The General Academic Self-Efficacy Scale: Psychometric Properties,Longitudinal Invariance,and Criterion Validity

Abstract

Academic self-efficacy (ASE) refers to a student’s global belief in his/her ability to master the various academic challenges at university and is an essential antecedent of wellbeing and performance. The five-item General Academic Self-Efficacy Scale (GASE) showed promise as a short and concise measure for overall ASE. However, of its validity and reliability outside of Scandinavia is limited. Therefore, this paper aimed to investigate the psychometric properties, longitudinal invariance, and criterion validity of the GASE within a sample of university students (Time 1: n = 1056 & Time 2: n = 592) in the USA and Western Europe. The results showed that a unidimensional factorial model of overall ASE fitted the data well was reliable and invariant across time. Further, criterion validity was established by finding a positive relationship with task performance at different time stamps. Therefore, the GASE can be used as a valid and reliable measure for general ASE.

Keywords

academic self-efficacy task performance longitudinal invariance longitudinal confirmatory factor analysis university students

Introduction

General academic self-efficacy (ASE) refers to students’ global belief in their ability to master the various academic challenges at university and is an essential antecedent of wellbeing and academic performance (Nielsen et al., 2018). Within university contexts, higher levels of ASE has been associated with lower levels of depression/stress/anxiety (Tahmassian & Jalali-Moghadam, 2011), better decision-making, motivation, and engagement, as well as higher levels of academic- and task-performance (Doo & Bonk, 2020; Tossavainen et al., 2021; Van Zyl et al., 2021). As a social-cognitive process, ASE is concerned with developing the belief in one’s ability to obtain and optimize the cognitive, behavioral, emotional, and social resources required to perform better at academic-related tasks (Nielsen et al., 2018). A meta-analysis showed that ASE is the strongest predictor of overall performance at university i.t.o grade point average (over and above personality, motivation and learning styles) (Richardson et al., 2012). Various studies have also found that ASE is a strong predictor of students overall task performance (i.e., the proficiency to perform well in academic tasks through making the right choices and to take the initiative to perform the most important or core tasks central to their academic studies on time, and to specification) (Campbell & Hackett, 1986; Lim & Bang, 2018; Tossavainen et al., 2021). When students feel competent in their own academic abilities, they are better able to utilize their capabilities to prioritize the completion of competing academic tasks more effectively, are less likely to be discouraged by setbacks, less likely to procrastinate, and invest more effort into their studies (Richardson et al., 2012; Tossavainen et al., 2021). Given ASE’s importance for performing well at academic tasks, it is not surprising that its development has become a central strategy for universities to enhance academic throughput (Meintjes, 2020).

As such, various psychometric instruments to measure ASE have been developed (cf. Dever & Kim, 2016; Lindstrøm & Sharma, 2011; Owen & Froman, 1988; Zimmerman et al., 1992). However, these instruments are exceptionally lengthy (ranging from 16 to 33) and have shown different factorial structures and varying ranges of internal consistency in different settings (Meintjes, 2020). This could lead to biased results and limits the potential for cross-cultural comparisons. Developed and validated in Scandinavia, the English adapted five-item General Academic Self-Efficacy Scale (GASE) showed promise as a short, clear, and concise measure for overall academic self-efficacy (Nielsen et al., 2018). The scale measures the global belief in one’s ability to perform and plan tasks associated with an academic degree (Nielsen et al., 2018). The GASE has shown to be a valid and reliable measure in various studies and proved to be invariant between genders (Bass, 2020; Hitches et al., 2021; Nielsen, 2020). However, its validity and reliability outside of Scandinavia are yet to be investigated.

Therefore, the purpose of this paper was to investigate the psychometric properties, longitudinal invariance, and criterion validity (i.r.o. task performance) of the GASE within the tertiary educational environment in Western Europe and the US. The study aims to provide researchers and practitioners with evidence that the GASE can be used as a valid and reliable tool to measure general academic self-efficacy in university contexts.

Methodology

Research Design

A longitudinal, electronic survey-based research design was employed to explore the psychometric properties, longitudinal invariance and criterion validity of the GASE. Data for this paper form part of a larger cross-cultural student wellbeing project obtained at two time-points over 3 months.

Participants

A purposive sampling strategy was employed to gather data from university students from one academic university in the USA and -Belgium as well as one technical university in the Netherlands (c.f. Table 1). At Time 1, the majority of the 1056 participants were English speaking (48.3%), American (46.8%)¹ females (54.9%) between the ages of 21 and 25 (58.80%) who held part-time employment (55.9%). At Time 2, the majority of the 592 matched participants were Dutch-speaking (71.3%), and Western European (66.4%) males (60.8%) between the ages of 21 and 25 (72.6%) who held part-time employment (62.8%).

Table 1.

Demographic characteristics of participants at Time 1 (N = 1056) and Time 2 (N = 592).

Item	Category	Time 1		Time 2
Item	Category	Frequency (f)	Percentage (%)	Frequency (f)	Percentage (%)
Gender	Male	473	44.80	360	60.8
	Female	580	54.90	230	38.9
	Other	3	0.30	2	0.3
Age (years)	18–20 years	370	35.00	128	21.6
	21–25 years	621	58.80	430	72.6
	26–30 years	36	3.40	18	3.0
	31 years and older	29	2.70	16	2.7
Nationality	European	408	38.60	393	66.4
	USA	494	46.80	65	11.0
	Other European	130	12.30	130	22.0
	Other American	24	2.30	4	0.70
Home language	English	509	48.20	78	13.2
	Dutch	435	41.20	422	71.3
	Other	112	10.60	92	15.5
Employment	Full time	50	4.70	13	2.2
	Part time	590	55.9	372	62.8
	Unemployed	416	39.4	207	35.0

Measuring Instruments

The General Academic Self-Efficacy scale (GASE: Nielsen et al., 2018) was used to measure academic self-efficacy. This five-item self-report scale measures academic self-efficacy on a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). An example item is: “I know I can pass the exam if I put in enough work during the semester.” Akanni and Oduaran (2018) reported acceptable levels of internal consistency with a Cronbach’s alpha of 0.81.

The Task Performance sub-scale of Koopmans et al.’s (2012) Individual Work Performance Scale was used to measure overall task performance. Task performance was measured by seven items on a six-point Likert-scale ranging from 1 (“Never”) to 6 (“Always”). An example of item is: “I knew how to set the right priorities.” In Europe, the scale also produced acceptable levels of internal consistency as represented by a McDonald’s Omega of 0.84 (Van Zyl et al., in press).

Statistical Analysis

Data were processed with JASP v. 0.15 (JASP, 2021) and Mplus v 8.6 (Muthén & Muthén, 2021) through Structural Equation Modelling (SEM) with the maximum likelihood estimator. The full maximum likelihood estimation method (FIML) was used to manage missing data.

First, a unidimensional confirmatory factor analytical (CFA) model was estimated for the scale at each time-point. Model fit was evaluated through conventional standards (c.f. Table 2).

Table 2.

Model Fit Statistics.

Fit Indices	Cut-Off Criterion	Sensitive to N	Penalty for Model Complexity
Absolute fit indices
Chi-square (χ2)	• Lowest comparative value between measurement models	Yes	No
Chi-square (χ2)	• Non-significant Chi-square value (p > 0.01)	Yes	No
Approximate Fit Indices
Root-Means-Square error of Approximation (RMSEA)²	• 0.06 to 0.08 (Marginally acceptable); 0.00 to 0.05 (Excellent)	No	Yes
	• Non-significant RMSEA estimate (p > 0.05)

Standardized Root Mean Square Residual (SRMR)	• 0.06 to 0.08 (Marginally acceptable); 0.01 to 0.05 (Excellent)	Yes	No
Incremental fit indices
Comparative Fit Index (CFI)	• 0.90 to 0.95 (Marginally acceptable fit); 0.96 to 0.99 (Excellent)	No	No
Tucker–Lewis Index (TLI)	• 0.90 to 0.95 (Marginally acceptable fit); 0.96 to 0.99 (Excellent)	No	Yes
Akaike Information Criterion (AIC)	• Lowest value in comparative measurement models	Yes	Yes
Bayes Information Criterion (BIC)	• Lowest value in comparative measurement models	Yes	Yes

Adapted from Hu and Bentler (1999), Van Zyl and Ten Klooster (2022) and Wong and Wong (2020).

Second, a longitudinal CFA (LCFA) approach was employed to determine the temporal stability of the scale’s factor structure. Here, academic self-efficacy at Time 1 was regressed on academic self-efficacy at Time 2. The model had to show (a) good data-model fit (c.f Table 2), (b) excellent measurement quality (λ > 0.40; p < 0.01) (c) a positive regression path between the factors (p < 0.01), and (d) the scale at Time 1 needed to explain at least 40% of the variance in Time 2 (Wong & Wong, 2020). Further, the average variance explained and reliability estimates (Cronbach Alpha >0.70; McDonald’s omega > 0.70) were computed (Wong & Wong, 2020).

Third, longitudinal measurement invariance (LMI) was used to determine the configural (similar factor structure), metric (similar factor loadings), and scalar invariance (similar intercepts) of the scale over time. Invariance was established by comparing these models based on the following criteria: changes in RMSEA (Δ < 0.015), SRMR (Δ < 0.02 for configural vs. metric/scalar; Δ < 0.01 metric vs. scalar), CFI (Δ < 0.01), and TLI (Δ < 0.01) (Wong & Wong, 2020). Differences in χ² were not considered (but reported for transparency) due to its sensitivity to both sample size and model complexity (Morin et al., 2020; Van Zyl & Ten Klooster, 2022).

Finally, criterion validity was established by estimating concurrent and predictive validity via a structural model. Concurrent validity was estimated by relating Academic Self-Efficacy at Time 1 to Task Performance at Time 1 and Academic Self-Efficacy at Time 2 to Task Performance at Time 2. To establish predictive validity, Academic Self-Efficacy at Time 1 was regressed on Task Performance at Time 2. To enhance fit, Item 1 and Item 2 of the Task Performance scale were permitted to correlate. For each regression, a significance level of p < 0.01 was set.

Results

Confirmatory Factor Analysis: Cross-Sectional and Longitudinal

The factorial validity of the GASE at each time point were explored through estimating a single first-order factor model. Here all items were specified to load directly onto a single factor. Observed items were used as indicators for the latent factor. No items were removed. The results summarized in Table 3 showed that the model fitted the data well at both Time 1 (χ²₍₁₀₅₆₎ = 17.73 p > 0.001; df = 5; CFI = 0.99; TLI = 0.98; RMSEA = 0.05 [.026, .075]; SRMR = 0.02; AIC= 11,905.73; BIC = 11,980.16) and Time 2 (χ²₍₅₉₂₎ = 12.24 p > 0.001; df = 5; CFI = 0.99; TLI = 0.98; RMSEA = 0.05 [.014, .085]; SRMR = 0.02; AIC= 6673.08; BIC = 6738.83).

Table 3.

Cross-sectional and Longitudinal Confirmatory Factor Analyses.

Model	Type	χ²	df	CFI	TLI	RMSEA		SRMR	AIC	BIC	aBIC	Meets Criteria
Cross-Sectional CFA
Model 1	Unidimensional Model at Time 1	17.73	5	0.99	0.98	0.05	[.026–.075]	0.02	11,905.73	11,980.16	11,932.52	Yes
Model 2	Unidimensional Model at Time 2	12.24	5	0.99	0.98	0.05	[.014–.085]	0.02	6673.08	6738.83	6691.21	Yes
Longitudinal CFA
Model 3	Longitudinal Factor Analysis: Unidimensional Model	45.60	29	0.99	0.99	0.02	[.008–.036]	0.02	18,015.40	18,194.04	18,079.70	Yes

χ² = Chi-square; df = degrees of freedom; TLI = Tucker–Lewis index; CFI = comparative fit index; RMSEA = Root Mean Square Error of Approximation [90%CI]; SRMR = standardized root mean square residual; AIC = Akaike information criterion; BIC = Bayes information criterion; aBIC = adjusted bayes information criterion. Bold: non-significant p >0.001.

Thereafter, a LCFA model was estimated. Here ASE at Time 1 was regressed on ASE at Time 2. Error variances between time-points were permitted to covary. Table 3 indicates that this model also fitted the data well (χ²₍₁₀₅₆₎ = 45.60 p > 0.001; df = 29; CFI = 0.99; TLI = 0.99; RMSEA = 0.02 [.008, .036]; SRMR = 0.02; AIC = 18,015.40; BIC = 18,194.04). GASE at Time 1 was also significantly related to GASE at Time 2 (β = 0.69; S.E = 0.03; t-value= 22.05; p < 0.01) and explained 47% of the overall variance.

Item Level Parameter Estimates and Internal Consistency

Table 4 summarizes the item level parameter estimates and internal consistency of the LCFA model. All factor loadings were significant and greater than 0.40 at both time-points. The AVE was 40% at Time 1 and 45% at Time 2. The scale showed acceptable levels of internal consistency at both Time Points with Cronbach’s alpha and McDonald’s Omega ranging from 0.74 to 0.78.

Table 4.

Item Level Parameter Estimates And Internal Consistency of the GASE.

Factor	Item Description	Time 1					Time 2
		λ	S.E.	AVE	ω	α	λ	S.E.	AVE	ω	α
Academic self-efficacy				0.40	0.74	0.74			0.45	0.78	0.78
GASE_1	I generally manage to solve difficult academic problems if I try hard enough	0.69	0.02				0.70	0.03
GASE_2	I know I can stick to my aims and accomplish my goals in my field of study	0.70	0.02				0.69	0.03
GASE_3	I will remain calm in my exam because I know I will have the knowledge to solve the problems	0.59	0.03				0.66	0.03
GASE_4	I know I can pass the exam if I put in enough work during the semester	0.67	0.02				0.72	0.03
GASE_5	The motto ‘if other people can, I can too’ applies to me when it comes to my field of study	0.47	0.03				0.55	0.03

λ: Standardized factor loadings; SE = standard error; AVE = average variance extracted; ω: McDonald’s Omega. α: Cronbach’s Alpha.

Longitudinal Measurement Invariance and Mean Comparisons

Measurement equivalence was investigated through LMI. The results summarized in Table 5 showed that metric, configural, and scalar invariance was established. No significant differences in terms of RMSEA (Δ < 0.015), SRMR (Δ < 0.02 for configural vs. metric/scalar; Δ < 0.01 metric vs. scalar), CFI (Δ < 0.01), and TLI (Δ < 0.01) between the configural, metric, and scalar invariance models were found (Wong & Wong, 2020). Therefore, the GASE showed to be a consistent measure over time and mean comparisons can be made. Latent Mean comparisons, with GASE at Time 1 set as the reference point, showed that GASE decreased slightly over three months (Δ

\bar{x}

= −0.20; S.E = 0.04; p = 0.00).

Table 5.

Longitudinal Invariance of the Unidimensional General Academic Self-Efficacy Scale.

No	Model	χ²	df	CFI	TLI	RMSEA		SRMR	Model Comparison	Δχ²	ΔCFI	ΔTLI	ΔRMSEA	ΔSRMR	Meets Invariance Criteria
M1	Configural Invariance	45.60	29	0.99	0.99	0.023	[.008–036]	0.023	M3 versus M1	43.55	−0.01	−0.01	0.015	0.016	Yes
M2	Metric Invariance	51.72	32	0.99	0.99	0.024	[.011–.036]	0.034	M2 versus M1	6.12	0.00	0.00	0.001	0.011	Yes
M3	Scalar Invariance	89.15	35	0.98	0.98	0.038	[.029–.048]	0.039	M3 versus M2	37.43	−0.01	−0.01	0.014	0.005	Yes

χ²= Chi-square; df = degrees of freedom; TLI = Tucker-Lewis Index; CFI = comparative fit index; RMSEA = root mean square error of approximation [90%CI]; SRMR = standardised root mean square residual.

Concurrent and Predictive Validity

Finally, separate structural models were estimated to evaluate the concurrent and predictive validity of the instrument (c.f. Table 6). Concurrent validity was established through finding a positive relationship between GASE and Task Performance at both Time 1 (χ²₍₁₀₅₆₎ = 215.56; df = 52; CFI = 0.97; TLI = 0.96; RMSEA = 0.06 [.047, .062]; SRMR = 0.03; AIC= 32,430.12; BIC = 32,618.68; β = 0.54; S.E= 0.03; R² = 0.29) and Time 2 (χ²₍₅₉₂₎ = 210.06; df = 52; CFI = 0.94; TLI = 0.92; RMSEA = 0.07 [.062, .082]; SRMR = 0.04; AIC = 18,412.02; BIC = 18,578.59; β= 0.58; S.E = 0.04; R² = 0.34).

Table 6.

Concurrent and Predictive Validity: General Academic Self-Efficacy and Task Performance.

Type	Regression Path			Standardized
Type	Regression Path			Beta	S.E	t-value	p	R²	Validity Established
Concurrent	Academic Self-Efficacy Time 1	→	Task Performance Time 1	0.54	0.03	18.82	0.00	0.29	Yes
Concurrent	Academic Self-Efficacy Time 2	→	Task Performance Time 2	0.58	0.04	16.09	0.00	0.34	Yes
Predictive	Academic Self-Efficacy Time 1	→	Task Performance Time 2	0.39	0.05	8.62	0.00	0.15	Yes

→ = regression; β = standardized beta; SE = standard error; p = statistical significance; R² = variance.

Predictive validity was also established through finding a positive relationship between GASE at Time 1 with Task Performance at Time 2 (χ²₍₁₀₅₆₎ = 174.633; df = 52; CFI = 0.96; TLI = 0.95; RMSEA = 0.05 [.040, .055]; SRMR = 0.04; AIC = 23,741.36; BIC = 23,929.93; β = 0.39; S.E = 0.08; R² = 0.15).

Discussion

The purpose of this paper was to investigate the psychometric properties, longitudinal invariance, and criterion validity (i.t.o. task performance) of the GASE within a US and Western European tertiary educational environment. The results showed that a single, first-order factorial model of overall academic self-efficacy fitted the data well, was reliable and invariant across time. In line with Nielsen et al. (2018), our findings show that the GASE measures general academic self-efficacy validly and reliably. Therefore, the mean scores of the GASE could be used by educational practitioners to measure ASE and track the effectiveness of educational programs or interventions aimed at enhancing ASE over time.

Further, criterion validity was established by finding a positive relationship with task performance at different time stamps. The results imply that when individuals hold active and positive beliefs about their abilities to plan/perform certain educational tasks or manage the challenges associated with their study programs, they are more likely to perform better in their academic-related tasks. This is because the feeling that one has mastery over the skills required to perform a given educational task enhances the engagement and motivation required to perform (Tossavainen et al., 2021). Further, according to Richardson et al. (2012) this could also be because holding a high level of ASE affects how obstacles or challenges are viewed (opportunities to learn vs. setbacks) which in turn leads to sustained task-related performance over time (Richardson et al., 2012; Tossavainen et al., 2021).

In conclusion, our results support the GASE as a valid and reliable measure for general academic self-efficacy within the current context. However, the study has its limitations. First, the sample is limited to a single US and two Western European universities. The results may therefore not be generalizable. Second, only (self-reported) task performance was used to establish criterion validity. Future research should include objective performance measures (e.g., grades) and other metrics associated with ASE, such as engagement, motivation, and resilience. Second, although various mechanisms were implemented to manage potential sample size attrition over time (e.g., students obtained course credit for participation; the follow-up assessment was kept as brief as possible; multiple reminders being sent: Mason, 1999) there was a 44% dropout between Time 1 and Time 2. This could have led to attrition bias which may affect the internal validity of the LFA and LMI assessments. However, the sample size at Time 2 is large enough to capture a full range of variation in responses, and therefore the configural, metric, and scalar invariance assessments are valid for the current study. Future research should attempt to manage the dropout rate through implementing more implicit and explicit incentives for participation at Time 2. Finally, with increased global competition between academic institutions, potential students and future employers may be interested in how effective academic programs are in developing more self-efficacious students. Given that the invariance between the two nations was not investigated due to sample size limitations, these types of comparisons cannot be made. Future research should aim to test the measurement equivalence of the scale across cultures.

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 iD

Llewellyn E. van Zyl

Notes

References

Akanni

A. A.

Oduaran

C. A.

(2018). Perceived social support and life satisfaction among freshmen: Mediating roles of academic self-efficacy and academic adjustment. Journal of Psychology in Africa, 28(2), 89–93. http://doi.org/10.1080/14330237.2018.1454582

Bass

K. L.

(2020). Can intentional instruction influence self-efficacy and increase academic performance? A pygmalion effect in Spanish language learners (Doctoral dissertation, Texas A&M University-Central Texas).

Campbell

N. K.

Hackett

(1986). The effects of mathematics task performance on math self-efficacy and task interest. Journal of Vocational Behavior, 28(2), 149–162. http://doi.org/10.1016/0001-8791(86)90048-5

Curran

P. J.

Bollen

K. A.

Chen

Paxton

Kirby

J. B.

(2003). Finite sampling properties of the point estimates and confidence intervals of the RMSEA. Sociological Methods & Research, 32(2), 208–252. http://doi.org/10.1177/0049124103256130

Dever

B. V.

Kim

S. Y.

(2016). Measurement equivalence of the PALS academic self-efficacy scale. European Journal of Psychological Assessment, 32(1), 61–67. http://doi.org/10.1027/1015-5759/a000331

Doo

M. Y.

Bonk

C. J.

(2020). The effects of self‐efficacy, self‐regulation and social presence on learning engagement in a large university class using flipped Learning. Journal of Computer Assisted Learning, 36(6), 997–1010. http://doi.org/10.1111/jcal.12455

Hitches

Woodcock

Ehrich

(2021). Shedding light on students with support needs: Comparisons of stress, self-efficacy, and disclosure. Journal of Diversity in Higher Education. Advance online publication. https://doi.org/10.1037/dhe0000328

L. T.

Bentler

P. M.

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

JASP . (2021). Jeffery’s amazing stats program (v.0.15). [Statistical Software]. https://jasp-stats.org

10.

Koopmans

Bernaards

C. M.

Hildebrandt

V. H.

Van Buuren

Van der Beek

A. J.

De Vet

H. C. W.

(2012). Development of an individual work performance questionnaire. International Journal of Productivity and Performance Management, 62(1), 6–28. https://doi.org/10.1108/17410401311285273

11.

Lim

H. A.

Bang

E. J.

(2018). The effects of music listening on affect, self-efficacy, mental exertion, and task performance of online learners. Journal of the Scholarship of Teaching and Learning for Christians in Higher Education, 8(1), 5–19. http://doi.org/10.31380/sotlched.8.1.13

12.

Lindstrøm

Sharma

M. D.

(2011). Self-efficacy of first year university physics students: do gender and prior formal instruction in physics matter? International Journal of Innovation in Science and Mathematics Education, 19(2), 1–19.

13.

MacCallum

R. C.

Browne

M. W.

Sugawara

H. M.

(1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods, 1(2), 130–149. http://doi.org/10.1037/1082-989x.1.2.130

14.

Mason

(1999). A review of procedural and Statistical methods for handling attrition and missing data in clinical research. Measurement and Evaluation in Counseling and Development, 32(2), 111–118. http://doi.org/10.1080/07481756.1999.12068976

15.

Meintjes

(2020). Depression and self-regulated learning as predictors of first-year students’ academic performance: A case study (Doctoral dissertation, University of the Free State).

16.

Morin

A. J. S.

Myers

N. D.

Lee

(2020). Modern factor analytic techniques: bifactor models, exploratory Exploratory Structural Equation Modeling 19 structural equation modeling and bifactor-ESEM. In Tenenbaum

Eklund

R. C.

(Eds.), Handbook of sport psychology (4th ed. Vol. 2, pp. 1044–1073). : Wiley Publishers.

17.

Muthén

L. K.

Muthén

B. O.

(2021). Mplus (Version 8.4). [Statistical software]. Muthén and Muthén.

18.

Nielsen

(2020). Measuring differences in Current Academic Learning Self-Efficacy (CAL-SE) and Current Academic Exam Self-Efficacy (CAE-SE) for social science students: A Rasch-based multi degree-program validity study. In Research on teaching and Learning in higher education (p. 57). Waxmann Verlag.

19.

Nielsen

Dammeyer

Vang

M. L.

Makransky

(2018). Gender fairness in self-efficacy? A Rasch-based validity study of the General Academic Self-efficacy scale (GASE). Scandinavian Journal of Educational Research, 62(5), 664–681. http://doi.org/10.1080/00313831.2017.1306796

20.

Owen

S. V.

Froman

R. D.

(1988, April). Development of a College Academic Self-Efficacy Scale. Paper presented at the Annual Meeting of the National Council on Measurement in Education, New Orleans, LA.

21.

Richardson

Abraham

Bond

(2012). Psychological correlates of university students’ academic performance: A systematic review and meta-analysis. Psychological Bulletin, 138(2), 353–387. http://doi.org/10.1037/a0026838

22.

Tahmassian

Jalali-Moghadam

(2011). Relationship between self-efficacy and symptoms of anxiety, Depression, worry and social avoidance in a normal sample of students. Iranian Journal of Psychiatry and Behavioral Sciences, 5(2), 91–98.

23.

Tossavainen

Rensaa

R. J.

Johansson

(2021). Swedish first-year engineering students’ views of mathematics, self-efficacy and motivation and their effect on task performance. International Journal of Mathematical Education in Science and Technology, 52(1), 23–33. http://doi.org/10.1080/0020739x.2019.1656827

24.

Van Zyl

L. E.

Rothmann

Zondervan-Zwijnenburg

M. A. J.

(2021). Longitudinal trajectories of study characteristics and mental health before and during the COVID-19 Lockdown. Frontiers in Psychology, 12, 1–13. http://doi.org/10.3389/fpsyg.2021.63353

25.

Van Zyl

L.E.

Ten Klooster

P.M.

(2022). Exploratory structural equation modelling: Practical guidelines and tutorial with a convenient online tool for mplus. Frontiers in Psychiatry, 12, 795672. http://doi.org/0.3389/fpsyt.2021.795672

26.

Van Zyl

L. E.

Van Vuuren

Roll

L. C.

Stander

M. W.

(In review). Person-Environment Fit and Task Performance: Exploring the Role of Grit as a Personal Resource. Current Psychology.

27.

Wong

(2020). Structural equation modelling: Applications using Mplus (2nd ed.). : Wiley & Sons.

28.

Zimmerman

B. J.

Bandura

Martinez-Pons

(1992). Self-motivation for academic attainment: The role of self-efficacy beliefs and personal goal setting. American educational research journal, 29(3), 663-676