Relatively little attention has been paid to the importance of principals in the production of student achievement. Most of the research has been done for teachers, and for good reason, as they directly affect student achievement. However, this is not to say that research on principals should be neglected. The economic question of interest is whether a good quality principal is better than hiring more mid-level teachers, as a better principal will provide a more efficient allocation of teachers towards students with different levels of achievement. Dhuey and Smith (2014) set out to find this relationship using principal fixed effects from a value-added model of using differenced FSA test scores from the same students in grade 4 and grade 7. The identification strategy was simply a policy rule set out by the Ministry of Education in British Columbia which rotated principals frequently through school districts. To make use of this policy rule, Dhuey and Smith carefully built an empirical model that accounted for all sources of endogeneity from principal mobility. Moreover, the fixed-effect model controlled for time-invariant characteristics such as ability, leadership style, and gender as, theoretically, they explain test score outcomes. Furthermore, controlling for school fixed effects, the authors use an interaction term that captures non-random rotation of principals in school districts that occur through matching of high value-added principals with low value-added schools. The interaction term is defined as a three-way interaction variable with the school, grade 5 and 6 principal. This variable, along with other neighbourhood and school-level factors should produce a causal effect of the principal effect on student achievement. Another possible source of endogeneity arises from observing a pseudo-principal effect. This is defined as observing a loss in gain test scores from an outgoing principal when it was nothing but a reversion to the mean test scores. For robustness, the authors check if there exists a turnover effect from previous lagged gains or losses in test scores. The outcome is, when large gains for two years in level test scores are observed, there exists a positive and statistically significant effect of 0.071 at the 5% significance level. Furthermore, there exists a weaker negative effect in turnover when large one-year losses in gain test scores are observed, however, this is only significant at the 10% significance level. However, most coefficients in this table are statistically insignificant, and the authors argue that the effect is not strongly supported. Finally, the results, on average, one standard deviation improvement in principal quality can boost student performance by 0.289 to 0.408 standard deviations in reading and math scores, respectively. Furthermore, the difference between a “good” principal and a median principal (principal at the 75th percentile – 50th percentile) improves test scores in reading and math by 0.170 and 0.193 standard deviations.
For robustness, the authors postulate that experience and tenure (length of stay in current school) does not affect the average principal if ability is accurately controlled for. The authors find no statistically significant effect in experience and tenure on test scores, and after controlling for school fixed effects, the effect was still insignificant. The conclusion drawn upon is, to boost the performance of underperforming students and reduce achievement gaps, efficient allocation of high ability principals to low-ability schools rather than using experience as a proxy for ability. Dhuey and Smith (2014) highlight some important questions, how can researches identify the best principals, and what are the fixed characteristics that make principals better than others? The goal of this paper is to provide answers to these questions in hopes to come to a policy conclusion that utilizes the main findings of the referenced research.
Dhuey and Smith (2018) later published a paper to address the questions listed above, titled how school principals influence student learning. Using data from grade 3 to grade 8 students in North Carolina, the principal effect is 0.17 and 0.12 in math and reading scores, respectively. Furthermore, using a similar school fixed effect interaction term, the authors find the principal effect shrinks to 0.04 and 0.02 standard deviations in math and reading scores, respectively. The authors try to address the reason in which high-value principals differ from low-value ones. Having an advanced degree increases value-added in reading and having a non-competitive doctorate increases math scores. This addresses one of Dhuey and Smith’s questions from the initial paper, suggesting that allocating principals of having advance degrees and non-competitive doctorates to low-value-added schools would lead to a more efficient allocation of principals. Perhaps, this would be a sufficient proxy for identifying high-ability principals. Dhuey and Smith (2018) find that incoming principals with no prior experience reduce attendance rates for students and the proportion of teachers with 11 years or more of experience. Furthermore, they are more likely to hire teachers with 0-3 years of experience. This result is interesting as it contrasts their initial work, due to finding no effect of experience on the average principal. However, in this study, there is a negative effect that arises from principals who have no prior experience. More research should be done relating to principals with advanced degrees and with no prior experience, as the research presented thus far suggests advance degrees to be a possible proxy for ability and experience and tenure should not affect the average principal, controlling for ability. The authors highlight a similar point, which states that without considering match effects in the movement of principals, there is a possibility to create a ‘zero-sum game’, where good principals benefit students, while bad principals negatively affect students, resulting in average achievement unchanged. As a result, if the school boards want to increase student achievement, an efficient allocation of principals is needed to raise overall student achievement.
The previous paragraph contradicts the results of Clark, Martorell, and Rockoff (2009) (hereafter, CMR) who find that education level is not likely to be a factor, rather training programs to be a factor. Furthermore, the results of CMR (2009) contrast Dhuey and Smith’s (2018) findings that experience likely does play a role in student achievement of principals. However, there is a positive effect of having a principal who previously held the title as an assistant principal in the same school.
The results of CMR (2009) do not provide the same robustness check as Dhuey and Smith (2014) do. After controlling for ability, experience plays no role for the average principal, however, CMR (2009) find inconsistent results. The difference is, perhaps CMR (2009) have not accurately controlled for ability in the same way Dhuey and Smith have done in their 2014 paper. Both papers, however, are controlling for school fixed effects using similar methods. Interestingly, CMR (2009) reports that the impact of school principals is felt during the first few years of principal tenure, suggesting most of the changes and initiatives to impact student achievement to be captured in this effect, rather than over time. Walsh and Dotter (2014) explore the effects in which the replacement of the current principal affects student achievement. This is of interest as it relates to the results of CMR (2009) and provides a research background for principal mobility policy rules which are implemented in British Columbia and North Carolina. The goal is to show if replacing poor-performing principals leads to positive student achievement outcomes. The authors find that after three years with the new principal reading scores increases by 4 percentage points or 0.09 standard deviations, with math scores also increasing as well, but this effect was not statistically significant. In contrast, the authors use a difference-in-difference methodology to search for the causal effect of principals on student achievement. Intuitively, it controls for unobservable in a similar method to the fixed effects model the previous authors used, but also, better highlights the pronounced effect of new incoming principals on student achievement. This research complements the work of Dhuey and Smith (2014) and CMR (2009). Dhuey and Smith (2014) postulate that the efficient allocation of teachers is the best way to raise student achievement and highlight possible inefficiencies that arise from non-matching mobility of principals. Moreover, Walsh and Dotter (2014) find that replacing poor-performing principals, thus promoting efficient matching, leads to an increase in student reading scores. However, the effect of math scores is inconclusive. Furthermore, CMR (2009) suggests that most of the effect on student achievement occurs early on the principal’s tenure. Walsh and Dotter (2014) find that the effect occurs within three years, as opposed to immediately. Interestingly, the research of Dhuey and Smith (2014) uses FSA test scores, which coincidentally, are performed every three years. Without forcing conclusions, and assuming homogeneous assumptions, this further supports their research, suggesting they accurately captured the main effect of principals in this period.
Research conducted from these papers has several main points that should be considered in policy implementation. Dhuey and Smith (2014) find that there is a positive and statistically significant effect of principals on student achievement, and moreover, is not affected by experience or tenure, but by ability. Furthermore, in their 2018 paper, they find that the effect of a poor-quality principal or one with no experience has strong negative effects on student achievement. Moreover, Walsh and Dotter (2014) find that replacing poor-quality principals leads to a positive effect on reading test scores within three-years of tenure. Dhuey and Smith (2018) find that a principal who holds an advanced degree improves reading scores, and a non- competitive doctorate improves math score. While CMR (2009), instead finds education to have no effect on student achievement, however, training programs have a positive effect on student achievement.
The main idea behind the research covered thus far is to promote principal mobility due to the positive effect of incoming high-quality principals. However, it should be based on an efficient matching of principals and schools, instead of the focus on experience, as it was previously in British Columbia, it should be on ability. Credentials such as training programs, or advance degrees could be proxies for this. An important consideration is that schools should stay away from principals with no experience, as this has negative effects on student achievement, Dhuey, and Smith (2018). Therefore, low value-added schools should stay away from such principals. However, a better policy consideration, taken from the results of CMR (2009), is to assign principals with no experience to assistant principals and promote them to the title of principal after the mobility of the outgoing principal. More efficiently, this can be done using new graduates from top training programs or with advance degrees matched to schools that require high-value added principals. This strategy utilizes findings from all papers covered in search for the most efficient allocation of principals. However, this may inconclusive, as Dhuey and Smith (2018) state, due to longitudinal data restrictions, identifying the true characteristics which identify ability, or effective principals is difficult, to say the least.
Clark, D., P.Martorell and J.Rockoff, 2009. “School Principals and School Performance.” Urban Institute Working Paper No.38. Washington, DC: Urban Institute
Walsh, E., and D.Dotter, 2014. “The Impact of Replacing Principals on Student Achievement in DC Public Schools,” Mathematica Policy Research Reports, Mathematica Policy Research.
Dhuey, E., and J.Smith, 2014. “How important are school principals in the production of student achievement?,” Canadian Journal of Economics, Canadian Economics Association, vol. 47(2), 634-63
Dhuey, E., and J.Smith, 2018. “How school principals influence student learning,” Empirical Economics, Springer, vol. 54(2), 851-82