Principal-Teacher Trust Is Linked to Teaching Quality Through Innovation and Learning Beliefs

What actually changes what happens inside a classroom? For decades, education researchers have pointed to school leadership as a key ingredient, yet the machinery connecting a principal’s relationship with teachers to teachers’ daily instructional choices has remained difficult to trace. A cross-sectional study from Türkiye, published in Frontiers in Psychology, examines that machinery using responses from 1,185 teachers spanning all twelve of the country’s statistical regions. The research reports that the quality of the relationship between school principals and teachers is associated with instructional practice mainly through two channels: how innovative teachers perceive their colleagues to be, and how confident teachers feel about their own capacity for professional learning.

The study focuses on a construct known as leader-member exchange, or LMX, which describes the quality of the relational bond between a leader and an employee. Grounded in social exchange theory, LMX comprises four dimensions: affect, meaning the emotional bond and mutual trust between principal and teacher; contribution, the perceived quantity and quality of effort each party invests toward shared goals; loyalty, the willingness to support the leader in the face of challenges; and professional respect, the esteem each holds for the other’s competence. High-quality exchanges, the theory holds, motivate employees to invest discretionary effort beyond their formal duties. The question the study poses is whether that discretionary effort is associated with better instructional practices, such as capturing students’ interest, ensuring active participation, and tailoring group work to student needs.

To answer it, the researcher tested a structural equation model linking LMX to instructional practices through two parallel mediators. The first, collective teacher innovativeness, captures a teacher’s perception of how open to change most teachers in their school are, based on items used in the 2018 TALIS survey. The second, learning effectiveness, reflects a teacher’s belief in their own motivation, ability, and perseverance as a learner, measured with the Learning Effectiveness subscale of the Teacher Agency Scale. The two constructs are conceptually distinct: one concerns what a teacher perceives colleagues to be doing, the other what a teacher believes about their own learning. Both were measured at the individual level, an important qualification, since the study did not collect school identifiers and therefore could not aggregate perceptions into genuine school-level characteristics.

The sample was built through proportional regional stratification across Türkiye’s twelve NUTS-1 regions, with invitations allocated in proportion to each region’s share of the national teacher population. Because no individual-level sampling frame covering all 1.19 million teachers was available, participation within selected schools was voluntary, making the sample a regionally stratified volunteer sample rather than a strict probability sample. Still, the achieved regional distribution tracked the national targets closely: no region’s share of the analyzed sample deviated from its share of the teacher population by more than 0.4 percentage points, and a chi-square goodness-of-fit test showed no significant difference. After excluding eighteen multivariate outliers identified by Mahalanobis distance, 1,185 cases remained, 78.4 percent of them female, with an average of 14.15 years of professional experience.

The measurement phase was thorough. The researcher compared four alternative measurement models estimated simultaneously on the same 29 items. A one-factor model, in which every item loaded on a single latent variable, fitted poorly, with a comparative fit index of 0.53, indicating that no single common factor, whether substantive or method-related, could account for the covariation among the items. A four-factor model treating LMX as unidimensional also fitted poorly. The seven-factor and second-order models both fitted acceptably, and the second-order specification, in which the four LMX dimensions serve as indicators of a single higher-order construct, was retained for being more parsimonious. Discriminant validity was assessed with the Fornell-Larcker criterion and heterotrait-monotrait ratios, all of which fell below the 0.85 threshold.

The structural results tell a notable story. Leader-member exchange showed only a small direct association with instructional practices, a standardized coefficient of 0.10. Its associations with the two mediators were larger: 0.34 with perceived collective innovativeness, the second largest coefficient in the model, and 0.25 with learning effectiveness. In turn, learning effectiveness carried the strongest association with instructional practices in the entire model, a moderate 0.41, while perceived collective innovativeness showed a smaller but reliable 0.21. The two indirect routes together carried a total indirect effect of 0.17, meaning roughly 63 percent of the overall LMX-instructional practices association, which totaled 0.27, flowed through the mediators rather than directly. The learning route, at 0.10, was about one and a half times the size of the collective route at 0.07, though the overlapping confidence intervals mean this difference is descriptive rather than formally tested.

Bootstrap analysis with 5,000 resamples supported both indirect pathways: the 95 percent confidence intervals for both specific indirect effects excluded zero. The full model explained 28.8 percent of the variance in instructional practices, a substantial proportion by conventional standards, though shared among five predictors. Leader-member exchange alone explained 11.6 percent of the variance in perceived collective innovativeness and 6.3 percent in learning effectiveness, indicating that most variation in the mediators lies outside the model. Among the control variables, professional seniority showed a very small negative association with instructional practices, while length of service with the current principal showed a very small positive one; gender and educational level were not associated.

The theoretical reading is a deliberate division of labor between two frameworks. Social exchange theory explains why teachers who experience their relationship with the principal as trustworthy, fair, and reciprocal become willing to invest effort in shared collegial activity and in professional learning whose returns are uncertain and deferred. Organizational learning theory explains how that effort may become changed classroom routine: new pedagogical knowledge must be created, shared, tried, and adjusted before it can alter established practice. Neither theory alone predicts the full chain. The pattern that emerges, with the direct LMX-practice link weakest and the learning-effectiveness link strongest, is consistent with this two-theory account, suggesting that relational leadership is related to teaching mainly through what it accompanies rather than on its own.

The Türkiye context adds institutional weight to the findings. The country’s education system is highly centralized: curricula, textbooks, assessment arrangements, and much of in-service training are determined nationally, leaving principals with limited formal control over classroom practice and teachers with relatively limited curricular autonomy. Under these constraints, the collegial group is one of the few arenas in which teachers can legitimately depart from routine, which may help explain why perceived collective innovativeness carries weight. The study frames this as an institutional rather than a cultural explanation, noting that no measure of cultural orientation was included and that national cultures are neither uniform nor deterministic.

The author is explicit about what the design cannot show. Because all constructs were measured with self-report scales completed by the same teachers at a single point in time, the findings describe associations consistent with the proposed model rather than causal effects. Reverse relationships remain possible: teachers who already use innovative practices and learn effectively may attract more supportive exchanges from their principals. A generally positive school climate could also influence all four constructs simultaneously. Diagnostics such as Harman’s single-factor test and full-collinearity values reduce but do not eliminate common-method concerns, and the unmodeled clustering of teachers within schools may have understated standard errors. The practical implications, including greater emphasis on relationship-building in principal professional development and stronger structural support for school-based professional learning, are therefore offered as areas for policy consideration rather than proven interventions, with any future trials needing independent evaluation.

Subject of Research: Associations between leader-member exchange, collective teacher innovativeness, learning effectiveness and teachers' instructional practices in Türkiye

Article Title: How is leader-member exchange related to teachers’ instructional practices in Türkiye? The mediating role of collective teacher innovativeness and learning effectiveness

Article References: Doğuş, Y. (2026). How is leader-member exchange related to teachers’ instructional practices in Türkiye? The mediating role of collective teacher innovativeness and learning effectiveness. Frontiers in Psychology, 17, Article 1960677. https://doi.org/10.3389/fpsyg.2026.1960677

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1960677

Keywords: leader-member exchange, school leadership, instructional practices, collective teacher innovativeness, learning effectiveness, teachers, Türkiye, structural equation modeling, social exchange theory, organizational learning theory, mediation analysis, leader-member