Background: Research has shown that teacher feedback is critical for student learning and well-being (Siegle & McCoach, 2007; Wisniewski et al., 2020). However, teachers often provide more positive feedback to students from high socioeconomic backgrounds than to those from low socioeconomic backgrounds (Rimm-Kaufman et al., 2000; Tobisch & Dresel, 2017). This tendency is potentially problematic because it can have significant implications for student achievement and perpetuate socioeconomic inequalities. This paper aims to investigate the effects of positive teacher feedback on student self-efficacy, a critical predictor of student motivation, well-being, and academic achievement (Stevens et al., 2006). By examining the relationship between positive feedback and self-efficacy, this study seeks to provide insights into how teachers can promote a more equitable learning environment for all students. Purpose: While previous studies have examined the impact of teacher feedback on student self-efficacy, these studies are often based on small, nonrepresentative samples (Au, 2020; Ruegg, 2018), pointing to the need for research that draws on representative samples to improve the generalizability of the findings. This study aims to contribute to the literature on the effects of teacher feedback on student self-efficacy by using nationally representative data from PISA 2018 to gauge the effects of positive teacher feedback on student self-efficacy. Research Design: This paper uses a potential outcomes framework for causal inference (Morgan & Winship, 2015). The potential outcomes framework is based on the concept of counterfactuals, which refer to hypothetical situations in which the outcome of interest would be different if the intervention had not been applied. The difference on an outcome variable Y between individuals who receive the treatment and those who do not is the average treatment effect (ATE). The expected change in Y due to treatment for members in the treatment group is the average treatment effect for the treated (ATT). The expected change in Y due to treatment for members in the control group is the average treatment effect for the control group (ATC). We use propensity scores to estimate each individual's probability of receiving the treatment based on covariates (Morgan & Winship, 2015; Pearl, 2000) and use matching and weighting to ensure that the results of the analysis are not biased by the confounding variables. Data and Analysis--Data & Variables: Data used in this study comes from the Programme for International Student Assessment (PISA) 2018. This paper focuses on the 4,838 students across 215 schools in the US sample, which represents 3,559,045 15-year-old students (OECD, 2019a). The outcome variable of interest is students' reading self-efficacy, which is measured by one item in the student questionnaire, "I am able to understand difficult texts." The causal variable is positive teacher feedback, which is measured by one item in the student questionnaire, "The teacher gives me feedback on my strengths in the (test language) subject." The covariates include student gender, socio-economic status, reading achievement, enjoyment of reading, competitiveness achievement motive, well-being, peer relationships, parental encouragement, and student-teacher relationships. Analytical Plan: We first present the estimate of the effects of positive teacher feedback on student reading self-efficacy by regressing student reading self-efficacy on positive teacher feedback, controlling for the covariates. To account for the complex survey design and to adjust estimates for the small amount of non-random missing data, two types of weights were applied when obtaining regression estimates: Final student weights (W_FSTUWT) and balanced repeated replication weights (2019b). Figure 1 presents the causal graph that models the theoretical relationship used for identifying covariates X to generate propensity scores for treatment assignment. To estimate propensity scores, we run a logistic regression model regressing treatment assignment on covariates X. We then offer matching estimates of the ATT and ATC using nearest neighbor matching based on estimated propensity scores. Next, we use inverse probability weighting (IPW) based on the propensity scores and the student-level final weight to generate weighted regressions estimates of the ATT and ATC. Last, we estimate the weighted regressions estimates of ATT and ATC after restricting the estimation sample to the region of common support on the propensity score. Results: Table 1 presents the descriptive statistics for the variables in the sample. Table 2 shows the logistic regression used to generate the propensity scores. Table 3 compares the balancing of treatment and control groups with and without using IPW. The weighted samples are more similar on the covariates than the unweighted sample, indicating that IPW successfully reduced potential confounding. Nearest neighbor matching produced ATT and ATC estimates of 0.07 (SE = 0.03, p < 0.05) and 0.10 (SE = 0.04, p < 0.01), respectively, which were slightly smaller than the weighted regression estimates. Weighted regression estimates with and without restricting the estimation sample to the region of common support were statistically indistinguishable, with an estimated ATT of 0.10 (SE = 0.02, p < 0.001) and estimated ATC of 0.12 (SE = 0.03, p < 0.001). Across all matching and weighting estimates, the estimated ATTs were smaller than the estimated ATCs, indicating that positive teacher feedback had a larger effect on reading self-efficacy for students who received positive teacher feedback less often than for students who received it more often. Missing data information is provided in Appendix A. Conclusion: Building on the extensive research that supports the positive impact of teacher feedback on student outcomes (Au, 2020; Hattie & Timperley, 2007; Ruegg, 2018; Wisniewski et al., 2020), this study finds compelling evidence that positive teacher feedback has a significant and positive effect on student reading self-efficacy. Moreover, the effect is larger for students who receive positive teacher feedback less often, which are often those from disadvantaged background (Rimm-Kaufman et al., 2000; Tobisch & Dresel, 2017). These findings underscore the importance of teachers' feedback practices and suggest that teachers can contribute to a more equitable learning environment by offering positive feedback regardless of students' socioeconomic backgrounds.