Rooted in the Big Five theory of personality, conscientiousness was defined as being orderly, dependable, and diligent (Costa and McCrae, 2008). Indeed, conscientiousness represented one’s traits, which included competence, order, dutifulness, achievement striving, self-discipline, and deliberation (McGhee et al., 2007), and these qualities were consistent with the academic orientation.

As traditionally used, the term “academic performance” refers to some method of expressing a student’s scholastic standing (Lavin, 1965). In higher education, students’ performance was usually expressed in terms of grade point average (GPA), which was the key criterion for postgraduate selection and graduate employment (Strenze, 2007).

Academic research on the positive correlation between conscientiousness and AP was extensive and sufficient (Vedel, 2014). The fact that conscientious students could achieve greater academic success had been verified repeatedly, both in high school (Trautwein et al., 2009) and at college levels (Chamorro-Premuzic and Furnham, 2008). Besides, compared with other personality of the Big Five theory, such as extraversion, agreeableness, neuroticism, and openness, conscientiousness was the strongest and most reliable predictor of academic achievement (Trapmann et al., 2007). However, the academic community had not yet reached any consensus on the mediating mechanism between conscientiousness and AP. Therefore, what mediates the relationship between these two variables needs further exploration.

First, according to the Social Cognitive Theory, self-efficacy was represented as a situational heterogeneous concept affecting individuals’ construction of activities, goals, and persistent choices (Bandura, 1986a). In the educational context, academic self-efficacy refers to the judgment of a learner’s ability to achieve the desired performance in the academic field (Bandura, 1997).

Furthermore, research results in recent decades have confirmed that ASE was positively correlated with both conscientiousness (Chen et al., 2001) and AP (Zeinalipour, 2022). More importantly, it was verified that more conscientious students contributed to a higher level of ASE (Lee and Klein, 2002), and students with a stronger sense of ASE would positively influence students’ academic achievement at last (Di Giunta et al., 2013). In other words, academic self-efficacy was of obvious causal relationships with conscientiousness and AP, and this connection laid the foundation for ASE to become a potential mediating variable to bridge the above path.

Second, academic motivation, which originated from the Self-Determination Theory, was referred to as the different loci of causality with regard to engagement in learning, which was often differentiated from extrinsic to intrinsic motivation (Ryan and Deci, 2000). More specifically, academic intrinsic motivation often refers to engaging in academic activities for their own sake, while extrinsic motivation means engaging in learning due to external factors (McGeown et al., 2014).

More importantly, AM was also closely linked to both conscientiousness (Apostolov and Geldenhuys, 2022) and AP (Gnambs and Hanfstingl, 2016). Furthermore, recent studies have testified that personality traits predicted 20.6% of the variance in intrinsic motivation and 5.1% in extrinsic motivation; specifically, only conscientiousness significantly predicted both intrinsic and extrinsic motivation (McGeown et al., 2014). Besides, AM accounted for 10% of the variance in academic achievement in university students based on a Turkish study (Kırkağaç and Öz, 2017). The strong causal relationship between AM, conscientiousness, and AP has been widely confirmed by academics. Consequently, AM would potentially play a mediating role in the connection between conscientiousness and AP.

Finally, ASE also correlated with motivation (Bandura, 1986b) and had repeatedly been verified to be a significant predictor of both intrinsic and extrinsic motivation (McGeown et al., 2014), which meant that the path from ASE to AM is also viable. Therefore, in together, based on the above findings on the causal relationships among conscientiousness, ASE, AM, and AP, the path starting from conscientiousness to AP through ASE and AM seems logical and feasible. Previous studies had attempted to incorporate conscientiousness, ASE, and AP with mediation models for research (Cupani and Pautassi, 2013), but there was still limited research on the assumed dual mediation in this article.

In the field of educational research, demographic information, especially gender, had been frequently used as a moderating variable (Choma et al., 2010) and was of a solid research foundation. Moreover, the research on gender serving as a moderator in its relationship with self-efficacy showed diversity, and the results of gender moderating ASE were not uniform. Some studies found no difference between male and female students, while others suggested that male students may have higher levels of self-efficacy and better science learning outcomes (Nurhasnah et al., 2022). For instance, a meta-analysis of 187 studies (including 247 individual studies, N = 68,429) on gender differences in ASE found that women reported a higher level of self-efficacy in the arts than men; meanwhile, men reported stronger self-efficacy in the sciences than women (Huang, 2013). In addition, studies tested how gender might influence the relationship between self-efficacy, academic success, and science identity (Miles and Naumann, 2021).

Overall, these findings highlighted the significance of considering gender when examining ASE, as it played an important role in shaping students’ educational beliefs and motivations. Based on previous research, it seems plausible that gender may play a moderated mediation role in the pathway from conscientiousness to ASE, AM, and AP.

The hypothesized model (Figure 1) proposed in this study was used to verify the dual mediating role of ASE and AM in the path from conscientiousness to AP and, further, to prove the moderated dual mediating role of gender in the above relationship.

The respondents of this study were 304 college students from China, and the sample covered 32 of the 34 provincial-level administrative regions (provinces, autonomous regions, municipalities directly under the central government, and special administrative regions) in China. The overall characteristics of the respondents were as follows: by gender, 40.1% were male and 59.9% were female and by age, 12.5% of the respondents were no older than 19 years of age, 30.6% were 20 years of age, 26.6% were 21 years of age, 22.7% were 22 years of age, and 7.6% were older than 22 years. From the perspective of only children, 41.4% of the respondents were not the only child, and 58.6% were the only child. In addition, 8.2% of the respondents came from single-parent families, and 91.8% came from two-parent families. In terms of socioeconomic status, 25.7% of the respondents came from wealthy families, 53.6% from average families, and 20.7% from less wealthy families. Regarding the participants’ current educational background, 49.7% were 4-year undergraduate students and 50.3% were 3-year college students.

The sampling process was conducted from 17 June to 28 June 2024. The sampling subjects were 870,000 Chinese college students registered on the Wenjuanxing platform. The sampling method was simple random sampling. First, the computer software randomly selected 3,782 registered college students on the Wenjuanxing registration platform. Then, an invitation letter, including a consent letter and a questionnaire, was sent to invite these selected students. In the invitation letter, the purpose of the study, the research procedures, benefits, risks, confidentiality, voluntary participation, and the withdrawal terms were explained to potential participants. Only students who agreed to participate and confirmed the consent form were included in the formal sampling procedure. Afterwards, 1,323 students agreed to participate in the study who had confirmed the consent form. Then, the participants were asked to complete an online questionnaire through the Wenjuanxing platform within 20 minutes, and 1,040 participants submitted the results. Participants who answered fully and submitted the questionnaire will receive a 5-yuan reward. Next, the research team checked the results and deleted invalid questionnaires. Specifically, a total of 737 results were judged to be invalid. Finally, 304 participants were marked as valid results and used for further analysis.

As for data analysis, the analysis tools were SPSS Ver.25 and SPSS PROCESS macro-Ver. 4.2. More specifically, frequency analysis was adopted to illustrate the characteristics of the participants. Next, the reverse question was addressed, and Cronbach’s α was tested to ensure the reliability of the variables. Afterward, to find out the answer to the first research question, Pearson’s bivariate correlation method was applied to demonstrate the relationships among the main variables. Finally, to address the second research question, the SPSS PROCESS macro model No.83 was undertaken to show the moderated dual mediation path among the variables. In addition, to analyze the moderated mediation effect, prior to the analysis, conscientiousness was mean-centered, the confidence level of the output confidence interval was 95%, and the number of bootstrap samples for the percentile bootstrap confidence interval was 5,000.

The correlations within the main constructs are shown in Table 1 through Pearson correlation analysis. Specifically, conscientiousness was positively correlated with ASE (r = 0.618, p < 0.001), AM (r = 0.664, p < 0.001), and AP (r = 0.598, p < 0.001). ASE was positively correlated with AM (r = 0.624, p < 0.001) and AP (r = 0.615, p < 0.001). AM was positively correlated with AP (r = 0.586, p < 0.001). In the meantime, the correlation coefficients between the constructs were all less than 0.7, suggesting no multicollinearity problem. In addition, regression analysis was conducted to calculate the variance inflation factor (VIF) and tolerance values, with AP as the dependent variable and conscientiousness, gender, ASE, and AM as independent variables. The results showed that the tolerance values of all independent variables were between 0.48 and 1, and the VIF value of the model was 2.079, so there was no problem of multicollinearity in the results. Furthermore, through frequency analysis, the value of conscientiousness averaged at 4.63 (range 1 to 6), ASE averaged at 3.66 (range 1 to 5), AM averaged at 5.77 (range −18 to 18), and AP averaged at 3.84 (range 1 to 5).

To find out the moderated mediation effect of gender on the path from conscientiousness to ASE, AM, and AP, model No.83 of SPSS PROCESS macro was applied for analysis. The results are displayed in Tables 2, 3 and Figures 2, 3.

First, as shown in Table 2 and Figure 2, conscientiousness positively affected ASE (0.8460, p < 0.001), AM (3.0278, p < 0.001), and AP (0.2503, p < 0.001). Moreover, ASE positively affected AM (2.2193, p < 0.001) and AP (0.2882, p < 0.001). Finally, AM significantly influenced AP in a positive way (0.0273, p < 0.01).

Second, as shown in Table 2, the interaction of conscientiousness and gender significantly affected ASE (−0.1803, p < 0.05), and the increase in R² according to the interaction term between conscientiousness and ASE (∆R² = 0.0080, p < 0.001) was also significant and consequently had a moderating effect. More specifically, according to the value of gender, two conditions (male and female) were given, and the conditional effect of conscientiousness on ASE according to the value of gender was found to be significant for both genders (p < 0.001).

The moderating effect of gender between conscientiousness and ASE was also shown in Figure 3. To confirm the pattern of interaction, the change was analyzed by dividing gender into two conditions: male and female. The degree of ASE was moderated in that conscientiousness would increase under these two conditions. When the gender was female, the slope of increasing ASE as conscientiousness increased was relatively gentle, but when the gender was male, the slope of increasing ASE as conscientiousness increased was relatively steep.

Finally, as shown in Table 3, to verify the moderated mediation effect, the conditional and non-conditional indirect effect of gender was analyzed. The direct effect of conscientiousness on AP was of significant positive effect (B = 0.2503, p < 0.001), and it was significant as zero did not exist in 95% of the confidence interval (0.1393 to 0.3613). Furthermore, the conditional indirect effect of ASE in the relationship between conscientiousness and AP was significant in both conditions of gender. Moreover, in the effect of conscientiousness on AP, the indirect effect of AM was of a significant positive effect (B = 0.0825), and it was significant because zero did not exist in 95% of the confidence interval (0.0215 to 0.1527). Furthermore, in the effect of conscientiousness on AP, the conditional indirect effect of ASE and AM was significant in both conditions (male and female) of gender. The adjusted index of moderated mediation was −0.0109, which was significant because there was no zero in the 95% confidence interval (−0.0261 to −0.0001). Given these results, the moderated mediating effect of male or female status was verified in the path from conscientiousness to AP, ASE, and AM.