Based on the interbehavioral perspective of psychology and educational psychology (Kantor, 1959; Kantor, 1975), several classifications of teaching performance in didactic situations have been proposed. As a first theoretical proposal and based on the notion of language games to characterize the scientific practice of the psychologist (Ribes, 1993; Ribes et al., 1996), it has been specified that there are five categories to describe the performance of the teacher during the practice of teaching scientific disciplines, including psychology: Cognitive exploration, Explanation of criteria, Illustration, Practice and Evaluation (Carpio et al., 1998). Subsequently, in the work of Ibáñez and Ribes (2001), a classification of teaching performance behaviors can be identified in four dimensions: (1) Clearly expressing to students how the established criterion is achieved, (2) Exemplifying to students the performance that leads to the achievement of the specified criterion (or criteria), (3) Arranging the necessary situations for the student to practice the ways in which the specified criterion is achieved, and (4) Informing students promptly about their successes or errors throughout the process of applying knowledge.

Later works of psychologists under an interbehaviorist perspective in educational psychology, were consolidating a model of didactic performance, mainly in the context of secondary and university education, of up to seven criteria or dimensions of the didactic performance of the teacher: Didactic Planning, Competency Exploration, Criteria Prescription/Explicitness, Illustration, Supervision/Supervised Practice, Feedback, and Evaluation (Ribes, 1993; Morales et al., 2013; Morales et al., 2017; Reyna and Hernández, 2017). These specific criteria or categories of teacher didactic performance would imply two general categorical elements that define it, the first referring to the teaching modality of the referent object, and the second referring to the evaluation of student learning progress. However, neither the seven specific criteria of teacher didactic performance nor the two generic categories had been empirically validated in classroom situations or didactic interactions.

A study with observational records and student self-reporting of teacher performance in high school science classes in Mexico (Bazán-Ramírez et al., 2022) constituted a first contribution of empirical evidence to validate the relevance of five criteria or dimensions of teacher didactic performance during classes: Exploration of competences, Explicitation of criteria, Illustration, Feedback and Evaluation. Bazán-Ramírez et al. (2023), observing Peruvian university students of biological sciences, obtained convergent and divergent validity of constructs of a self-report and confirmed the multidimensional structure of the teacher’s didactic performance with six criteria: Competence exploration, Explicitness of criteria, Illustration, Practice supervision, Feedback and Evaluation.

With Peruvian graduate students in educational sciences, construct validity was obtained for a restricted model of five criteria of teacher didactic performance with in-variance by sex, age and level of studies (Bazán-Ramírez et al., 2022). Likewise, by means of second-order models, they confirmed the structure of the two generic components of Teaching and Evaluation. In the first case, Teaching performance was formed by two variables: Explicitation of Criteria and Illustration. In the second case, Formative evaluation was constituted by the variables: Monitoring (supervision), Feedback and Assessment. With Peruvian university students of professional careers in sciences, engineering, humanities and literature, social sciences and health sciences, at the beginning of in-person classes after the COVID-19 Pandemic, Cuellar-Quispe et al. (2023) validated these constructs of explanation of criteria, illustration, supervision of learning practices and activities, feedback, and evaluation. The research findings show that there is good virtual teaching performance, and that university professors demonstrate their ability to teach in a hybrid way.

Based on the theoretical approaches and these empirical studies derived from them, Figure 1 characterizes an interactive model of didactic performance, thus ex-tending the restricted models validated by Bazán-Ramírez and colleagues presented in the previous paragraphs. For the specific case of teacher performance, this model includes two generic domains (second-order factors) related to each other: Teaching and Formative evaluation, which configure specific didactic teaching performance criteria. These domains also demarcate the possibility of structuring morphologically and functionally different but interrelated interactive episodes and involve identifying the organization of interactive didactic performances in two generic categories: teaching and feedback. Black bidirectional arrows mark the relationship of six didactic performance criteria that are directly functionally related to each other; however, blue arrows illustrate the establishment of other functional relationships between the criteria or areas of didactic teacher-student interaction.

It should be noted that the first criterion of the teacher’s didactic performance, didactic planning, even though it is carried out prior to the development of instruction through the various didactic and interactive sequences, always influences them, and is functionally related to the other teacher’s performances, both in teaching and formative evaluation. The results of these performances of the teacher in the didactic behavior of his students will also allow the teacher to make the necessary adjustments, oriented in the didactic interaction and in the learning needs of his students and their individual and group characteristics and capabilities.

The first two questions on the content and construct validity of a scale of self-evaluation of university teachers on their own teaching performance, under a psychological model of teaching performance in the university context, justify the importance for the development of educational psychology, regardless of the theoretical root model, to recover in current research on measurement and evaluation of behaviors, the integration of substantive theory in psychological measurement (Mislevy, 1993; Cole, 1993; Bejar, 1993; Martínez and Moreno, 2002). Thus, obtaining content and construct validity in the self-assessment of teaching performance also implies testing the validity of the categories and criteria derived from a particular theory of university teachers’ teaching performance.

On the other hand, question 3 on the invariance of the self-assessment of teaching performance is justified because a self-assessed behavioral measurement instrument must be able to withstand differences of origin, such as gender, nationality and initial professional training of the subjects, in order to explain differences in the teaching performance of university teachers. Likewise, the fourth question is justified due to the importance of personal factors, such as gender, initial training, and the geographic context where they carry out their teaching activity, in explaining the didactic performance of university teachers.

Cross-sectional study of a methodological, instrumental (Ato et al., 2013), predictive and comparative nature (Ato and Vallejo, 2015).

A total of 203 university professors participated, of which 64 professors taught at a public university in northwestern Mexico; 38 taught in Education Sciences and 26 in Psychology. The other stratum of 139 professors, from two public universities in Peru, taught in Education (54) and Psychology (85). As shown in Table 2, of the total sample, 128 were women and the teachers’ ages ranged from 27 to 76 years with an average age of around 50 years. In terms of academic training, 18 teachers only had a bachelor’s degree, 99 had a master’s degree and 86 had a doctorate.

Only those who agreed to give informed consent were included in the study and teachers from other professions who filled out the instrument virtually or presented inconsistent or atypical response patterns according to the multivariate Mahalanobis distance statistic were excluded.

We took the seven categories or criteria of the interbehavioral model of teaching performance (Kantor, 1975; Carpio et al., 1998), and derived statements (items) that allow teachers to assess themselves with respect to these competency criteria of teaching performance. Each of these seven criteria is described below:

Self-assessment of teacher didactic performance. Following the didactic performance model developed by interbehavioral psychologists (Carpio et al., 1998; Irigoyen et al., 2011; Silva et al., 2014; Ribes, 1993; Morales et al., 2017; Velarde-Corrales and Bazán, 2019; Bazán-Ramírez et al., 2022), a self-assessment questionnaire of the teacher’s didactic performance was designed in seven categories (dimensions or performance criteria). This questionnaire is composed of 28 statements (items), with a response scale: 1 = Never, 2 = Almost never, 3 = Almost always and 4 = Always Its consisting of 28 statements (items), with a response scale: 1 = Never, 2 = Almost never, 3 = Almost always and 4 = Always. Table 3 presents each item according to the didactic performance dimension, in which the teacher must choose one of the four response options, according to the frequency with which he/she behaved in correspondence with the statement, during his/her classes in the semester just concluded.

The information collection process was carried out by means of an online form between March and May 2024, both in Mexico and Peru. The online form explained the objectives of the research, the ethical safeguards for the care of personal data, the no-penalty withdrawal and the need for voluntary participation expressed by selecting the corresponding option on the form. In this option the informed consent was provided online which after reading each participant could sign the commitment by marking the option I accept or the option I do not accept. Once the virtual applications were concluded at the end of May, the information from Google forms was retrieved in the EXCEL program and the database was prepared to be exported to the SPSS and R programs.

The descriptive analysis of the items, as well as the descriptive analyses of the self-assessment of the teacher’s teaching performance for each criterion, and the comparative analyses of the general teaching performance and its second-order factors were performed with SPSS software version 25 for Windows. In the contrast of means with Student’s t-test and one-way analysis of variance, effect sizes were assessed with Cohen’s d (0.10 small effect, 0.50 medium effect, 0.80 large effect) and omega squared (0.01 small effect, 0.06 medium effect, 0.14 large effect) (Cohen, 1992).

As for the evidence of content-based validity of the questionnaire, it was quantified with the Aiken V coefficient from the assessment of 7 judges on the criteria of clarity, pertinence and relevance of the items (Sireci and Faulkner-Bond, 2014). The Aiken V values vary between 0 (total disagreement among the judges) and 1 (total agreement among the judges). A satisfactory evaluation of the items on each criterion is assumed if the point estimate of Aiken’s V is ≥0.70 and for the lower limit of the 95% confidence interval values greater than 0.59 (Penfield and Giacobbi, 2004).

For validity evidence based on the internal structure of the construct used the confirmatory factor analysis strategy, in which a 7-factor multidimensional model was initially examined, followed by a second-order factor model with two generic factors (teaching and formative assessment) containing in its configuration the specific teaching performance criteria (primary factors). As the items are categorical, the variance-adjusted weighted least squares (WLSMV) estimation method based on poly-choric correlation matrix was used (DiStefano and Morgan, 2014). For the evaluation of the fit of the models examined, the chi-square goodness-of-fit test (χ2) was taken into account, which must offer a p-value greater than 0.05 to be considered a good fit, as well as the recommended robust fit indices such as CFI (Comparative Fit Index), TLI (Tucker Lewis Index), RMSEA (Root Mean Square Error of Approximation) and SRMR (Standardized Root Mean Square Residual). If CFI and TLI present indexes ≥0.90 they indicate adequate fit and good fit if they are ≥0.95, as for RMSEA it is considered adequate fit when its index is ≤0.08 and good fit when it is ≤0.05, likewise if SRMR presents an index ≤0.08 it means that it presents adequate fit and if it is ≤0.06 it means good fit (Penfield and Giacobbi, 2004; DiStefano and Morgan, 2014). These analyses were run with R program version 4.0.2 and RStudio environment with Lavaan 0.6–7 and semTools 0.5–3 packages.

The evidence of internal consistency reliability was estimated by taking into consideration the nature of the categorical graded items. For the didactic performance criteria scores and the overall questionnaire score, the ordinal alpha, McDonald’s omega and composite reliability coefficients were estimated. For the latent constructs (second-order factors and general factor) the H coefficient was also estimated (Hu and Bentler, 1999; Kline, 2015).

Both Mexican and Peruvian participants were informed about the objectives, benefits and negative risks of the research. They were asked to read the informed consent form, and if they wished to participate in the study, and if they wished to participate in the study, they were asked to sign it, as well as the authorization for the publication of the derived data, after an explanation of the ethical considerations of the study. In the case of the application in Mexico, the informed consent was printed and signed in person; in the case of the application in Peru, the consent was obtained online and signed virtually.

For both countries, the privacy and confidentiality of the data was protected. The data collected were used only for the purposes of this research. The dignity and rights of the participants were respected, avoiding any form of coercion or prejudice. Finally, the research was handled with integrity and honesty. The data are presented in an accurate, transparent and unbiased manner.

As shown in Table 4, according to the values of skewness and kurtosis, the items do not present normal distribution because there are values greater than ±1.5. The table also shows that the corrected homogeneity indexes (CHI) are ≥0.30 for all items, indicating that the items have a high discriminative capacity (DiStefano and Morgan, 2014).

According to the results of Aiken’s V (Table 5) all the items satisfy the criteria of clarity, pertinence and relevance because the point estimates are greater than 0.70 and the lower limits of the 95%CI are greater than 0.59, the critical values recommended as satisfactory (Cohen, 1992). Therefore, the data support the evidence of content-based validity of the teacher didactic performance self-assessment questionnaire.

Figure 2 presents the multifactorial model of teacher didactic performance examined with confirmatory factor analysis. The estimator used was WLSMV (Diagonally Weighted Least Squares with Mean and Variance corrected), which is recommended when there are categorical items.

The overall evaluation of the multifactor model was satisfactory according to the fit indices: χ2(329) = 411.07, p = 0.001, CFI = 0.96, TLI = 0.95, RMSEA = 0.05 [0.03, 0.06], SRMR = 0.07. Likewise, the evaluation of the individual parameters evidenced high standardized factor loadings (>0.50). These results support the existence of validity based on the internal structure of the construct.

It can also be seen in Figure 1 that almost all the factor loadings are greater than the covariances between the factors, which means that there is divergent and convergent validity. It can also be seen in Figure 1 that all the factor loadings are high (>0.50), implying the existence of convergent validity (Kalkbrenner, 2021). As can be seen in Table 5, almost all of the values of the square root of Average Variance Extracted (√AVE ≥ 0.74) are greater than the interfactor correlations of the multidimensional model (more than 80% are less than 0.74, see Figure 1), which means that the evidence of discriminant validity is ac-acceptably met (Muñiz et al., 2005).

Figure 3 presents the model resulting from the interactive analysis between general second-order factors (Teaching and Formative Assessment Categories) that configure the specific didactic performance criteria. A strong covariance (ϕ = 0.96) can be appreciated between the second-order factors.

In accordance with the theoretical approach, the interactive episode teaching (first second-order factor) contains four primary factors (didactic performance criteria: instructional planning, competency exploration, explicitness of criteria and illustration), according to the path coefficients, the criterion with the greatest variance explained corresponds to explicitness of criteria (R2 = 0.90) and the criterion with the least variance explained corresponds to Illustration (R2 = 0.39). Likewise, the second interactive formative evaluation episode (the other general factor of second order) contains in its structural configuration three primary factors (didactic performance criteria: supervision of practice, feedback and evaluation) highlighting the criterion supervision of practice with the greatest variance explained (R2 = 0.84) and the criterion with the least variance explained corresponds to evaluation (R2 = 0.61).

It is also observed in Figure 3 that all the primary factors are configured by four items as established by the theoretical design. The relationships between the primary factors and the items in all cases present high factor loadings (λ > 0.50). The interactive model presents indices that denote the presence of good fit: χ2(342) = 446.804, p = 0.000, CFI = 0.95, TLI = 0.94, RMSEA = 0.05 [0.038, 0.066], SRMR = 0.08. All these results support satisfactory evidence of validity based on the internal structure of the construct for a second-order model.

Having established the factor structure of the self-assessment questionnaire of teacher didactic performance, the measurement invariance was analyzed according to sex, nationality and professional discipline of origin. Table 6 shows that the configural invariance (unrestricted model for parameter estimation) presents a very satisfactory fit for the three sociodemographic conditions (CFI > 0.95 and RMSEA <0.08). As for the nested models (metric, scalar and strict) the recommended values of ∆CFI (> − 0.01) and ΔRMSEA (<0.015) denote equally the fulfillment of invariance for the three conditions.

Table 7 shows the reliability coefficients estimated by internal consistency for categorical data. The reliability for the overall score of the questionnaire is higher than 0.93 according to Cronbach’s alpha ordinal and McDonald’s omega coefficients, and the re-liability for the construct is of high precision (H > 0.95). On the other hand, the reliability for the interactive episodes (second order factors) evidence high precision for the scores because the ordinal alpha and omega internal consistency coefficients are ≥0.90; likewise, these two general factors as latent constructs are of high reliability according to coefficient H (≥0.94). Finally, the reliability for the seven factors [didactic performance (criteria) varies between 0.77 and 0.88 according to McDonald’s ordinal alpha and omega coefficients], denoting the presence of high precision for their scores.

Table 8 identifies the self-assessment of didactic performance in the competency criteria taking into consideration the lower and upper limits of the CI95% of the mean, the coefficient of variation and the standardized Z scores. Lower values in the coefficient of variation (CV) indicate greater homogeneity while higher positive Z-scores indicate greater presence of the measured attribute.

According to the context, the competency criteria in which teachers stand out are in the explicitness of criteria, instructional planning and supervision of practice. And the competencies in which strengthening is required are competency exploration, illustration, feedback and evaluation.

Table 9 presents the results of comparative analyses with the analysis of Student’s t-test. In the comparison according to the professional training discipline and the gender factor, significant differences (p < 0.05) were observed for a small effect size (d > 0.20), with higher self-evaluation scores for teachers of education and women in general didactic performance, as well as in the teaching and formative evaluation factors. On the other hand, in the comparison by nationality, no differences were observed (p > 0.05, d < 0.20) in general didactic performance or in the general factors of teaching or formative evaluation.

Table 10 presents an ordinal regression model to evaluate the impact of the factors gender and professional training discipline on the probability that university teachers have a regular, acceptable or optimal teaching performance.

The estimated logistic model is statistically significant (χ2 = 18.840, gl = 2, p < 0.001), which means that personal factors predict the occurrence of didactic performance categories. Likewise, Pearson’s Chi-square goodness-of-fit tests (χ2 = 6.976, gl = 4, p = 0.137) and the Deviance (χ2 = 7.050, gl = 4, p = 0.133) evidence good fit of the data. On the other hand, the parallel lines test presented satisfactory results for the fulfillment of the proportionality assumption (χ2 = 1.356, gl = 2, p = 0.508).

Nagelkerke’s Pseudo R-squared provides a 10.2% explanation for the overall logistic model of two predictors. Among university professors, specifically women are 2.05 times more likely to have an optimal didactic performance compared to men. Regarding the discipline of professional training of origin, professors of education are 2.62 times more likely to achieve optimal didactic performance in contrast to professors of psychology.