Academic confidence most strongly predicts Peruvian students’ perceived research skills

In a cross-sectional study of 650 undergraduates at a private university in Peru, academic self-efficacy was the strongest statistical predictor of students’ self-perceived research skills, while a positive attitude toward scientific research also contributed significantly. The researchers analyzed questionnaire data from the Peruvian Union University’s Juliaca campus during the 2025-II academic semester. Multiple regression found that academic self-efficacy had a standardized coefficient of 0.678 and attitude toward research had a coefficient of 0.145; both associations were statistically significant at p < 0.001. Together, the two psychological variables accounted for 56.5% of the statistical variance in students’ self-rated research skills.

The result concerns perceptions rather than demonstrated competence. Self-perceived research skills were measured with a 17-item scale covering thinking skills, academic writing and information communication, while academic self-efficacy captured students’ general belief that they could organize and execute actions needed to meet academic goals. All three main constructs were assessed through self-report instruments, making the strong correlation between academic self-efficacy and perceived research ability difficult to disentangle from conceptual overlap, a general tendency to evaluate oneself favorably across domains, or common-method effects. The authors stress that the regression coefficients represent statistical associations, not evidence that confidence or favorable attitudes cause better research performance.

The study used an explanatory–associative, non-experimental design. Data were collected once through a virtual Google Forms survey, with no manipulation of attitudes, self-efficacy or research training. Participants were selected by non-probability convenience sampling from a population of 8,000 undergraduates. The final sample included students from all faculties and academic cycles who voluntarily completed the questionnaires, with incomplete and duplicate records removed before analysis. Of the participants, 52.8% were women and 47.2% men; 56% were under 20, and 46.2% were in their ninth academic semester. Administration was the largest degree program, representing 35.7% of the sample, followed by Environmental Engineering at 14%.

Three validated scales supplied the measurements. The Attitude Toward Scientific Research Scale contained 28 items divided among interest in research, research vocation and evaluation of research, with responses recorded on a five-point strongly disagree–strongly agree scale. The nine-item Academic Self-Efficacy Scale used a four-point never–always format and was treated as a global measure of efficacy beliefs. The Self-Perceived Research Skills Scale included 17 items spanning thinking, writing and information-communication abilities. Confirmatory factor analyses in the present sample supported the instruments’ factorial structures. Chi-square-to-degrees-of-freedom ratios ranged from 1.703 to 1.752, while comparative fit indices ranged from 0.923 to 0.978. RMSEA values were between 0.052 and 0.076, and standardized root mean square residuals ranged from 0.025 to 0.035.

Before interpreting the regression, the researchers evaluated the model’s statistical assumptions. Tolerance and variance inflation factor values were both 0.815 and 1.228, respectively, indicating that multicollinearity was not a serious concern. A Durbin–Watson statistic of 1.96 was close to the expected value of 2, and residual normality was supported by a Shapiro–Wilk result of W = 0.990 with p = 0.081. Residual skewness of −0.22 and kurtosis of 0.48 were within acceptable limits. Visual inspection of residual-versus-predicted plots showed no substantial departures from linearity or homoscedasticity. Maximum Cook’s distance was 0.006 and maximum leverage was 0.007, and standardized residuals ranged from −2.51 to 2.68, so no observations were removed as unduly influential outliers.

The correlations provided a more detailed picture of the relationships. Attitude toward research was moderately associated with academic self-efficacy at r = 0.431 and with self-perceived research skills at r = 0.437; both p-values were below 0.001. Academic self-efficacy and self-perceived research skills had a strong correlation of r = 0.740, also with p < 0.001. Because this latter association was estimated from self-reported measures collected at the same time, it may partly reflect students’ general confidence in their academic capabilities rather than a specific, empirically demonstrated mastery of research tasks. The descriptive means were 100.02 for attitude toward research, 24.92 for academic self-efficacy and 61.27 for self-perceived research skills, with standard deviations of 23.04, 5.63 and 11.18, respectively.

The multiple regression model was statistically significant, with F = 421.016 and p < 0.001. Academic self-efficacy carried by far the greater standardized weight: its coefficient was 0.678, compared with 0.145 for attitude toward research. In unstandardized terms, a one-unit increase in the academic self-efficacy score was associated with a 1.346-unit increase in perceived research skills, while a one-unit increase in the attitude score was associated with a 0.071-unit increase. The model’s R² was 0.565 and adjusted R² was 0.564. These figures describe the model’s statistical explanatory performance for this sample; they do not show what would happen if educators increased either psychological factor or how much of the difference would transfer to a broader student population.

The findings fit a sociocognitive interpretation of university research training, in which efficacy beliefs are related to effort, persistence and willingness to approach difficult academic tasks. They also align with arguments that students who value research may be more inclined to engage in investigative activities. The authors cite research showing that positive attitudes can coexist with methodological weaknesses, including difficulties among Peruvian students in formulating problems and analyzing data with statistical techniques. Academic self-efficacy, attitude toward research and self-perceived research skills are related but distinct constructs: the first concerns general academic capability beliefs, the second captures evaluative and affective orientations toward research, and the third reflects students’ judgments of their own research abilities. Their shared reliance on self-assessment nevertheless makes the especially strong r = 0.740 relationship difficult to interpret independently of measurement effects.

For universities, the results suggest a case for examining psychological conditions alongside technical research instruction. The authors propose gradual, experiential training in which students can observe, practice, analyze and reflect on research procedures, supported by specific feedback, research advising, collaborative groups and access to databases, statistical software, laboratories and repositories. They also point to extracurricular opportunities such as scientific clubs, symposia and institutional journals. These are implications drawn from observed associations, not interventions validated by the present study. The design cannot determine whether mentoring, feedback or improved facilities would raise academic self-efficacy, change attitudes or improve actual research ability. Future work should therefore use longitudinal or intervention-based designs, objective performance measures and assessments of students’ real research competencies.

Several limitations constrain how broadly the results can be applied. Convenience sampling from one university campus limits representativeness, and the uneven distribution of students across degree programs and academic stages may have influenced the scores. Nearly half the sample occupied the ninth semester, so students with more accumulated exposure to research methods may have differed from those in earlier years. The study did not control for academic semester, field of study, age or gender, and the investigators note that these factors should be examined in future models. The single-time-point design also prevents assessment of change over time, while the self-report format introduces potential social-desirability and self-assessment biases. The strongest defensible conclusion is therefore that, among these Peruvian students, greater academic confidence was closely associated with stronger perceived research skills, and favorable attitudes toward research made a smaller but significant contribution. Whether those beliefs correspond to better research outcomes remains a separate question requiring direct performance assessment and experimental or longitudinal evidence.

Subject of Research: Associations between academic self-efficacy, attitudes toward research and self-perceived research skills among Peruvian university students

Article Title: Attitude toward scientific research and academic self-efficacy as statistical predictors of self-perceived research skills in university students in Peru

Article References: Torres Medina, W. E., & Montalvo Apolín, D. E. (2026). Attitude toward scientific research and academic self-efficacy as statistical predictors of self-perceived research skills in university students in Peru. Frontiers in Education, 11, Article 1926983. https://doi.org/10.3389/feduc.2026.1926983

Image Credits: AI Generated

DOI: 10.3389/feduc.2026.1926983

Keywords: academic self-efficacy, research attitudes, research skills, university students, Peru, self-report bias, cross-sectional study, higher education, psychological predictors, regression analysis, Attitude, toward