variable. Results are discussed in terms of their fit with SCCT and their practical implications.

Keywords: Social Cognitive Career Theory; College student performance; College student retention; Self-efficacy; Cognitive ability

Article Outline

1. Introduction

Social Cognitive Career Theory (SCCT) was developed by Lent, Brown, and Hackett (1994) to explain and predict the processes by which vocational and academic interests are developed, vocational and academic choices are made, and various levels of work and academic performance are attained. SCCT’s interest and choice models have attracted impressive research attention, yielding numerous individual studies (see [Lent, 2005] and [Swanson et al., 2000]) and meta-analyses (e.g., [Lent et al., 1994] and [Rottinghaus et al., 2003]) of individual hypotheses. A recent two stage, meta-analytic test of the full interest and choice models largely supported SCCT’s major hypotheses of interest development and choice making—that the relations of self-efficacy and outcome expectations to choices are both direct and partially mediated by interests, and interests have a direct relation to choices (Sheu et al., 2006).

While research (both primary and meta-analytic) relevant to SCCT’s performance model is also abundant, it has tended to appear in more diverse publication outlets (often in industrial–organizational psychology and personnel journals) and has focused more on work than academic outcomes. These studies were also often not derived directly from SCCT’s performance model, but their findings can be used to test various hypotheses of the model. Finally, unlike the interest and choice models, no single large-scale test of the complete performance model (i.e., containing all SCCT predictors) has appeared in the literature—although numerous studies have examined subsets of the model (e.g., one or more social cognitive predictors of performance or persistence).

SCCT’s performance model, as illustrated in Fig. 1, suggests that work and academic performance is a function of five conceptually distinct, but interrelated (in a reciprocal manner) cognitive and behavioral variables—general cognitive ability, past performance, outcome expectations, self-efficacy beliefs, and goal mechanisms.

Display Full Size version of this image (7K)

Fig. 1. Social Cognitive Career Theory Performance Model.

More specifically, with respect to academic performance, SCCT hypothesizes that general cognitive ability (e.g., as indexed by SAT or ACT scores) and past performance (e.g., high school GPA) both influence college student performance (e.g., GPA) and persistence (e.g., retention) in two ways: directly and indirectly through the mediating paths to student’s self-efficacy beliefs and outcome expectations. (Self-efficacy refers to confidence in one’s ability to accomplish academic tasks successfully; outcome expectations are beliefs about the consequences of engaging in academic tasks, e.g., that such efforts will lead to valued outcomes.)

Thus, according to SCCT, students who do well in, and graduate from, college do so, in part, because they have developed through their prior educational and social learning experiences the component academic (e.g., study, test-taking, writing) skills required for college success. They also do well in college because they have developed, via past performance accomplishments, cognitive aptitudes, and other sources of feedback (e.g., social encouragement, modeling), strong and robust academic self-efficacy beliefs and outcome expectations that lead them to approach rather than avoid challenging academic tasks, put effort into their college work, and persist in their academic careers despite encountering occasional set-backs and difficulties.

SCCT also posits that self-efficacy and outcome expectations affect performance and persistence at least partly through the intervening influence of students’ performance goals. For instance, students with stronger self-efficacy beliefs and outcome expectations will tend to set and work towards more challenging academic goals than will those with weaker efficacy beliefs or less positive outcome expectations. More challenging or difficult goals are, in turn, assumed to motivate students to work harder toward goal fulfillment, leading to more favorable academic outcomes.

As noted earlier, there has been a good deal of research on various paths of SCCT’s performance model in academic settings. The aspect of the model that has generated the most research involves the hypothesized relations of self-efficacy beliefs to academic performance and persistence outcomes. Studies of these relations have been synthesized in two comprehensive meta-analyses ([Multon et al., 1991] and [Robbins et al., 2004]). Meta-analytically-derived correlations between self-efficacy and college student academic performance, specifically, have ranged from .38 (Multon et al., 1991) to .50 (Robbins et al., 2004), with the latter providing the more accurate (least biased) estimate. That is, Robbins et al. (2004) fully corrected correlations for the biasing effects of sampling error, measurement error, and range restriction, whereas Multon et al. (1991) only corrected the observed correlations for sampling error (Robbins et al. reported an identical correlation to that of Multon et al. when similarly correcting only for sampling error). Robbins et al. (2004) also reported a fully corrected correlation of .36 between academic self-efficacy beliefs and college retention criteria. Thus, it appears that, when considered at the bivariate-level, academic self-efficacy beliefs are strongly related to college performance, and moderately related to retention in college.

A large body of research, although not developed specifically to test SCCT hypotheses, has also revealed substantial relations between measures of general cognitive ability (e.g., SAT, ACT, MAT, and GRE scores) and a variety of academic performance outcomes. Three recent meta-analyses have reported amazingly consistent fully corrected, bivariate correlations (r = .39) of GPA to ACT/SAT (Robbins et al., 2004), MAT (Kuncel, Hezlett, & Ones, 2004), and GRE (Kuncel, Hezlett, & Ones, 2001) scores. These findings suggest that general cognitive ability, no matter how measured, also contributes, in a very consistent manner, to both college and graduate school performance. Retention data provided in these same meta-analyses suggest that general cognitive ability might relate somewhat more to degree completion among graduate students (fully corrected rs of .22 in [Kuncel et al., 2001] and [Kuncel et al., 2004]) than undergraduate students (fully corrected r of .12 in Robbins et al., 2004), but correlations exceeded conventional levels of statistical significance in both cases. Finally, consistent with SCCT hypotheses, the bivariate relation between indices of past (high school) academic performance and college performance was also significant (r = .45 in Robbins et al., 2004).

Other paths in the model (i.e., the relations of abilities and past performance accomplishments to self-efficacy beliefs, self-efficacy beliefs to goals, and goals to academic performance and persistence) have been studied less frequently, but tests have largely supported SCCT’s hypotheses about the relations among these variables. For example, the Robbins et al. (2004) meta-analysis, found support, via fully corrected bivariate correlations, for each of the above hypothesized paths— rs of .28 and .70 for the relation of academic self-efficacy, respectively, to general cognitive ability and high school GPA; r of .49 between academic self-efficacy beliefs and academic goals; and rs of .18 for the relation of academic goals to performance, and .34 for the relation of academic goals to persistence.

These data, especially the fully corrected correlations reported in meta-analyses that provide estimates of construct-level versus measurement-level relations, offer strong support for each path in SCCT’s academic performance model (except that outcome expectations have rarely been the focus of research). However, bivariate correlations, even fully corrected meta-analytic ones, are problematic when used to evaluate the validity of complex, multivariate models. In particular, bivariate relations do not provide evidence of the unique influence of variables on each other in complex models. Rather, bivariate correlations are typically inflated estimates of the unique “influence” of one variable on another because they do not partial out the influence of other variables that might covary with either variable in the bivariate correlations. Thus, for example, the strong correlation that has been reported for the relation of self-efficacy beliefs and academic performance is undoubtedly influenced by past performance that covaries with both self-efficacy beliefs and current performance indices.

Indeed, some investigators (e.g., [Heggestad and Kanfer, 2005] and [Vancouver et al., 2001]) have argued that self-efficacy has no predictive power at all and is simply a proxy for past performance. Although Bandura and Locke (2003) summarized research that effectively countered this argument, the fact remains that some of the variance accounted for by self-efficacy in the prediction of various performance outcomes is not wholly unique, but rather accounted for also by past performance accomplishments and other covariates (e.g., general cognitive ability). Tests of complex, multivariate models, like SCCTs academic performance model, therefore, require a methodological and statistical approach (i.e., structural equation modeling) that allows one to estimate the unique relations among variables in the model and to estimate the fit of the model to the data when all relevant variables are included.

The primary purpose of this study is to provide the first structural tests of SCCT’s academic performance model. More specifically, because the bivariate relations reviewed earlier suggest that some paths in the model may yield different estimates when performance versus persistence is modeled, we conducted separate tests of the model’s ability to predict performance (college GPA) and persistence (retention) criteria. Also, since research has suggested that past performance (e.g., high school GPA) may be more strongly related to academic self-efficacy beliefs than are indices of general cognitive abilities (e.g., ACT or SAT scores), we ran two different tests of the performance and persistence models—one employing cognitive ability measures and the other past performance indices as predictors.

We also chose to model general cognitive ability and past academic performance separately because we wanted to explore whether the self-efficacy beliefs derived from these two sources of efficacy information related equally or differently to academic performance and persistence outcomes—whether or not they provided equally good efficacy-related information about a student’s likely success and persistence in college. Although SCCT makes no differential predictions in this regard, there are some reasons why they might not be equally good sources of efficacy information. For example, since past high school grades are somewhat influenced by the high schools in which they are attained (e.g., students in high performing high schools may receive lower grades than less able students from academically weaker high schools), self-efficacy beliefs derived from this source of information may not as accurately reflect a student’s likely success in college than information derived from cognitive ability measures. On the other hand, since high school grades are influenced by motivational as well as ability inputs, the self-efficacy beliefs derived from past high school performance may do a better job of predicting college success and persistence than those derived solely from cognitive indices of academic performance.

Finally, since meta-analytic estimates were available on all paths of the model, except for paths involving outcome expectations, we used meta-analytic estimates to create the correlation matrices that served as inputs for the four path-analytic model tests. This procedure of combining meta-analysis and path-analytic procedures to test complex multivariate models was first recommended by Viswesvaran and Ones (1995) and has since been employed successfully to test, for example, multivariate models of training motivation (Colquitt, LePine, & Noe, 2000), job satisfaction (Ilies & Judge, 2003), life satisfaction (Heller, Watson, & Ilies, 2004), and leadership performance (Podaskoff, MacKenzie, & Bommer, 1996).

In sum, we first employed previously reported meta-analytic data to examine the associations among the major constructs in SCCT’s academic performance model. We next created four correlation matrices using relevant meta-analytic estimates to fill each cell. We then used these as input to conduct four path analyses of SCCT’s academic performance model in college students—two of these analyses modeled academic performance and two modeled academic persistence. One version each of the performance and persistence analyses employed general cognitive ability as a predictor, while the other used high school performance.

2. Methods

2.1. Literature search

To identify all possible meta-analyses that could provide estimates of the relations among constructs in the model, we first searched the PsycINFO data-base using meta-analysis combined with the following other search term: self-efficacy, outcome expectations, academic ability, academic skills, Graduate Record Examination, GRE, LSAT, SAT, ACT, Miller Analogies Test, goals, grade point average, retention, college performance, and college persistence. We then used the reference lists of articles generated via the computer data-base and hand searched the tables of contents of relevant journals, ultimately generating a final set of nine meta-analyses, published between 1983 and 2004, relevant to at least one cell in the input correlation matrices. Of these, eight focused on the relations of general cognitive abilities or past academic performance to current academic performance or persistence, two meta-analytically explored the relations of self-efficacy beliefs to academic performance and persistence, and one provided data relevant to all paths except for those involving outcome expectations.

We subsequently engaged in a comprehensive literature search of the primary literature, using PsycINFO and hand searches, to identify individual studies of the relations of outcome expectations to the other constructs in the SCCT performance model. Our goal was to conduct our own meta-analyses of outcome expectations relations to fill in relevant cells in the input correlation matrices. Unfortunately, only outcome expectation-self-efficacy relations had been studied with sufficient frequency to yield valid meta-analytic estimates. We, therefore, eliminated outcome expectations from our analyses and instead tested a reduced version of SCCT’s performance model.

2.2. Rules for inclusion

After appropriate meta-analyses were identified, two criteria were used in decisions about whether to include them in subsequent path-analytic tests. First, it was important to ensure that all included meta-analyses used similar samples of participants (e.g., it would be inappropriate to include in the same matrix meta-analytic estimates based on high school, college, and graduate student samples). Second, all meta-analytic correlations included in the matrices had to be derived in the same way. There are, for example, a variety of effect size estimates used in the literature ranging from uncorrected correlations to those that have been fully corrected for the biasing effects of major statistical artifacts, including sampling error, measurement error on the predictor and criterion, and range restriction on the predictor and criterion. Our preference was to employ fully corrected correlations because these have been shown to provide the least biased estimates of relevant population parameters (Hunter & Schmidt, 1990).

One of the nine identified meta-analyses (Robbins et al., 2004) met both criteria in that it (a) used homogeneous samples of full-time college students who were enrolled in four year institutions, and (b) provided fully corrected effect size estimates relevant to all paths in the model (except those involving outcome expectations). Thus, the fully corrected correlations provided in this meta-analysis were used to fill in the cells in the four correlation matrices that were created as input for the four path analyses.

Robbins et al. (2004) meta-analytically explored the relations of a large number of academic and psychosocial variables to college performance and persistence. We extracted from Robbins et al., 2004 Table 7 (p. 272) only those fully corrected correlations that were relevant to the SCCT academic performance model—that is, correlations involving SAT/ACT scores, high school GPA, academic self-efficacy, academic goals, college GPA, and retention.

Table 1 presents the correlation matrix used for the academic performance model tests, while Table 2 is the correlation matrix used for the persistence model tests.

Table 1.

Correlation input matrix for the performance model tests

Variable	1	2	3	4
1. Ability	—	.283 (639)	.146 (6929)	.388 (16,648)
2. Self-efficacy	.703 (7436)	—	.489 (1311)	.496 (9598)
3. Goals	.180 (12,143)	.489 (1311)	—	.179 (17,575)
4. Performance	.448 (17,196)	.496 (9598)	.179 (17,575)	—

Note. ACT/SAT as ability is above the diagonal. High school GPA as ability is below the diagonal. Numbers in parentheses are the sample sizes. All correlations were from Robbins et al. (2004) and represent meta-analytic estimates that were fully corrected from sampling and measurement error, and range restriction. Harmonic mean for ACT/SAT as ability is 2143. Harmonic mean for high school GPA as ability is 999.

Table 2.

Correlation input matrix for the persistence model test

Variable	1	2	3	4
1. Ability	—	.283 (639)	.146 (6929)	.124 (3053)
2. Self-efficacy	.703 (7436)	—	.489 (1311)	.359 (6930)
3. Goals	.180 (12,143)	.489 (1311)	—	.340 (20,010)
4. Persistence	.246 (5551)	.359 (6930)	.340 (20,010)	—

Note. ACT/SAT as ability is above the diagonal. High school GPA as ability is below the diagonal. Numbers in parentheses are sample sizes. All correlations are from Robbins et al. (2004) and represent meta-analytic estimates that were fully corrected for sampling and measurement error, and range restriction. Harmonic mean for ACT/SAT as ability is 806. Harmonic mean for high school GPA as ability is 962.

2.3. Path-analytic procedures

We tested the four models using LISREL 8 (Joreskog & Sorbom, 1996). An additional decision that researchers employing meta-analytic estimates in path analyses and structural equation modeling must make is how to equate sample sizes within each cell in the input correlation matrix when each cell is based on meta-analyses employing different sample sizes (see, for example, the cell ns in Table 1 and Table 2). The most common solutions to holding sample sizes constant across cells have been to use either the arithmetic mean or the harmonic mean. We chose to use the harmonic mean because it tends to yield the least biased estimates of standard errors of parameter estimates (Viswesvaran & Ones, 1995). The formula for the harmonic mean is k/(1/N₁ + 1/N₂ + …….1/N_k), where k refers to the number of correlations in the matrix and N refers to the sample sizes in each cell. As the formula reveals, the harmonic mean also gives less weight to cells with large N’s and is, therefore, the more conservative of the two sample size equating methods. The footnotes to Table 1 and Table 2 indicate the harmonic means that we used in each of the four path analyses.

We judged model fit in each of the four analyses on the basis of three primary criteria—the Comparative Fit Index (CFI), Goodness of Fit Index (GFI) and the Standardized Root Mean Squared Residual (SRMR). We chose CFI and SRMR because past Monte Carlo studies (Hu & Bentler, 1999) of the behavior of different fit indices have revealed that a combination of CFI values greater than or equal to .96 and SRMR values of .10 or less never rejected a correct model. In addition, individually, CFI values exceeding .94 and SRMR values of less than .06 are generally considered to indicate excellent model fit, while CFI’s of .90 to .94 and SRMR’s of .06 to .10 are indicative of acceptable, but marginal fit (Joreskog & Sorbom, 1996). We also included the GFI as a third primary estimate of model fit because it is an index of absolute fit. The CFI, on the other hand, is an estimate of comparative fit (versus a null relations baseline model) and can yield high values because of a very poorly fitting baseline model rather than an adequately fitting tested model. Consistently high CFI and GFI values would suggest that the tested model adequately fit the data.

Although the normal theory weighted least squares χ² is often reported in structural equation model studies, we chose not to use this as a primary index of fit because it usually rejects well-fitting models with samples sizes as large as those employed in our model tests. The Root Mean Square Error of Approximation (RMSEA), another frequently employed index of model fit, is also very sensitive to large sample sizes and sometimes rejects well-fitting models because it is derived from the normal theory χ² rather than from the residual correlation matrix. The SRMR by contrast is calculated directly from the residual correlation matrix creating a situation where the residuals and SRMR can accurately suggest a good fitting model in the presence of an unacceptably high RMSEA (Browne, MacCallum, Kim, Anderson, & Glaser, 2002). Therefore, like the χ², the RMSEA was also considered as a secondary index of fit in this study.

3. Results

3.1. Academic performance model

Results of the path-analytic tests of SCCT’s academic performance model revealed that the fit of the model was somewhat better when ACT/SAT scores were used to operationalize prior ability/prior performance (CFI = .99, SRMR = .01, GFI = .99) than when high school GPA was used for this purpose (CFI = .94, SRMR = .05, GFI = .97). Chi square tests and RMSEA values associated with the former (ACT/SAT) model confirmed its excellent fit to the data [χ² (1) = .18, p = .68; RMSEA = .01], while those associated with the latter (high school GPA) model also supported the better fit of the general cognitive ability model versus the high school GPA model [χ² (1) = 69.55, p = .001; RMSEA = .26]. Inspection of the fitted residuals for the high school GPA model revealed them to be sufficiently small (mdn = .00, range = −.16 to +.01) to suggest, along with the CFI, SRMR, and GFI, that the GPA model fit the data sufficiently to allow us to report and interpret each model’s standardized path estimates (see Browne et al., 2002). These are displayed in Fig. 2. All coefficients are statistically different from 0 in both models, except for those estimating the strength of the paths from goals to performance.

Display Full Size version of this image (5K)

Fig. 2. Social Cognitive Career Theory Performance Model Tests. Standardized path coefficients obtained from the high school GPA model test are in parentheses.

We also tested for differences in standardized path estimates between the two (i.e., ACT/SAT versus high school GPA) models. This involved subtracting the two parameter estimates and dividing the result by the standard error of difference for the parameter estimates. The result is a z score, where a value of 1.96 or greater allows one to reject the null hypothesis of no difference between parameter estimates at p = .05.

These tests revealed that high school GPA contributed significantly more to the prediction of academic self-efficacy beliefs than did ACT/SAT scores (z = 15.16, p < .001), but that ACT/SAT scores had a stronger direct relation to college performance (GPA) than did high school GPA, z = 2.14, p = .032. Although the path coefficient from self-efficacy to academic performance was higher in the ACT/SAT model than in the high school GPA model, the difference in these path estimates did not reach conventional levels of statistical significance, z = 1.52, p = .12. Together, these results suggest that the “effects” of past academic performance on college performance is largely indirect via self-efficacy beliefs, while the influence of general cognitive ability, as measured by ACT/SAT scores, is more direct. Indeed, the total indirect effect of high school GPA (.70 × .39 = .27) on academic performance was higher than its direct effect (.18), while the opposite was true of ACT/SAT scores (.28 × .46 = .13 indirect effect vs. a .27 direct effect).

3.2. Academic persistence model

The fit of the two models of academic persistence (retention), one using ACT/SAT scores and the other using high school GPA as indices of aptitude/past performance, was largely consistent with the tests of fit of the two academic performance models. The fit of the ACT/SAT model was excellent [CFI = .99; SRMR = .01; GFI = .99; RMSEA = .01; χ² (1) = .07, p = .80], while that of the high school GPA model was more marginal [CFI = .93, SRMR = .05, GFI = .97; RMSEA = .26; χ²(1) = 66.97, p < .001] but interpretable (fitted residuals: mdn = .00, range = −.16 to .00). The standardized parameter estimates for the two models are displayed in Fig. 3.

Display Full Size version of this image (5K)

Fig. 3. Social Cognitive Career Theory persistence model tests. Standardized path coefficients obtained from the high school GPA model test are in parentheses.

There were several notable differences between the performance and persistence models. First, in contrast to the prediction of college grades, neither ACT/SAT scores nor high school GPA had a significant direct effect on retention. Rather, their effects on retention were more indirect via their relations to self-efficacy beliefs and goals, with high school GPA having somewhat greater total effects on persistence (.44) than ACT/SAT scores (.39). Second, goals related significantly to retention in both versions (i.e., using ACT/SAT or high school grades) of the persistence model, though goal-outcome paths were not significant in either version of the performance model. Third, there were no significant differences in the parameter estimates of the two variants of the persistence model, except that high school GPA again accounted for more variance in academic self-efficacy beliefs than did ACT/SAT scores, z = 15.16, p < .001, and the self-efficacy beliefs derived from these two sources of information were equally predictive of academic persistence.

4. Discussion

The results of the current study generally provided strong support for SCCT’s model of academic performance and persistence with one exception—there was a near zero path from goals to college GPA in the models of academic performance. SCCT hypothesizes that self-efficacy beliefs (and outcome expectations) lead to higher academic performance, in part, because persons with higher versus lower self-efficacy beliefs establish and work toward more challenging academic goals. The results of the current investigation supported the first part of this mediator hypothesis; namely, that self-efficacy beliefs are substantially related to academic goals. However, it did not appear that goals contributed to the grade point averages that students attained in college. Rather, the effect of self-efficacy on academic performance appeared to be more direct than being mediated by goal mechanisms.

There are at least two explanations for the latter findings. First, it is possible that goals actually contribute little to academic performance over and above other, more substantial contributors (e.g., general cognitive abilities, self-efficacy beliefs) and may actually provide somewhat redundant motivational information to students. Note, for instance, that the motivational properties of both self-efficacy beliefs (e.g., Bandura, 1997) and goals (e.g., Locke & Latham, 2002) are hypothesized to be identical—they each serve to facilitate approach versus avoidance behavior, promote sustained effort, and foster persistence in the face of difficulties. Thus, although self-efficacy beliefs do seem to lead students to set more challenging academic goals, it may be that the motivational properties of having robust academic self-efficacy are more important to academic attainment than are those associated with goals—or, at least, goals do not provide students with unique motivational incentives that are not already provided by self-efficacy beliefs.

The second explanation is more methodological than substantive and, we think, may be the more plausible reason why goals did not seem to contribute significantly to academic performance. Goals were primarily measured as intentions to complete college rather than as academic performance goals (e.g., goals to get an A in math courses), regardless of whether the criterion was college grades or retention, in the individual studies included in the Robbins et al. (2004) meta-analysis. Thus, goals in our performance meta-analyses did not match well with the outcome (GPA) they were intended to predict. It is also possible that they did not represent a high degree of challenge for many students and, thus, were of limited utility in motivating grade performance. These measurement issues may well have attenuated goal-GPA relations. The fact that goals predicted commensurate outcomes (i.e., retention) well in the tests of the persistence model supports this interpretation.

Several writers have observed that social cognitive variables tend to serve as more reliable predictors when they match criterion variables appropriately in terms of content, specificity, and other relevant dimensions (e.g., Lent & Brown, 2006). The fact that goals predicted a well-matched criterion (retention) better than they did a poorly-matched one (GPA) in our analyses is consistent with this interpretation. Thus, it would be premature to conclude from these findings that goals are not good predictors of academic performance. Future research could better test this relationship by ensuring closer correspondence between the operationalization of goals (e.g., level of grades to which students aspire) and that of the performance criterion (e.g., actual grades that students attain).

There were several other findings that may have important theoretical and practical implications. First, consistent with prior research in the higher education literature (e.g., Tinto, 2006) indices of academic aptitude (general cognitive ability and high school performance) had no direct relations to college retention outcomes. Rather, their influences on retention were largely indirect via their influences on self-efficacy beliefs and goals to complete college. This suggests that academically able students are no more likely to finish college than are less able ones unless they develop strong confidence in their college academic capabilities and robust goals for college completion. Other important non-ability influences on college completion rates (e.g., social integration, achievement motivation) have also been identified in the higher education literature ([Robbins et al., 2004] and [Tinto, 2006]), and our results simply support prior findings that academic aptitude, in and of itself, is not a particularly good predictor of who will and who will not finish college. It would be useful for future research to assess whether and how such other non-ability variables may complement self-efficacy and goals as predictors or determinants of college retention.

Second, we found that the two indices of academic aptitude (general cognitive ability and high school performance) related, as predicted by SCCT, to college performance, albeit via somewhat different mechanisms. Prior high school performance seemed to inform students’ self-efficacy beliefs to a greater degree than did general cognitive ability, but cognitive ability seemed to have a stronger direct relation with college performance than did past high school performance. The former finding is consistent with SCCT hypotheses that past performance accomplishments in similar contexts are the most potent sources of efficacy beliefs in academic and other contexts, and that their influences on performance are, at least partly, mediated by self-efficacy beliefs.

The practical implications of the cognitive ability, self-efficacy, and academic performance relations merit consideration. These findings suggest that students may not rely greatly on their performance on college entrance tests to inform their self-efficacy beliefs, even though the cognitive abilities measured by these tests may, according to our results, do a better job than past high school performance in predicting how well students will do in college. Although there are a number of reasons why performance on measures of cognitive ability may not greatly inform students’ self-efficacy beliefs (e.g., more experience with academic course work than with standardized tests, the bad press that admissions tests have received in recent years), our data along with a growing body of other research on the predictive power of college and graduate school admissions tests (e.g., [Kuncel et al., 2001] and [Kuncel et al., 2004]) suggest that some students may benefit if they are assisted to make greater use of their test performance in building their academic self-efficacy beliefs. For instance, counselors might help students, especially those who might have under-performed in high school, understand that college admissions tests do, in fact, validly measure the types of cognitive abilities important to college success, and that high cognitive abilities bode well for future college as well as occupational ([Schmidt, 2002] and [Schmidt and Hunter, 2004]) success.

In terms of limitations, because we used meta-analytic estimates to build the correlation matrices in this study, we could only include variables in the path-analytic tests for which meta-analytic estimates were available or could be derived. Thus, we were not able to test the full SCCT academic performance model because outcome expectations have received very limited research attention in the academic performance and retention literatures—despite the fact that this construct is central to social cognitive theories of work and academic performance and would be expected to add significant unique variance to the prediction of college performance and persistence outcomes. Outcome expectations, therefore, deserve more concerted research attention in future tests of Social Cognitive Career Theory’s performance hypotheses in the academic context.

Another caveat is that our approach to combining meta-analytic and path-analytic methods to test theoretical models is a relatively new development in the literature and has not received from statisticians and methodologists the same degree of investigation as have meta-analyses and path analyses individually. Although the two stage approach is being increasingly employed as a theory testing tool, future research might uncover some assumptive or statistical problems with the approach that have not yet been anticipated or studied.

Despite these limitations, the results of this study suggest that the SCCT model of academic performance and persistence may offer a valid framework for understanding the process by which students achieve academic success and persistence in college. It might, therefore, be used as a template for intervention efforts designed to promote college student performance and persistence. The study also suggested that the mechanisms by which high school performance and general cognitive abilities contribute to students’ academic performance may be somewhat different, with the former (high school performance) informing self-efficacy beliefs more than the latter, despite the fact that general cognitive ability seemed to contribute more substantially, beyond self-efficacy beliefs, to college student performance.

In light of these findings, further research might focus on explaining why performance on cognitive ability measures is not as good a source of efficacy information for college students than high school GPA. It may be that grades serve as a rough index of academic ability for students yet also convey additional data to them about how well they have been able to motivate themselves to achieve academically in relevant past situations (i.e., students may view their grades as an amalgamation of ability and motivational or self-regulatory information). An additional direction for research might be to explore whether interventions designed to foster better linkages between cognitive ability scores and academic self-efficacy can promote improved college success for students who have under-achieved in high school (i.e., test scores may offer credible information about future academic potential, despite spotty past performance). It would also be valuable to extend tests of the SCCT performance model to students of different academic levels (e.g., high school students) and cultural contexts, using longitudinal designs. Finally, it would be useful to test the model’s utility in predicting the performance of adult workers. In sum, more research is needed on factors that might moderate overall model fit and on interventions that might be derived from the model and the current results.

References

Bandura, 1997 A. Bandura, Self-efficacy: The exercise of control, Freeman, New York (1997).

Bandura and Locke, 2003 A. Bandura and E.A. Locke, Negative self-efficacy and goals revisited, Journal of Applied Psychology 88 (2003), pp. 87–99. Abstract | Abstract + References | PDF (89 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (87)

Browne et al., 2002 M.W. Browne, R.C. MacCallum, C.T. Kim, B.L. Anderson and R. Glaser, When fit indices and residuals are incompatible, Psychological Methods 7 (2002), pp. 403–421. Abstract | Abstract + References | PDF (168 K) | Full Text via CrossRef

Colquitt et al., 2000 J.A. Colquitt, J.A. LePine and R.A. Noe, Toward an integrative theory of training motivation: A meta-analytic path analysis of 20 years of research, Journal of Applied Psychology 85 (2000), pp. 678–707. Abstract | Abstract + References | PDF (2370 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (119)

Heggestad and Kanfer, 2005 E.D. Heggestad and R. Kanfer, The predictive validity of self-efficacy in training performance: Little more than past performance, Journal of Experimental Psychology: Applied 11 (2005), pp. 84–97. Abstract | PDF (215 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (3)

Heller et al., 2004 D. Heller, D. Watson and R. Ilies, The role of person versus situation in life satisfaction: A critical examination, Psychological Bulletin 130 (2004), pp. 574–600. Abstract | Full Text + Links | PDF (176 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (26)

Hu and Bentler, 1999 L. Hu and P.M. Bentler, Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives, Structural Equation Modeling 6 (1999), pp. 1–55.

Hunter and Schmidt, 1990 J.E. Hunter and F.L. Schmidt, Methods of meta-analysis: Correcting for error and bias in research findings, Sage, Newbury Park, CA (1990).

Ilies and Judge, 2003 R. Ilies and T.A. Judge, On the heritability of job satisfaction: The mediating role of personality, Journal of Applied Psychology 88 (2003), pp. 750–759. Abstract | Full Text + Links | PDF (239 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (8)

Joreskog and Sorbom, 1996 K.G. Joreskog and D. Sorbom, LISREL 8: User reference guide [computer software manual], Scientific Software International., Chicago, IL (1996).

Kuncel et al., 2001 N.R. Kuncel, S.A. Hezlett and D.S. Ones, A comprehensive meta-analysis of the predictive validity of the Graduate Record Examinations: Implications for graduate student selection and performance, Psychological Bulletin 127 (2001), pp. 162–181. Abstract | Abstract + References | PDF (2020 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (64)

Kuncel et al., 2004 N.R. Kuncel, S.A. Hezlett and D.S. Ones, Academic performance, career potential, creativity, and job performance: Can one construct predict them all, Journal of Personality and Social Psychology 86 (2004), pp. 148–161. Abstract | Full Text + Links | PDF (109 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (41)

Lent, 2005 R.W. Lent, A social cognitive view of career development and counseling. In: S.D. Brown and R.W. Lent, Editors, Career development and counseling: Putting theory and research to work, Wiley, New York (2005), pp. 101–127.

Lent and Brown, 2006 R.W. Lent and S.D. Brown, On conceptualizing and assessing social cognitive constructs in career research: A measurement guide, Journal of Career Assessment 14 (2006), pp. 12–35. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (4)

Lent et al., 1994 R.W. Lent, S.D. Brown and G. Hackett, Toward a unifying social cognitive theory of career and academic interest, choice, and performance, Journal of Vocational Behavior 45 (1994), pp. 79–122 [Monograph]. Abstract | PDF (2449 K)

Locke and Latham, 2002 E.A. Locke and G.P. Latham, Building a practically useful theory of goal setting and task motivation, American Psychologist 57 (2002), pp. 705–717. Abstract | Abstract + References | PDF (404 K) | Full Text via CrossRef

Multon et al., 1991 K.D. Multon, S.D. Brown and R.W. Lent, Relation of self-efficacy beliefs to academic outcomes: A meta-analytic investigation, Journal of Counseling Psychology 38 (1991), pp. 30–38. Abstract | PDF (740 K) | Full Text via CrossRef

Podaskoff et al., 1996 P.M. Podaskoff, S.B. MacKenzie and W.H. Bommer, Meta-analysis of the relationships between Kerr and Jermier’s substitutes for leadership and employee job attitudes, role perceptions, and performance, Journal of Applied Psychology 81 (1996), pp. 380–399.

Robbins et al., 2004 S.B. Robbins, K. Lauver, H. Le, D. Davis and R. Langley, Do psychosocial and study skill factors predict college outcomes A meta-analysis, Psychological Bulletin 130 (2004), pp. 261–288. Abstract | Full Text + Links | PDF (177 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (36)

Rottinghaus et al., 2003 P.J. Rottinghaus, L.M. Larson and F.H. Borgen, The relation of self-efficacy and interests: A meta-analysis of 60 samples, Journal of Vocational Behavior 62 (2003), pp. 221–236. SummaryPlus | Full Text + Links | PDF (127 K) | View Record in Scopus | Cited By in Scopus (12)

Schmidt, 2002 F.L. Schmidt, The role of general cognitive ability and job performance: Why there cannot be debate, Human Performance 15 (2002), pp. 187–210. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (29)

Schmidt and Hunter, 2004 F.L. Schmidt and J. Hunter, General mental ability in the world of work: Occupational attainment and job performance, Journal of Personality and Social Psychology 86 (2004), pp. 162–173. Abstract | Full Text + Links | PDF (92 K) | Full Text via CrossRef

Sheu et al., 2006 Sheu, H. B., Miller, M., Lent, R. W., Brown, S. D., Hennessy, K., & Duffy, R. (2006). The social cognitive choice model: Comparison of two meta-analytic approaches. Presented in the symposium Testing social cognitive career theory with diverse methodological tools, American Psychological Association, New Orleans, LA.

Swanson et al., 2000 J.L. Swanson and P.A. Gore Jr., Advances in vocational psychology theory and research. In: S.D. Brown and R.W. Lent, Editors, Handbook of counseling psychology (3rd ed.), Wiley, New York (2000), pp. 233–269.

Tinto, 2006 V. Tinto, Research and practice in student retention: What next, Journal of College Student Retention: Research, Theory, and Practice 8 (2006), pp. 1–19. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (0)

Vancouver et al., 2001 J.B. Vancouver, C.M. Thompson and A.A. Williams, The changing signs in the relationships between self-efficacy, personal goals, and performance, Journal of Applied Psychology 86 (2001), pp. 605–620. Abstract | Abstract + References | PDF (1506 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (29)

Viswesvaran and Ones, 1995 C. Viswesvaran and D.S. Ones, Theory testing: Combining psychometric meta- analysis and structural equation modeling, Personnel Psychology 48 (1995), pp. 865–885. Full Text via CrossRef

Portions of this paper were presented in the symposium, Testing Social Cognitive Theory with Diverse Methodological Tools (Robert W. Lent, Chair) at the annual meetings of the American Psychological Association, New Orleans, LA, August 13, 2006. We thank Nicholas Joyce, Justin Li, and Drs. Fred Bryant and Matthew Miller for their assistance with this research.
Corresponding author. Fax: +1 312 915 6660.