Development and Validation of the Academic Persistence Outcome Expectations Scale

Abstract

Outcome expectations, an integral theoretical component of social cognitive career theory, remains almost completely unexamined in the domain of academic persistence, or the decision a student makes to remain in college. This study sought to develop a theoretically derived scale to measure outcome expectations. An initial item pool was developed and sent to a sample of college students. A second, confirmatory sample of undergraduate students was collected via an online crowdsourcing platform. Results suggested the presence of a two-factor structure was the most parsimonious solution across both samples. The two factors retained across both samples reflected positive and negative outcome expectations that students perceived about remaining in college for the year. Limitations and implications are discussed.

Keywords

social cognitive career theory outcome expectations self-efficacy academic persistence retention

For every ten students that begin college in the United States, seven will return for a second year and five will complete a bachelor’s degree in five years (ACT, 2017; NCS, 2016). For students who begin college and do not complete it, money spent on education that did not lead to any kind of meaningful credential and lost wages associated with limited participation in the workforce represent a real economic cost. Though some students gain the skills necessary for competitive employment without completing a credential, many students leave institutions without one end up worse off financially than when they started (Schneider & Yin, 2011). Those who complete a bachelor’s degree stand to earn a substantially higher income than those who do not (Autor, 2014) and though good paying jobs without a bachelor’s degree exist, often some kind of formal post-secondary training (i.e., an associate’s degree) is necessary to access these jobs (Carnevale et al., 2018).

This problem is far from new. In fact, what is remarkable is how relatively stable the number of departing students have been over time (Habley et al., 2012; Pantages & Creedon, 1978; Summerskill, 1962). Given the costs to the student and society writ large that are associated with going to college and departing prior to completing a credential, scholarship to identify the underlying mechanisms of this behavior becomes critical. To this end, academic persistence has emerged as a construct to understand the personal variables associated with remaining in college or persisting to graduation in the social science literature (Melguizo, 2011). In psychological science, academic persistence is broadly conceptualized as an observed outcome anticipated by motivational constructs such as identity, self-determination, or expectancy beliefs interacting with contextual and environmental variables (Brown et al., 2008; Chen & Jang, 2010; Destin & Williams, 2020; Schunk & DiBenedetto, 2020). One of the more prominent theories within the vocational and educational literature, Social Cognitive Career Theory (Lent & Brown, 2013; Lent et al., 1994, 2000), has largely conceptualized academic persistence for all college majors through tests of the performance model where key constructs such as self-efficacy, outcome expectations, and goals are informed by prior performances to predict persistence across time (Brown et al., 2008). Overall, key assumptions of the model have been supported in the domain of academic persistence (Bolkan et al., 2021; Brown et al., 2008; Lee et al., 2015; Kahn & Nauta, 2001; Wright et al., 2013a). Tempering these supportive results, however, is that they have mostly focused on the role of self-efficacy with less attention paid to the role of outcome expectations (Brown et al., 2008; Fouad & Guillen, 2006; Lent & Brown, 2019). Thus, the present study seeks to develop a theoretically supported factor structure to measure outcome expectations in the situation of academic persistence.

Theoretical Background

The parent theory to SCCT, Social Cognitive Theory (Bandura, 1986), posits that human behavior is not the product of internal unidirectional determinants but exists through the triadic and reciprocal interaction of the person, the environment, and the person’s behavior via key cognitive processes capable of forecasting actions and consequents based on consistent feedback from the environment (Bandura, 1986, 2001). A central construct to social cognitive theory is self-efficacy (Bandura, 1977, 1986, 1997, 2001). Self-efficacy is representative of human agency and proposed to be a causal determinant to human agency, action, achievement, and well-being (Bandura, 1997, 2001). Self-efficacy is not a global trait, but rather a future oriented belief that individuals hold about their capacity in each situation that vary in terms of their level, strength, and generality (Bandura, 1977, 1997).

Outcome expectations are defined as a co-determinant of behavior with self-efficacy playing a central role in human forethought and agentic action (Bandura, 2001). They are future oriented judgments about the consequences of engaging in each behavior or set of behaviors that can be proximal or distal to the behavior or performance (Bandura, 1986, 1997, 2001; Lent et al., 1994). These forecasting judgments can be either distinctly positive or negative expectancies and are thought to conform to three distinct subtypes: physical, social, and self-evaluative (Bandura, 1997; Lent et al., 1994). Physical outcome expectations refer to experiences one believes that a certain action is likely to experientially cause. Social outcome expectations refer to social effects that a certain behavior may evoke. Self-evaluative outcome expectations refer to how an individual will evaluate their own performance affectively (Bandura, 1986; Fouad & Guillen, 2006). Additionally, outcome expectations are often regarded as consequents in relation to self-efficacy in predicting behavior, however, this point has been sharply debated as some empirical evidence exists that implies the contrary (Kirsch, 1995; Williams, 2010). For the purposes of the present research, ambiguity on this point emphasizes the need for a psychometrically robust instrument.

Academic Persistence within the SCCT Performance Model

Within SCCT, the performance model conceptualizes academic persistence broadly as a performance itself. Theoretically, the tenets of the performance model equally apply to academic persistence and academic performance. There are two propositions for this model (propositions #8 and #9 within the original SCCT framework). The first proposes that self-efficacy beliefs both directly and indirectly, through performance goals, influence performance. Outcome expectations influence performance only indirectly through goals. The second states that previous ability or aptitude has a direct effect on performance and indirect effect through self-efficacy beliefs (Lent et al., 1994). Regarding empirical support, Brown et al. (2008) demonstrated in their meta-analytic work that measurements of prior performance accomplishments such as GPA, SAT/ACT scores predict student retention through the self-efficacy and goals, noting that there was insufficient evidence to model outcome expectations.

Yet very few studies have investigated the model with outcome expectations included using the SCCT performance model since the original theoretical articulation (Lent et al., 1994). Kahn and Nauta (2001) administered measures of self-efficacy, outcome expectations, and goals to four hundred college freshmen at two time points and tracked their freshman to sophomore persistence demonstrating that outcome expectations and performance goals predicted persistence administered in the second semester of the freshman year into the sophomore year above prior performance accomplishments. Outcome expectations in this model were measured by three items derived from previous research (Bean, 1985) where students were asked to rate the utility of graduation on various distal outcomes such as work and income on a four-point Likert type scale. More recently, Bolkan and colleagues (2021) tested a modified version of the performance model. In their study, contextual influences, such as academic advising, took the place of prior performance accomplishments in predicting the roles of self-efficacy, outcome expectations, and goals on four-year graduation rates in a sample of college students after covarying for race, gender, major changes, and prior performance accomplishments. The resulting model supported a direct path between academic advising and graduation rates, a direct path between self-efficacy and graduation rates, and indirect paths to four-year graduation rates through self-efficacy and goals. Notably, outcome expectations did not predict goal setting behavior in this study as would be theoretically expected. Outcome expectations here were measured by a five-item scale, which asked students to rate the prospects of graduating in four years as “Important”, “Worthwhile”, “Positive”, “Wise”, and “Good” (McCroskey & Richmond, 1996).

Outcome Expectations Scales in Persistence Intentions Outside of the Performance Model

Research on academic persistence within the original performance model may be scarce is due to the study of persistence intentions in the context of other SCCT models, most notably those focused on predicting the decision engineering students make to remain in their majors (e.g., Flores et al., 2021; Lent et al., 2013; 2016). Yet in some of these model tests, the complex theoretical structure of outcome expectations remains unrepresented thus providing limited information about which outcome expectations are particularly influential or benign (Fouad & Guillen, 2006). Indeed, Lent and Brown (2019) acknowledged this limitation in outcome expectations measurement in their recent review of meta-analytic findings noting that outcome expectations scales tend to be written to only capture positive outcome expectations to the exclusion of other types of expectations that Bandura (1986) had theorized since, in this case, it could be reasoned that different facets of outcome expectations might influence academic persistence asymmetrically. In other words, negative outcome expectations may be more powerful in predicting lack of academic persistence than positive outcome expectations are in predicting persistence in college, or some types of expectations (social, physical, or self-expectations) are more important than others.

A few exceptions are noteworthy. First, Lee et al. (2018) found multidimensional support for a 21-item outcome expectations instrument assessing negative outcome expectations in engineering. The retained factors were: cultural-related stressors, personal life and work balance, job characteristics, and social costs. Here, these factors represented interrelated domains of outcome expectations rather than the structure Bandura (1986) hypothesized. Secondly, Byars-Winston et al. (2016) fit a CFA to a modification of a commonly utilized outcome expectations scale in persistence intentions research (Lent et al., 2005) with a large sample of racially diverse science majors. They modified the scale to include five items that specifically measured physical, social, and self-evaluative outcome expectations in a better effort to capture all of the theorized areas of consequence. Results here supported a unidimensional structure in the entire sample, but statistical fit of the correlated factors CFA model greatly diminished in group invariance testing with women of color. Finally, Lent et al., (2013) modified their own original scale (Lent et al., 2003) with additional items given the scale had “typically correlated as expected with measures of self-efficacy, interests, and major choice goals in prior research, it has not always accounted for unique predictive variance in multivariate analyses (p. 25).” Results of an exploratory factor analysis supported a two-factor structure for twelve modified items, which were labeled extrinsic and intrinsic outcome expectations. Thus, the results of these three different scales measuring outcome expectations in the same or very similar domain yield conflicting information about the underlying factor structure of outcome expectations. Yet if the measure conforms to a theoretical structure, it should be reproducible within the same domain (Messick, 1995). However, this has not been demonstrated empirically with current persistence outcome expectations scales.

Current Study

It is clear from the extant literature that academic persistence outcome expectations is a construct that appears to be poorly measured in academic persistence research despite its rich theoretical structure (Bandura, 1986) and the history of strong empirical support of SCCT in general in academic persistence research (Brown et al., 2008; Lent & Brown, 2019). Studies have led to differing conceptualizations of outcome expectations scales in academic persistence that may affect inferences drawn from model tests of the performance model in academic persistence research (Bolken et al., 2021; Kahn & Nauta, 2001; Lent et al., 2005; 2013). Further, measurement models that have been suggested in the related area of persistence intentions towards remaining in a specific college major have captured outcome expectations only in part and thus may not reflect the entire theoretical range of outcome expectations, leaving their structural validity in question (Fouad & Guillen, 2006; Lent & Brown, 2019; Messick, 1995). In light of this, the goals of the project are to investigate the original factor structure, as theorized by Bandura (1986) and retained by Lent et al. (1994), of outcome expectations in this domain and to develop a theoretically useful instrument for research and practice in assessing outcome expectations in academic persistence.

Study 1 Method

Participants

A sample of undergraduate students at a large Midwestern university in an urban setting (n = 216) was recruited to pilot a new measure of outcome expectations. Students in the sample were enrolled in undergraduate coursework in a department of educational psychology housed in the school of education. The mean age was 21 years old. Most of the sample (n = 175) identified as cisgender women while a minority (n = 37) of participants identified as cisgender men and few (n = 4) participants identified as transgender, nonbinary, and/or genderqueer. The racial background for much of the sample was White (n = 128) followed by those endorsing two or more racial backgrounds (n = 33), Latinx/Hispanic (n = 27), Black American (n = 12), Asian American/Pacific Islander (n = 11), and Native American (n = 5). Generational status in college was measured by a parents’ completed education as reported by our participants. If both parents were reported to possess less than a bachelor’s degree, then the student was a first-generation college student. By this metric, a small subsample of students (n = 72) were first generation college students. Academic standing was also measured by self-report. Here, 60 students were freshman, 42 students were sophomore, 57 were juniors, and 55 were seniors. Two students did not disclose their academic standing. Many participants identified as coming from the middle class (n = 132) or the working class (n = 64). The remaining students endorsed either an impoverished (n =11) or affluent (n = 9) background.

Procedures

The protocol was approved by the institutional IRB. Items were initially generated by the first author after content review of theoretical work on the outcome expectations construct (Bandura, 1977; 1986; 1997; Lent & Brown, 2006) and data collected from a diverse group of 10–15 college students participating in two focus groups about their expectations for persisting in college as a part of an ongoing program evaluation of major exploration coursework (Schams et al., 2022). Students were asked how they would feel if they remained in college, what others would think if they were to choose a major or remain in college, what rewards or losses they anticipate if they remained in college, and what they would think of themselves if they remained in college. Students were allowed to respond freely. This process revealed the following themes across the two focus groups: “college will put me “on-track” with my age cohort,” “through college I fulfill societal expectations,” “college is means to an employment and monetary outcome,” and tacit social pressure to choose a major as means of progressing towards graduation.

Items were written and then reviewed by two vocational psychologists—one of whom had published influential papers in the area of outcome expectations and another with general familiarity with the construct. There was no disagreement between the two content experts that the items measured outcome expectations, only disagreement about the class of outcome expectation, most notably between physical and self-evaluative outcome expectations. The initial item set was retained in light of the slight mismatch in expertise though additional items were then written and reviewed to better capture this distinction. The entire pool of items was then sent to the sample of students who completed the item pool and a brief demographic questionnaire in exchange for payment. Students in this sample had the option of receiving either a gift card to a popular online store or a small credit direct deposited into a private account for participation. Participants were more likely to select the gift card over the credit (p = .85). Qualtrics software was used to administer the survey electronically. The scale items were presented in random order to minimize order effects. The item pool contained negatively worded items to both capture negative outcome expectations and minimize acquiescence bias. The instructions read as follows: “Below are some statements that may reflect your expectations of what will happen to you if you choose to remain in college. Please indicate how strongly you agree/disagree with the following set of statements by using the following five-point scale.” The scale anchors were “Disagree,” “Somewhat Disagree,” “Not Sure” “Somewhat Agree” and “Agree.” The question stem read: “If I decide to stay in college this year…” and then the respondent was presented with a series of items (Lent & Brown, 2006).

Plan for Analysis

The general plan for analysis at this stage of scale development was to conduct an exploratory factor analysis to ascertain the latent factor structure of the items to establish construct validity. Data were prepared using the jamovi software program (The jamovi project, 2022), an R-based software program available for statistical analyses (www.jamovi.org) and then analyzed in Mplus 8.3 (Muthén & Muthén, 2009-2017). Statistical power was assessed post-hoc in R (version 4.1.3; R Core Team, 2022) using the semPower package (Moshagen & Erdfelder, 2016).

For factor models, a geomin rotation was employed with 100 random starts (Hattori et al., 2017). Likert-type items are treated continuously for all analyses (Robitzsch, 2020). For guidance in factor retention, we relied on parallel analysis (Lim & Jahng, 2019). For items to be retained, we employed guidance suggested by Marsh et al. (2005), namely that items will have a large loading on a given target factor and with a relatively small cross-loading on a non-target factor and low correlated residuals as indicated by modification indices. Additionally, we required any retained factor to be greater than three items (Muthén & Muthén, 2009). Reliability was judged by evaluations of Cronbach’s alpha and McDonald’s omega total as indicators of scale’s internal consistency. The latter has been suggested by various methodologists as a more accurate estimate of reliability under certain conditions (McNeish, 2018; Raykov & Marcoulides, 2019).

After an exploratory model is retained, we fit a confirmatory model to ensure our solution is independent of the rotational strategy and to obtain bootstrapped parameter estimates. For all models, model fit was judged using previously established combination rules for fit statistics. The fit statistics we examined to aid us in model retention included RMSEA ≤ 0.08, CFI and TLI ≥ .90, and SRMR < 0.08. These are not “golden rules,” but guidelines to judge model fit (Hu & Bentler, 1999; Marsh et al., 2004). The model chi-square was also considered, though not exclusively, due to its known vulnerabilities to sample size.

Study 1 Results

Negatively worded items were reversed coded and all items were then evaluated for difficulty and discrimination. Two items that discriminated poorly with the overall construct were removed from further analysis (items “I will feel happy” and “I will be doing what society expects of me”). Two additional items were removed for content redundancies about feelings of accomplishment. Table 1 contains the remaining items. Here, parallel analysis suggested a four-factor structure though several items contained comparatively low item-loadings represented on their target factors. Additionally, model fit for the four-factor structure was poor (M1, Table 2). Here, we employed trimming criterion in which relatively high loadings on target factors (λ ≥ 0.50) and with relatively low cross-loadings (λ ≤ 0.40) were retained. The resulting trimming procedure retained 16 items on two factors, which was supported by a second parallel analysis procedure and model fit statistics (M2, Table 2). The first factor was negative outcome expectations (α = 0.80) students perceived for their decision to remain in college during the academic year while the second factor represented positive outcome expectations (α = 0.85). Omega reliability statistics were also within the acceptable range (Table 2). The geomin rotated factor correlation was statistically significant in the expected direction (r = 0.27, p < 0.05) after accounting for reversed coding.

Table 1.

Outcome Expectations Item Pool.

Item Description
1. I will be worried about missed opportunities elsewhere
2. I will feel anxious
3. I will feel depressed
4. I would feel as if I have accomplished something
5. I will be on track with my peers
6. I will be able to find a good job
7. I will be able to meet my needs
8. My parents will be proud of me
9. My friends will respect me
10. I will make less money
11. I will have no time to do anything else
12. I will not have enough money to pay my bills
13. I will be wasting my time because the economy does not support a college education anymore
14. I will be proud of myself
15. I will feel like I’m going in the right direction with my career
16. I will reach my potential
17. I will be able to pay for my lifestyle down the road
18. I will get the job I want
19. I will have to take on another job
20. I will not be able to see my friends as often as I like
21. I will feel impressed with myself
22. I will not have enough time to do what I want to do
23. I will feel as though I am consistently behind
24. I will be closer to beginning my career
25. I will feel like I belong somewhere
26. I will have hope for a better future
27. I will not be able to manage all of my responsibilities
28. I will be able to be financially independent
29. I will feel excited
30. I will feel afraid for my future
31. I will feel stressed
32. I will feel happy
33. My friends will be proud of me
34. I will lose contact with someone I care about
35. I will feel proud of myself
36. I will feel like my career is moving forward
37. I will feel optimistic about my future
38. I will feel like I have purpose
39. I will be doing what society expects of me
40. I will be letting my family down

Note. The item stem reads: “If I choose to stay in college this year…” The two poorly discriminating items and the two redundant items were removed from the table to ease interpretation.

Table 2.

Summary table of goodness of fit statistics for factor analytic models.

Study	Model	χ2	CFI	TLI	RMSEA	90% CI RMSEA	SRMR
1	M1. EFA	1393.26	0.76	0.71	0.06	0.06 – 0.07	0.04
	M2. EFA Red.	180.08	0.91	0.89	0.05	0.04 – 0.07	0.04
	M3. S1 CFA	188.36	0.92	0.91	0.06	0.04 – 0.07	0.06
2	M4. S2 CFA	175.75	0.94	0.93	0.04	0.03 – 0.06	0.05

Note. All chi-square models are statistically significant (p ≤ 0.01).

A confirmatory factor analysis was then fit to the reduced item set to ensure the result was not excessively dependent on the rotational strategy and to obtain bootstrapped parameter estimates in the first sample. Items were restricted to load only on one factor in the two-factor solution. The estimation method was full information maximum likelihood with fixed factor variances. Correlated errors were not allowed in this model. The resulting model was a good fit to our data thus supporting the presence of a two-factor structure (M3, Table 2). Factor correlations between positive outcome expectations and negative outcome expectations was statistically significant in the expected direction (r = 0.35, p ≤ 0.01). A post-hoc power analysis revealed that adequate statistical power was obtained in this analysis.

Study 2 Method

To confirm our findings from the first sample, we sought to validate the factor structure from Study 1 first by conducting a second confirmatory analysis in another sample with measures theoretically related to the construct.

Participants

The protocol for this study was approved by the institution’s IRB. The second sample was collected from an online crowdsourcing platform known as Prolific Academic (https://prolific.ac/). Prolific Academic offers the option to screen participants based on demographic information provided to the company rather than relying on information provided to the researchers by the participants themselves. When a survey is released on Prolific, only participants that meet the study’s eligibility criteria can view it. This provides some assurance that the participant pool matches the demographic of interest (Palan & Schitter, 2018). Research has supported Prolific Academic as a source for obtaining quality data comparable to its competitors (Peer et al., 2017). As Prolific Academic insists that a fair wage be paid to all workers, participants in this present study were reimbursed at the rate of $6.54/hour. The average completion time was 9 minutes.

Data from 301 participants was collected. To be eligible for this study, participants must have been registered to complete surveys in exchange for payment on the Prolific platform. Participants must have reported to Prolific that they were based in the United States, were currently students enrolled in a university, and were between the ages 18–22. The last criterion was necessary because a filter that distinguished whether a participant was a non-traditional undergraduate student, graduate student, or a full or part-time student did not exist on Prolific during data collection. Participants were also asked to indicate whether they were enrolled in a post-secondary institution, if they planned to continue their studies in the next semester, their current major, and how they are receiving their instruction (e.g., online vs. in-person education). Participants must have been enrolled in a post-secondary institution and have planned on attending school next semester to be eligible to control for otherwise naturally departing students. Ineligible participants received payment for participation.

The mean age in this sample was 21 years old. Regarding gender identity, the majority identified as cisgender men (n = 160), followed by cisgender women (n = 122), transgender (n = 10), and genderqueer (n = 1). Most of the sample was White (n = 157) followed by Asian American (n = 68), Latinx/Hispanic (n = 39), individuals of mixed race (n = 38), Black American (n = 28), and Native American (n = 9). Regarding academic standing, 81 were freshman, 89 students were sophomores, 74 were juniors, and 57 were seniors. A majority of this sample also identified as coming from the middle class (n = 154) or the working class (n = 90). The remaining students endorsed either coming from impoverished backgrounds (n = 35) or affluent ones (n = 22). Most were enrolled in a four-year college or university (n = 254) or community college (n = 40). The remaining students were in either in a technical school or military academy (n = 7). Among those who had declared their major, college majors were diversely represented by business, humanities (e.g., Film, Music), helping professions (e.g., Social Work, Education), and STEM (e.g., Computer Science, Mathematics). Most students had taken at least one course online (n = 190) with a minority of students completing their coursework fully online (n = 14).

Measures

College Self-Efficacy

The College Self-Efficacy Inventory (CESI; Solberg et al., 1993) consists of 20 items that load on three subscales considered to be central to a student’s experience in college. These are course self-efficacy, social self-efficacy, and roommate self-efficacy. The stem asks participants to rate their confidence on certain tasks in social and educational domains. Example items include: “participate in class discussions” and “making new friends in college.” Higher scores on the measure reflect greater self-efficacy in the behaviors thought to exist in the general domain of attending college. For this study, only the course and social self-efficacy subscales will be used to capture the experiences of students who may not be living on campus or with a roommate. Subscales for course self-efficacy (α = 0.81) and social self-efficacy (α = 0.82) were internally consistent in this sample.

College Outcome Expectations

The College Outcome Expectations Scale (COE) is a measure developed by Flores et al. (2008) to measure the outcome expectations an individual has for a college education. This measure was chosen because it most closely resembles the measure under construction in this study though differs in focus (e.g., expectancies for college experiences vs. expectancies for persisting in college). The overall total score was internally consistent in this sample (α = 0.93).

Academic Goal Progress

Progress towards academic goals was measured by the Academic Goals Scale (Lent et al., 2005), which was a scale originally developed by Lent et al. (2003) and then refined by Lent et al. (2005). The scale prompt is: “How much progress do you think you are making toward each of the following goals at this point in time?” followed by seven items each listing an academic goal. The response format is a 1-5 Likert scale with higher scores being indicative of progress towards goals. Studies that have employed this scale in model tests have reported good to excellent observations of internal consistency (.84 - .90; Lent et al., 2005, 2007). For this study, internal consistency was high for the total score (α = 0.91).

Academic Persistence Outcome Expectations

The Academic Persistence Outcome Expectations Scale was developed in Study 1 and contains two hypothesized factors: positive outcome expectations and negative outcome expectations. It contains sixteen items broadly capturing two dimensions of outcome expectations. For this study, internal consistency was within the acceptable range on both positive (α = 0.88) and negative (α = 0.78) outcome expectations.

Plan for Analysis

At this stage of scale development, the overall goal was to test the factor structure with confirmatory factor analysis and to examine corollary relationships with previously established measures. Data were analyzed in Mplus and jamovi, under the same conditions as Study 1. This included reverse coding the negatively worded items prior to analysis and using the same model fit indices as specified in study 1 as well as the same estimator method without the use of correlated errors. Correlations between the hypothesized factors and established measures were also examined.

Study 2 Results

Results of the confirmatory factor analysis on a two-factor structure suggested the model was a good fit to the data (M4, Table 3). Inspection of the loadings in the factor structure revealed that one item previously retained in the negative outcome expectations factor in study 1 loaded comparatively poorly in study 2. For all other items, loadings were within the confidence interval reported in Study 1. Correlations between the two subscales and previously established measures are in Table 4. Here, we observed correlations to be in the expected directions theoretically. For example, positive academic persistence outcome expectations correlated with course self-efficacy (r = 0.40, p ≤ 0.05) to the same magnitude as college outcome expectations with which it was also strongly correlated (r = 0.78, p ≤ 0.05). Additionally, negative academic persistence outcome expectations correlated relatively strongly with progress on academic goals (r = 0.42, p ≤ 0.05) compared to positive academic persistence outcome expectations (r = 0.34, p ≤ 0.05) and college outcome expectations (r = 0.37, p ≤ 0.05) potentially signaling that negative outcome expectations may be slightly more robust in this domain.

Table 3.

Standardized factor loadings for retained models, by study.

Factor	Item	Study 1			Study 2
Factor	Item	EFA	CFA	95% CI	CFA	95% CI
Negative APOE (ω = 0.77-0.81)	I will feel anxious	0.56	0.59	0.41–0.75	0.61	0.48–0.74
	I will feel depressed	0.59	0.60	0.45–0.74	0.68	0.58–0.80
	I will have no time to do anything else	0.65	0.62	0.46–0.76	0.59	0.45–0.72
	I will not have enough money to pay my bills	0.56	0.55	0.40–0.68	0.49	0.35–0.63
	I will not be able to see my friends as often as I like	0.50	0.50	0.31–0.67	0.29	0.12–0.46
	I will feel as though I am consistently behind	0.70	0.69	0.56–0.82	0.65	0.49–0.80
	I will not be able to manage all my responsibilities	0.60	0.62	0.47–0.76	0.52	0.36–0.68
	I will feel stressed	0.56	0.55	0.39–0.72	0.58	0.46–0.71
Positive APOE (ω = 0.85-0.88)	I will feel like I’m going in the right direction with my career	0.57	0.58	0.36–0.80	0.80	0.72–0.88
	I will be closer to beginning my career	0.66	0.64	0.45–0.83	0.71	0.60–0.82
	I will have hope for a better future	0.52	0.52	0.29–0.76	0.76	0.67–0.85
	I will feel like my career is moving forward	0.85	0.84	0.74–0.95	0.65	0.54–0.77
	I will feel optimistic about my future	0.69	0.70	0.47–0.94	0.55	0.42–0.68
	I will feel like I have purpose	0.79	0.79	0.65–0.93	0.71	0.57–0.84
	I will feel excited	0.61	0.62	0.41–0.83	0.75	0.62–0.87
	My friends will be proud of me	0.67	0.66	0.49–0.83	0.64	0.52–0.76

Table 4.

Correlations between APOE and theoretically relevant measures.

	1	2	3	4	5	6
1. Course Self efficacy	1
2. Social Self-efficacy	0.54	1
3. College outcome expectations	0.40	0.35	1
4. Progress on academic goals	0.59	0.24	0.34	1
5. Negative academic persistence outcome expectations	0.32	0.16	0.37	0.42	1
6. Positive academic persistence outcome expectations	0.40	0.37	0.78	0.34	0.41	1

Note. All correlations are statistically significant (p ≤ 0.05).

Discussion

This study aimed to examine the factor structure of outcome expectations and to provide a useful scale to help standardized findings across studies and eventually allowing for meta-analytic inference for the utility of the performance model in studying academic persistence (Brown et al., 2008; Fouad & Guillen, 2006). To this end, an overinclusive item pool was developed and rated by two psychologists familiar with the construct. The item pool was sent out to a sample of university college students at a local university. What was found was the presence of a two-factor structure for outcome expectations that corresponds to positive and negative outcome expectations (Lent & Brown, 2006). It seems that in the domain of academic persistence, students’ expectancy beliefs most strongly conform to the anticipated gain (positive outcome expectation) or loss (negative outcome expectation) when prompted to think about what will happen if they attend college for a given year.

The strength of the present research is that it creates a useful tool to measure outcome expectations in academic persistence by providing a structurally valid scale that has been validated on a wide variety of college students in a nationwide sample. This study identified two factors that are worth further psychometric exploration in response to recent calls for outcome expectations scales to measure both positive and negative factors (Lent & Brown, 2019). The APOE is a correlated-factors model thus one or more subscale can be used together or separately depending on the goal of the research. If researchers are to use the instrument, we advise first fitting an item level CFA and reporting reliability statistics, ideally McDonald’s Omega in light of differences in factor loadings observed across both samples. In addition to statistical fit, corollary evidence from the second sample demonstrated good convergent validity with college self-efficacy and goals and, in the case of positive outcome expectations, good concurrent validity with college outcome expectations thus providing further support for the validity of both subscales.

Our other aim was to examine the factor structure of outcome expectations in this domain. Here, we retained a two-factor structure capturing the positive and negative types of outcome expectations beliefs in academic persistence (Bandura, 1986; 1997; Lent & Brown, 2006). This finding converges with research in other outcome expectations scales in SCCT (Anderson et al., 2016) as well as the greater SCT paradigm (Resnick, 2005; Rollnick et al., 1996) when researchers attempt to model both the rewards and losses participants anticipate in light of a given action. These findings do broadly support the theoretical structure of outcome expectations given the two separate factors do reflect the two types of outcome expectations that are theorized (Bandura, 1986, 1997). Yet what remains in question is the subclass structure of outcome expectations since, theoretically, it would be expected these three classes would be both distinctly represented, which was not the case here. Though some studies have found a structure that conforms to theoretical subclass structure (Byars-Winston et al., 2016; McAuley et al., 2010; Wójcicki et al., 2009) we note that these tend to capture mostly positive outcomes. In fact, we are not aware of other studies that have attempted to model both the types and the subclass structure of outcome expectations. Other studies that have modeled one type of outcome expectations have retained either unidimensional structures (e.g., Bieschke, 2000; Oliveira et al., 2016) or structures capturing related domains in outcome expectations (e.g., Lee et al., 2018; Wright et al., 2013b).

There might be a variety of reasons to explain these findings. Outcome expectations are typically not regarded to be as important as self-efficacy within self-efficacy theory (Bandura, 1997). Thus, little theoretical guidance has been offered in terms of exactly what constitutes a different class of outcome expectations. This could lead to widespread interpretation in terms of what constitutes a given class of outcome expectations. For instance, consider monetary value. Money can be perceived as a social outcome expectation due to its association with social benefits. It also may be considered a physical outcome expectation because of the experiential pleasure that is potentially associated with earning a reward. If one were to write two items (e.g., “Money will allow me to live the life I want”, “If I have a lot of money, my friends and family will approve of my lifestyle”) to capture this, it is plausible these items could be highly correlated with each other since affording a desired a lifestyle with associated access to social benefits can be considered earning a reward. Because a respondent would be thinking of money when interacting with either item, it would make sense that both items are better thought of as capturing the similar anticipated reward (or positive outcome expectation). If so, this might have theoretical implications if the subclasses (physical, social, and self-evaluative) are thought to contain both positive and negative outcome beliefs (Bandura, 1997) since it seems psychometric evidence thus far might better support a structure where the subclasses are represented at the item level on positive and negative outcome expectations factors.

Limitations and Future Directions

There are several limitations about our effort here. First, to guarantee a college student population through Prolific Academic, it was necessary to restrict the sample to traditionally aged college students who were currently attending a full-time university. This effectively limits the validity test to traditionally aged full-time college students rather than capturing the experiences of non-traditional college students in the second sample. This could be one explanation for observed differences in some factor loadings between the two samples. Further, a majority of the sample is white thus providing limited information for students of color. Thus, future directions should include measurement invariance testing would also be a necessary step to test by gender, race, and status as a traditional versus non-traditional student. It would also be useful to establish predictive validity that involved linking scores on this scale to objective and subjective indicators of persistence. Although information on whether students planned on enrolling in college for the next term was collected, it was used primarily as a screener question in the Prolific Academic sample, thus there was no variance in the indicator. Persistence information was not collected in the first sample. Further, it may be that more complex analytic methods than utilized or reviewed here are needed to identify the structure as originally hypothesized. We hope that our discussion encourages such research.

Conclusion

Our findings suggest that students’ responses appear to be influenced by anticipating future rewards or punishments within the environment. Neither the class of positive or negative outcome expectation, nor the distance to the performance of staying in college, appeared to matter to students as much as the type of outcome expectation. Conceptualizing the scale in this manner, the results support a two-factor structure across two samples and provide evidence of validity in the second sample. These findings provide support for a reliable and valid scale that can be used to measure outcome expectations in the academic persistence domain.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Michael B. Kozlowski

Nadya A. Fouad

References

ACT (2017). National collegiate retention and persistence to degree rates. https://www.act.org/content/dam/act/unsecured/documents/retain_2015.pdf

Anderson

C. B.

Lee

H. Y.

Byars-Winston

Baldwin

C. D.

Cameron

Chang

(2016). Assessment of scientific communication self-efficacy, interest, and outcome expectations for career development in academic medicine. Journal of Career Assessment, 24(1), 182–196. https://doi.org/10.1177/1069072714565780

Autor

D. H.

(2014). Skills, education, and the rise of earnings inequality among the "other 99 percent. Science (New York, N.Y.), 344(6186), 843–851. https://doi.org/10.1126/science.1251868

Bandura

(1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84(2), 191–215. https://doi.org/10.1037/0033-295X.84.2.191

Bandura

(1986). Social foundations of thought and action: A social cognitive theory. Prentice-Hall, Inc.

Bandura

(1997). Self-efficacy: The exercise of control. W H Freeman/Times Books/ Henry Holt & Co.

Bandura

(2001). Social cognitive theory: An agentic perspective. Annual Review of Psychology, 52, 1–26. https://doi.org/10.1146/annurev.psych.52.1.1

Bean

J. P.

(1985). Interaction effects based on class level in an explanatory model of college student dropout syndrome. American Educational Research Journal, 22(1), 35–64. https://doi.org/10.3102/00028312022001035

Bieschke

K. J.

(2000). Factor structure of the research outcome expectations scale. Journal of Career Assessment, 8(3), 303–313. https://doi.org/10.1177/106907270000800307

10.

Bolkan

Pedersen

W. C.

Stormes

K. N.

Manke

(2021). Predicting 4-year graduation: Using social cognitive career theory to model the impact of prescriptive advising, unit load, and students’ self-efficacy. Journal of College Student Retention: Research, Theory & Practice, 22(4), 655–675. https://doi.org/10.1177/1521025118783485

11.

Brown

S. D.

Tramayne

Hoxha

Telander

Fan

Lent

R. W.

(2008). Social cognitive predictors of college students’ academic performance and persistence: A meta-analytic path analysis. Journal of Vocational Behavior, 72(3), 298–308. https://doi.org/10.1016/j.jvb.2007.09.003

12.

Byars-Winston

Rogers

Branchaw

Pribbenow

Hanke

Pfund

(2016). New measures assessing predictors of academic persistence for historically underrepresented racial/ethnic undergraduates in science. CBE—Life Sciences Education, 15(3), ar32. https://doi.org/10.1187/cbe.16-01-0030

13.

Carnevale

A. P.

Strohl

Ridley

Gulish

(2018). Three educational pathways to good jobs: High school, middle skills, and bachelor’s degree. Georgetown University Center for Education and the Workforce. https://cew.georgetown.edu/cew-reports/3pathways/

14.

Chen

K.-C.

Jang

S.-J.

(2010). Motivation in online learning: Testing a model of self-determination theory. Computers in Human Behavior, 26(4), 741–752. https://doi.org/10.1016/j.chb.2010.01.011

15.

Destin

Williams

J. L.

(2020). The connection between student identities and outcomes related to academic persistence. Annual Review of Developmental Psychology, 2(1), 437–460. https://doi.org/10.1146/annurev-devpsych-040920-042107

16.

Flores

L. Y.

Navarro

R. L.

DeWitz

S. J.

(2008). Mexican American high school students' postsecondary educational goals: Applying social cognitive career theory. Journal of Career Assessment, 16(4), 489–501. https://doi.org/10.1177/1069072708318905

17.

Flores

L. Y.

Navarro

R. L.

Lee

B. H.

Diaz

Martinez

(2021). Social cognitive predictors of Latinx and White engineering students’ academic satisfaction and persistence intentions: Exploring interactions among social identities and institutional context. Journal of Vocational Behavior, 127, 103580. https://doi.org/10.1016/j.jvb.2021.103580

18.

Fouad

N. A.

Guillen

(2006). Outcome expectations: Looking to the past and potential future. Journal of Career Assessment, 14(1), 130–142. https://doi.org/10.1177/1069072705281370

19.

Habley

W. R.

Bloom

J. L.

Robbins

(2012). Increasing persistence: Research-based strategies for college student success. John Wiley & Sons.

20.

Hattori

Zhang

Preacher

K. J.

(2017). Multiple local solutions and geomin rotation. Multivariate Behavioral Research, 52(6), 720–731. https://doi.org/10.1080/00273171.2017.1361312

21.

L-t.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

22.

Kahn

J. H.

Nauta

M. M.

(2001). Social-cognitive predictors of first-year college persistence: The importance of proximal assessment. Research in Higher Education, 42(6), 633–652. https://doi.org/10.1023/A:1012225510213

23.

Kirsch

(1995). Self-efficacy and outcome expectancies: A concluding commentary. In Maddux

J. E.

(Ed.), Self-efficacy, adaptation, and adjustment: Theory, research, and application (pp. 331–345). Plenum Press. https://doi.org/10.1007/978-1-4419-6868-5_12

24.

Lee

H. S.

Flores

L. Y.

Navarro

R. L.

Kanagui-Muñoz

(2015). A longitudinal test of social cognitive career theory’s academic persistence model among Latino/a and White men and women engineering students. Journal of Vocational Behavior, 88, 95–103. https://doi.org/10.1016/j.jvb.2015.02.003

25.

Lee

H. S.

Flores

L. Y.

Navarro

R. L.

Suh

H. N.

(2018). Development and validation of the negative outcome expectations scale in engineering (NOES-E). Journal of Career Assessment, 26(1), 52–67. https://doi.org/10.1177/1069072716679923

26.

Lent

R. W.

Brown

S. D.

(2006). On conceptualizing and assessing social cognitive constructs in career research: A measurement guide. Journal of Career Assessment, 14(1), 12–35. https://doi.org/10.1177/1069072705281364

27.

Lent

R. W.

Brown

S. D.

(2013). Social cognitive model of career self-management: Toward a unifying view of adaptive career behavior across the life span. Journal of Counseling Psychology, 60(4), 557–568. https://doi.org/10.1037/a0033446

28.

Lent

R. W.

Brown

S. D.

(2019). Social cognitive career theory at 25: Empirical status of the interest, choice, and performance models. Journal of Vocational Behavior, 115, 103316. https://doi.org/10.1016/j.jvb.2019.06.004

29.

Lent

R. W.

Brown

S. D.

Hackett

(1994). Toward a unifying social cognitive theory of career and academic interest, choice, and performance. Journal of Vocational Behavior, 45(1), 79–122. https://doi.org/10.1006/jvbe.1994.1027

30.

Lent

R. W.

Brown

S. D.

Hackett

(2000). Contextual supports and barriers to career choice: A social cognitive analysis. Journal of Counseling Psychology, 47(1), 36–49. https://doi.org/10.1037/0022-0167.47.1.36

31.

Lent

R. W.

Brown

S. D.

Schmidt

Brenner

Lyons

Treistman

(2003). Relation of contextual supports and barriers to choice behavior in engineering majors: Test of alternative social cognitive models. Journal of Counseling Psychology, 50(4), 458–465. https://doi.org/10.1037/0022-0167.50.4.458

32.

Lent

R. W.

Brown

S. D.

Sheu

H.-B.

Schmidt

Brenner

B. R.

Gloster

C. S.

Wilkins

Schmidt

L. C.

Lyons

Treistman

(2005). Social cognitive predictors of academic interests and goals in engineering: Utility for women and students at historically black universities. Journal of Counseling Psychology, 52(1), 84–92. https://doi.org/10.1037/0022-0167.52.1.84

33.

Lent

R. W.

Miller

M. J.

Smith

P. E.

Watford

B. A.

Lim

R. H.

Hui

(2016). Social cognitive predictors of academic persistence and performance in engineering: Applicability across gender and race/ethnicity. Journal of Vocational Behavior, 94, 79–88. https://doi.org/10.1016/j.jvb.2016.02.012

34.

Lent

R. W.

Miller

M. J.

Smith

P. E.

Watford

B. A.

Lim

R. H.

Hui

Morrison

M. A.

Wilkins

Williams

(2013). Social cognitive predictors of adjustment to engineering majors across gender and race/ethnicity. Journal of Vocational Behavior, 83(1), 22–30. https://doi.org/10.1016/j.jvb.2013.02.006

35.

Lent

R. W.

Singley

Sheu

H.-B.

Schmidt

J. A.

Schmidt

L. C.

(2007). Relation of social-cognitive factors to academic satisfaction in engineering students. Journal of Career Assessment, 15(1), 87–97. https://doi.org/10.1177/1069072706294518

36.

Lim

Jahng

(2019). Determining the number of factors using parallel analysis and its recent variants. Psychological Methods, 24(4), 452–467. https://doi.org/10.1037/met0000230

37.

Marsh

H. W.

Ellis

L. A.

Parada

R. H.

Richards

Heubeck

B. G.

(2005). A short version of the Self-Description Questionnaire II: Operationalizing criteria for short-form evaluation with new applications of confirmatory factor analyses. Psychological Assessment, 17(1), 81–102. https://doi.org/10.1037/1040-3590.17.1.81

38.

Marsh

H. W.

Hau

K. T.

Wen

(2004). In search of golden rules: Comment on hypothesis-testing approaches to setting cutoff values for fit indexes and dangers in overgeneralizing Hu and Bentler's (1999) findings. Structural Equation Modeling, 11(3), 320–341. https://doi.org/10.1207/s15328007sem1103_2

39.

McAuley

Motl

R. W.

White

S. M.

Wójcicki

T. R.

(2010). Validation of the multidimensional outcome expectations for exercise scale in ambulatory, symptom-free persons with multiple sclerosis. Archives of Physical Medicine and Rehabilitation, 91(1), 100–105. https://doi.org/10.1016/j.apmr.2009.09.011

40.

McCroskey

J. C.

Richmond

V. P.

(1996). Fundamentals of human communication: An interpersonal perspective. Prospect Heights, IL: Waveland Press.

41.

McNeish

(2018). Thanks coefficient alpha, we'll take it from here. Psychological Methods, 23(3), 412–433. https://doi.org/10.1037/met0000144

42.

Melguizo

(2011). A review of the theories developed to describe the process of college persistence and attainment. In Smart

J.C.

(Ed.), Higher education: Handbook of theory and research. Dordrecht: Springer. https://doi.org/10.1007/978-94-007-0702-3_10

43.

Messick

(1995). Validity of psychological assessment: Validation of inferences from persons’ responses and performances as scientific inquiry into score meaning. American Psychologist, 50(9), 741–749. https://doi.org/10.1037/0003-066X.50.9.741

44.

Moshagen

Erdfelder

(2016). A new strategy for testing structural equation models. Structural Equation Modeling, 23(1), 54–60. https://doi.org/10.1080/10705511.2014.950896

45.

Muthén

L. K.

Muthén

B. O.

(1998–2017). Mplus user’s guide (Eighth Edition). Los Angeles, CA: Muthén & Muthén.

46.

Muthén

L. K.

Muthén

B. O.

(2009). Mplus short courses: Topic 1: Exploratory factor analysis, confirmatory factor analysis, and structural equation modeling for continuous outcomes. Los Angeles CA: Muthén & Muthén. Retrieved from http://www.statmodel.com/course_materials.shtml

47.

National Student Clearinghouse Research Center (2016). First year persistence and retention. https://nscresearchcenter.org/wp-content/uploads/SnapshotReport28a.pdf

48.

Oliveira

Í. M.

Taveira

M. d. C.

Cadime

Porfeli

E. J.

(2016). Psychometric properties of a career exploratory outcome expectations measure. Journal of Career Assessment, 24(2), 380–396. https://doi.org/10.1177/1069072715580577

49.

Palan

Schitter

(2018). Prolific. ac—a subject pool for online experiments. Journal of Behavioral and Experimental Finance, 17, 22–27. https://doi.org/10.1016/j.jbef.2017.12.004

50.

Pantages

T. J.

Creedon

C. F.

(1978). Studies of college attrition: 1950–1975. Review of Educational Research, 48(1), 49–101. https://doi.org/10.2307/1169909

51.

Peer

Brandimarte

Samat

Acquisti

(2017). Beyond the Turk: Alternative platforms for crowdsourcing behavioral research. Journal of Experimental Social Psychology, 70, 153–163. https://doi.org/10.1016/j.jesp.2017.01.006

52.

Raykov

Marcoulides

G. A.

(2019). Thanks coefficient alpha, we still need you. Educational and Psychological Measurement, 79(1), 200–210. https://doi.org/10.1177/0013164417725127

53.

R Core Team (2022). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/

54.

Resnick

(2005). Reliability and validity of the outcome expectations for exercise scale-2. Journal of Aging & Physical Activity, 13(4), 382–394. https://doi.org/10.1123/japa.13.4.382

55.

Robitzsch

(2020). Why ordinal variables can (almost) always be treated as continuous variables: Clarifying assumptions of robust continuous and ordinal factor analysis estimation methods. Frontiers in Education, 5, 589965. https://doi.org/10.3389/feduc.2020.589965

56.

Rollnick

Morgan

Heather

(1996). The development of a brief scale to measure outcome expectations of reduced consumption among excessive drinkers. Addictive Behaviors, 21(3), 377–387. https://doi.org/10.1016/0306-4603(95)00068-2

57.

Schams

Fouad

A Burrows

G Ricondo

Song

(2022). Effect of a class-level intervention on career indecision variables. The Career Development Quarterly, 70(2), 162-171. https://doi.org/10.1002/cdq.12292

58.

Schneider

Yin

(2011). The high Cost of low graduation rates: How much does dropping Out of college really cost?. American Institutes for Research. https://www.air.org/sites/default/files/downloads/report/AIR_High_Cost_of_Low_Graduation_Aug2011_0.pdf

59.

Schunk

D. H.

DiBenedetto

M. K.

(2020). Motivation and social cognitive theory. Contemporary Educational Psychology, 60, 101832. https://doi.org/10.1016/j.cedpsych.2019.101832

60.

Solberg

V. S.

O'Brien

Villareal

Kennel

Davis

(1993). Self-efficacy and Hispanic college students: Validation of the college self-efficacy instrument. Hispanic Journal of Behavioral Sciences, 15(1), 80–95. https://doi.org/10.1177/07399863930151004

61.

Summerskill

(1962). Dropouts from college. In Sanford

(Ed.),The American college: A psychological and social interpretation of the higher learning (pp. 627–657). John Wiley & Sons Inc. https://doi.org/10.1037/11181-019

62.

The jamovi project (2022). Jamovi (version 2.2) [computer software]. https://www.jamovi.org

63.

Williams

D. M.

(2010). Outcome expectancy and self-efficacy: Theoretical implications of an unresolved contradiction. Personality and Social Psychology Review: An official Journal of the Society for Personality and Social Psychology, Inc, 14(4), 417–425. https://doi.org/10.1177/1088868310368802

64.

Wójcicki

T. R.

White

S. M.

McAuley

(2009). Assessing outcome expectations in older adults: The multidimensional outcome expectations for exercise scale. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 64B(1), 33–40. https://doi.org/10.1093/geronb/gbn032

65.

Wright

S. L.

Jenkins-Guarnieri

M. A.

Murdock

J. L.

(2013a). Career development among first-year college students: College self-efficacy, student persistence, and academic success. Journal of Career Development, 40(4), 292–310. https://doi.org/10.1177/0894845312455509

66.

Wright

S. L.

Wright

D. A.

Jenkins-Guarnieri

M. A.

(2013b). Development of the social efficacy and social outcome expectations scale. Measurement and Evaluation in Counseling and Development, 46(3), 218–231. https://doi.org/10.1177/0748175613484042