Do charter schools alleviate the negative effect of teacher turnover?

Abstract

Using data on charter and public school districts in Texas, we test the hypothesis that the labor practices in charter schools, in particular their ability to easily dismiss poorly performing teachers, diminishes the negative effect of teacher turnover on student achievement and graduation rates in comparison to public schools. We find some support for this hypothesis, and discuss implications for theory and practice.

Keywords

charter schools graduation rates labor practices test scores Texas

Introduction

Since the passage of the No Child Left Behind Act in 2001, US state governments have intensified attempts to introduce market-based reforms into the system of primary and secondary education (Vergari, 2007) in an effort to meet increasingly demanding levels of student achievement mandated by the law. The policymakers responsible for these market-based solutions believe that consumer choice and competition among schools will increase educational innovation and liberate schools from ‘burdensome bureaucratic regulations’ in the public sector (Lubienski, 2003), although no clear research supports these notions (Schneider, 2016). These trends are not unique to the USA: the shift towards different forms and extents of privatizing education is global (Ball, 2012; Verger et al., 2016), and charter schools are just one instance of neo-liberal, market-driven reforms in education.

The proliferation of charter schools is a prominent aspect of such ongoing reforms in American education. Although the details of charter schools and charter policies may differ across states (Malloy and Wohlstetter, 2003; Podgursky and Ballou, 2001), charter schools are publicly-funded but privately-owned institutions which are free to operate separately from traditional government regulations or bureaucracy (Vergari, 2007). Among the most dramatic aspects of deregulation in charter schools has been the creation of an entire educational sector in which tenure for school teachers is not available. Educators in charter schools are generally employed ‘at will’, both as a means to reduce labor costs and to easily terminate underperforming teachers (Bankston et al., 2013; Stuit and Smith, 2012). Charter school teachers are, unlike public school teachers, also generally not required to maintain teaching certification, but are often expected to work longer hours than their public-school colleagues, lack long-term employment protection such as tenure, and tend to be given raises on a merit or pay-for-performance basis, rather than salary schedules, as is the case in many public schools (Podgursky and Ballou, 2001).

The unifying assumption behind these practices is that they facilitate the reward of good teachers, and permit the easy dismissal of underperforming teachers. Because teacher quality affects student achievement, we should see both higher turnover in charter schools, as well as higher student achievement. Do any of these fundamental differences between charter and public school labor policy affect academic outcomes for students? The goal of this article is to determine if teacher turnover in charter schools may represent a healthy optimization of the teacher workforce. Although the general direct negative effect of turnover is unlikely to disappear, it is likely that the negative effect of turnover on student achievement is dampened in charter schools, compared to traditional public schools.

Literature review

Charter schools vs traditional public schools

The relationship between educational labor policies and student achievement is insufficiently researched and understood. The lack of sufficient research in this area is in part owing to the difficulty in gathering data on educational labor conditions (as few organizations are likely to collect granular data on labor conditions unless specifically required to do so), as well as the difficulty of separating teacher and school effects from the overwhelming impact of non-labor variables (such as socioeconomic status) upon student achievement. The present study is no exception in succumbing to data limitations: although qualitative research attempting to directly analyze the labor practices in question would be ideal, we use known differences between charter and traditional schools (as elaborated in the review below) and multiple control variables, to infer the effect of charter school labor practices on student achievement vs alternative explanations.

The distinct organizational practices of charter schools are known and well-documented, but their effects and implications have not been sufficiently investigated: much of the existing work focuses on elaborating the distinctions, and does not typically address the question of the impact of these practices. For example, Lubienski (2003) analyzed the available material on charter schools and found that, although didactic and pedagogical methods within charter schools ‘tend towards the familiar’ (that is, classroom methods in charter schools are more similar to public schools than they are different), charter schools are organizationally distinct from traditional public schools. Torres (2014) writes that charter schools have higher levels of turnover than traditional public schools – often double the typical rate of turnover. Torres has also uncovered that voluntary turnover in charter schools is relatively low, whereas turnover among longer-tenured teachers is relatively high. If this is indeed the case, the findings suggest that charter school teachers may decline in quality over time (a finding that Torres attributes to workload-related burnout), and consequently these teachers may begin to negatively affect student outcomes as they mature in the organization. On the other hand, Harris and Sass (2011) find a positive relationship between tenure and teacher performance (and thus student performance) in public schools. Podgursky and Ballou, in their study of charter school personnel policies, note that charter schools are ‘fairly aggressive in ridding themselves of ineffective teachers’ (Podgursky and Ballou, 2001: 26), and link their aggressive dismissal rates to the lack of unions and labor bureaucracy.

Teacher quality

Although teacher quality in public and charter schools is not of direct concern for the present study, it is nevertheless an essential important control variable needing elaboration, in order to reduce the possibility of alternative explanations for differences in student outcomes (if any). Most research in the area of student outcomes (in both charter and public schools) tends to focus on the effects of teacher quality upon student achievement and in the traits of successful teachers (Rivkin et al., 2005) or upon the traits of higher-performing teachers and students together (Aaronson et al., 2007). The organizational context in which these traits are put to use are generally seen as a theoretical backdrop rather than a critical part of the puzzle of student achievement.

Teacher quality is an important predictor of student success, as noted by Aaronson et al. (2007) and Rivkin et al. (2005). Low-performing teachers have negative effects on student achievement (Hanushek, 2012). Although ‘teacher quality’ is a loaded and vague term, research suggests that it can indeed be modeled (Collier and Millimet, 2009). For example, it has been approximated through measures such as years of experience, teacher salary, and advanced degrees in specific subject matter areas, and such measures have demonstrated student achievement (Clotfelter et al., 2007; Darling-Hammond, 2000; Figlio, 1997; Hanushek and Rivkin, 2006; Hanushek et al., 2005; Harris and Sass, 2011). Indeed, most scholars in the field of educational administration implicitly recognize that teachers have an effect upon student achievement, both individually and as a group. It stands to reason, then, that a concern with educational policy and improving student achievement should inevitably turn to the role of teachers, and specifically to the labor conditions of teachers. It seems intuitive that educational labor conditions must affect teaching performance in some manner, and that this must, in some sense, have an effect upon student outcomes.

Teacher labor conditions

The effects of labor policy and labor organization upon student achievement can now be studied, thanks to the rise (in both real numbers and relative popularity) of charter schools in the United States, and the sufficient time they have been in operation. Charter schools in the United States have a vastly different set of labor standards and practices than traditional American public schools (Malloy and Wohlstetter, 2003; Podgursky and Ballou, 2001). The differences between labor conditions in charter and public schools, numerous as they are, may have observable effects on teacher performance and, consequently, on student performance. This unique situation – two vastly different approaches to educational labor organization co-existing alongside each other in the same country – provides a laboratory for testing numerous aspects of educational theory, organizational theory, and managerial science and labor policy.

One variable directly connected to labor conditions is the level of turnover. Charter schools in the United States have notably higher turnover among staff, especially teachers (Stuit and Smith, 2012). Some charter schools incorporate high levels of teacher turnover into their organizational design – regarding attrition as ‘a given’ – and tend to regard their high levels of turnover as an aspect of their organizational selectivity, even seeing it as an advantage rather than a problem (Torres, 2014).

This attitude towards turnover on the part of many charter schools tends to contradict traditional educational and organizational scholarship. Much mainstream organizational theory suggests that tenure (that is to say, the opposite of turnover – employees remaining with a particular employer for extended periods of time) is associated with stronger work performance (Ng and Feldman, 2010), and hence, in the case of educational organizations, with increased student achievement. This makes intuitive sense, as teachers are expected to grow in experience (and thus quality) as their years of tenure increase – albeit up to a point, as some have suggested that in the ‘teacher lifecycle’, the effect of experience peaks before end of career (Day and Gu, 2007). Consequently, we would expect high levels of teacher turnover to negatively impact student achievement and organizational effectiveness in educational institutions – especially given that some studies have found that higher-quality teachers are less likely to exit the profession (Krieg, 2006). It seems that a rational school – charter or not – would attempt to reduce turnover among educators as much as possible, at least among the well-performing employees.

Effects of turnover

Management literature tends to have a pessimistic outlook on turnover. Ng and Feldman (2010) suggest that tenure – as opposed to turnover – has a positive relationship with job performance. Watlington et al. (2010) note that turnover of ‘high-quality’ teachers has a generally negative impact upon student performance and school district financial capabilities, although it is unclear from their research if they consider the potentially positive impact of low-quality teachers leaving the teaching profession.

Ronfeldt et al. (2013) offer evidence to suggest that teacher turnover has a negative effect upon math and English test scores in New York City public schools, and attribute this to both the ‘compositional’ effects of turnover (e.g. more-experienced teachers being replaced by less-experienced teachers) and ‘disruptive’ effects (the loss of effectiveness suffered by an organization when experienced employees depart). They also find that the impact of turnover is greater for schools which educate larger numbers of minority and low-income students. Interestingly, they find that turnover in schools appears to have a multi-year effect, even upon the students of those teachers who remain employed. This suggests, at least to some degree, that turnover has a strong impact upon public school organizations as a whole and public school educational delivery as a whole, rather than simply individual classes or students. Others, such as Hanushek (2012), suggest that the ‘cost’ of retaining poorly-performing teachers may have a greater impact on school performance (and hence student performance) than the turnover associated with their termination.

Carruthers (2012) conducted a study on the effect of teacher effectiveness in North Carolina charter schools. She points out that charter schools tend to improve in effectiveness (measured largely by test scores) as they mature, and posits that this may be the result of charter teachers growing in experience or, in her words, ‘development of the faculty’s value added, cohesiveness, and collaboration’ (Carruthers, 2012: 291), and that this development may be fundamentally different than the faculty development in the public sector. Hanushek and Rivkin (2010) have likewise performed studies which reveal that teachers in Texas who leave schools for reasons other than retirement (turnover) tend to perform less well than those who stay.

Although the conventional wisdom in organization studies is that turnover generally harms organizational performance, many suggest that greater organizational flexibility in hiring and firing practices – namely, a greater ability to terminate underperforming employees – may lead to increased organizational performance (Abelson and Baysinger, 1984; Dalton and Todor, 1979; Dalton et al., 1981; Park et al., 1994). Consequently, it is possible that the labor flexibility inherent in charter school’s hiring and firing practices will address teacher performance problems and thus improve student outcomes or, at a minimum, prevent them from worsening significantly as a result of the turnover. As charter schools have more organizational capacity to reduce the numbers of low-performing teachers through termination, from this perspective, turnover is a way to reduce numbers of teachers who are not performing well within the organization. For charter schools, then, the goal may be not so much to retain teachers as a class: it may be more valuable to remove low-performing teachers before they damage the educational attainment of their students. This attitude would seem to fit with the charter school movement’s orientation towards competition and market-based strategies for increasing educational performance: poor-performing teachers are weeded out over time, as better-performing teachers slowly rise to the top of their career fields and receive benefits commensurate with their performance. Consequently, we should expect to see charter schools with high levels of teacher turnover compared to traditional public schools (TPS), but this increased turnover should also be associated with an increase in positive student outcomes – or at a minimum, associated with greatly dampened negative effects.

Theory and hypotheses

As Ronfeldt et al. have noted, there is ‘surprisingly little research’ which attempts to determine the causal relationships between teacher turnover and student achievement (Ronfeldt et al., 2013: 5). Confusingly, turnover among teachers is theorized to have both negative and positive effects upon organizational and student performance, depending upon the assumptions about the teachers in question. The limited scholarship in this area can be divided into two general categories: those who believe that teacher turnover has generally negative effects upon student achievement, owing to the loss of high-quality experienced teachers and the associated organizational dysfunction (Guin, 2004; Ronfeldt et al., 2013), and those who believe that the turnover produced by the quitting or termination of under-performing teachers will be a net positive for organizational performance and, therefore, student achievement (Hanushek, 2012). Both sides can claim a sound logical basis for their reasoning: certainly, if mostly high-performing teachers are being forced out of schools, then student achievement will likely suffer. If mostly low-performing teachers are being forced out – either through terminations or increased organizational expectations – then student achievement will likely improve, assuming that their replacements are more highly-performing, as suggested by some (Torres, 2014).

As stated earlier, charter schools in the United States are mostly free from the traditional labor bureaucracy which surrounds many public educational institutions. Charter school administrators, compared to traditional public school administrators, have a relatively free hand to set organizational goals, required skills and competencies for teachers, faculty compensation and the ability to hire and fire at will.

Consequently, it is possible that charter schools and traditional public schools have different experiences of turnover and teacher attrition. It may be that traditional public schools primarily lose teachers through voluntary turnover (e.g. highly-qualified teachers leaving for better-paying jobs or careers with more developmental opportunity in industry or government), whereas charter schools experience turnover primarily as a loss of poorly-performing teachers through termination or organizational pressure to achieve results.

One of the reasons for the ambiguity in the research of turnover is that organizational data rarely capture voluntary vs forced turnover. Voluntary turnover can have a much more negative impact as it likely means that qualified teachers leave to pursue better options – and it is unlikely that any school or organization would terminate high-quality employees. If this is true, then it follows that greater latitude in hiring and firing decisions will result in greater turnover of employees that the organization considers ‘low-quality’, whereas less latitude will result in a greater admixture of low- and high-quality employees. Accordingly, teacher quality at a school must be controlled for in any analysis of turnover.

We expect teacher turnover to have different effects upon charter schools than on traditional public schools. In particular, the effect may still be negative, but significantly less severe than what would be observed in public schools. For example, turnover in public schools would generally be harmful to student achievement because they tend to lose high-quality teachers to attrition, whereas charter schools may tend to lose low-quality teachers to attrition, owing to the greater labor flexibility inherent in charter school organizations. If true, this theory would be an advancement in educational policy studies, as it suggests a direct causal (and empirical) link between at least one aspect of educational labor policy and student achievement.

Because turnover is expected to have negative overall effects upon organizational effectiveness in general (Hur, 2013; Ng and Feldman, 2010), and turnover of high-quality teachers is expected to have negative effects on public school effectiveness in particular (Guin, 2004; Harris and Sass, 2011; Watlington et al., 2010), which in turn is expected to have negative effects upon student achievement (Aaronson, 2007; Guin, 2004), we make the following hypothesis.

Hypothesis 1: Increased turnover negatively affects student achievement

We theorize charter schools are more likely to be forcing out mostly lower-quality teachers, which should ultimately improve student achievement in charter schools. Tenure and career longevity of high-quality teachers has been shown to have positive effects upon various measures of student achievement, including standardized test scores (Harris and Sass, 2011), whereas turnover of low-quality teachers is expected to have a similar positive effect upon test scores (Hanushek, 2012). The assumption is plausible, as charter school districts are known to have generally higher levels of teacher turnover than traditional public school districts (Stuit and Smith, 2012), but we theorize that this turnover is primarily of lower-performing teachers, whereas turnover in traditional public school districts consists primarily of higher-performing teachers. Consequently, charter school districts will maintain positive student outcomes even with high levels of turnover, whereas increasing levels of turnover will have worse effects upon public school district student performance. Hence, we make the following hypothesis.

Hypothesis 2: In charter schools, the negative effect of turnover on student achievement is reduced in comparison to traditional public schools.

Methodology, data, and variables

The Texas Educational Agency (TEA) produces comprehensive annual data sets describing the Texas school districts, called ‘TEA Snapshot’, providing comprehensive statistical data for any Texas school district organization (public or charter) which receives funds from the state during the school year. The analysis below is based on the 2012–2013 TEA Snapshot. School districts and organizations are required to keep exhaustive statistical data on students, student outcomes, teachers, labor conditions (salary, turnover, teacher experience, teacher education), and so forth (TEA, 2015b).

The study population is comprised of all Texas charter school organizations and traditional public school districts for the 2012–2013 school year. School districts and charter school organizations have been chosen over individual schools, as the district-level analysis best represents the organizational level which is in charge of labor decisions, goals, and standards. Texas was chosen among other states, as Texas is perceived as having an extremely ‘permissive’ environment in terms of charter school formation and labor policies, as well as being an early adopter of charter school policies, and being one of the few states having a mature and well-established charter school industry (Podgursky and Ballou, 2001). Charter schools are organized into ‘charter management organizations’, which are often conglomerations of smaller schools united by a single administrative body, and which operate under a single educational and labor policy – an arrangement substantially similar to traditional public school districts. Despite the slightly different terminologies, both charter management organizations and public school districts are broadly similar organizations, created to accomplish broadly similar goals, and can be compared to each other with some degree of accuracy (Farrell et al., 2012; Lake et al., 2010). There are 1228 public school districts and charter management organizations in the State of Texas for the 2012–2013 school year.

Five dependent variables

The study utilizes five measures of student achievement as listed below: four of them are the passing rates for different mandatory test subjects, and the fifth variable is the 5-year graduation rate.

STAAR test pass rates (math, science, writing, and social studies)

The State of Texas Assessments for Academic Readiness (STAAR) is the mandatory standardized test which must be administered in every local educational organization (school district or charter management organization) that receives funding from the State of Texas. It should be noted that this variable is not an ‘average test score’, but is instead the percentage of students in a particular school district or organization who pass a particular test in a particular year. For example, a ‘math pass rate’ of 76 would mean that 76% of students in a particular school district or charter organization managed to pass their standardized math test in 2012–2013. For this project, we will be examining the ‘core’ tests: STAAR Math, STAAR Writing, STAAR Social Studies, and STAAR Science. Although the use of standardized testing to gauge student achievement is controversial – to say the least – students are generally unable to matriculate without passing their respective STAAR exams, meaning that the exams themselves provide a ‘de-facto’ measurement of achievement; you can’t achieve anything in Texas academically, by definition, if you do not pass the tests.

Graduation rates

For a more robust measurement of student achievement that goes beyond mere standardized test pass rates, we have also chosen to examine 5-year longitudinal graduation rates, which is defined as the percentage of students in a particular cohort who graduate within 5 years of the beginning of their 9th-grade school year. Some collegiate bodies such as the National Collegiate Athletic Association (NCAA) have suggested that, although standardized test scores are not perfectly correlated with graduation rates, they are generally predictive of a student’s overall chances of graduation (Welch, 1999). As graduating from secondary school is virtually the definition of academic success for teenage students, and because graduation rates are at least theoretically distinguishable from standardized test score pass rates (as it is possible to pass a standardized test but later drop out for other reasons), we can conclude that overall graduation rates are an alternative empirical measurement of student achievement, somewhat independent of standardized test scores or pass rates. Consequently, we would expect that the organizational disruption caused by turnover of high-quality teachers leads to lower student performance (Hur, 2013; Ronfeldt et al., 2013; Watlington et al., 2010). The 5-year graduation rate gives a generous window for examining student effectiveness, as it allows slow learners an extra year to graduate, rather than counting them as ‘un-graduated’ at the end of 4 years. Graduation rates for 2013 are calculated by tracking students who started 9th grade in 2007–2008 and who graduated within a year of 2012.

Independent variables

The two variables required to test the hypotheses formulated in the previous section are whether or not the district is charter, and teacher turnover. The variable Charter district is a binary variable which is determined by assigning the value ‘1’ to every school district listed as a ‘charter’ in the TEA Snapshot ‘Community type’ table, and assigning a ‘0’ to each school which is defined as something other than a charter school (TEA, 2015a). Private schools (which generally do not receive funding from the state) are excluded from this analysis.

The variable Teacher turnover is the percentage of full time equivalent (FTE) teachers in a particular school organization (both public and charter) at the beginning of the year, who were not employed there at the end of the school year. Teachers are tracked state-wide by social security number, regardless of the school for which they are employed (TEA, 2015b). For example, a Teacher turnover score of 33 means that 33% of the teachers who began as full-time employees in a particular school year were no longer employed by that school district at the end of the school year.

Testing Hypothesis 2 requires the creation of an interaction term between charter status and teacher turnover. The variable Charter*Turnover or the charter–turnover interaction is an interaction term defined by Charter district (as above), multiplied by the Teacher turnover rate (also as above).

Control variables

To reduce the possibility of alternative explanations for charter–traditional district difference that might result from preexisting differences, the following control variables are included: the variable Poverty is the percentage of students in a given educational organization who are considered economically disadvantaged by the State of Texas. In practical terms, this means students who receive free and reduced-price lunches provided by the school (TEA, 2015b). Eligibility for free and reduced-price lunches is defined as those students whose families fall within 130% and 185% (respectively) of the Federal poverty line (USDA, 2013). These two categories are counted together when final results are tallied, meaning that a ‘poverty rate’ score of (for example) 85 means that 85% of the students in a school organization are children in families which earn 185% or lower of the federal poverty line. This number also includes other students whose families are considered economically disadvantaged in some way, such as those who receive food stamps, Pell grants or other public assistance (TEA, 2014b).

The variable Percent white is the percentage of students in a charter school organization or public school district who are considered to be ‘white’ persons, and is defined as the percentage of students in a particular school district who are reported as ‘white’ to the TEA through the PEIMS data system (TEA, 2015b).

The variable Total students is a surprisingly complex variable, given the huge number of reasons that a student can be counted (or not counted) as a student (TEA, 2014a), but can be summed up as the number of students actively ‘in membership’ at a particular school district by 26 October of the school year in question (TEA, 2015b). ‘Membership’ is defined as students who are present for at least, on average, 2 hours of educational instruction per day (TEA, 2014a: 274).

The variable Percent of teachers with an advanced degree represents the percentage of full-time teachers in a charter organization or public school district who have attained a master’s or doctoral degree (TEA, 2015b).

The variable Attendance rate is ‘the total number of days students were present’ during the measured school year, ‘divided by the total number of days students were in membership’ (TEA, 2015b).

The variable Charter*Size denotes an interaction term between charter school organization status and the number of students in membership at a charter school. The variable Average teacher experience in years represents a weighted average of the years of experience of the teachers in a particular charter organization or school district (TEA, 2015b). The variable Average teacher experience in years squared represents the square root of the percentage of teachers with advanced degrees in each charter organization or school district (to account for the possible curvilinear effect of experience on achievement, as suggested by some authors).

The variable Suburban districts represents a categorical variable to denote the charter organization or school district’s geographic presence in a suburban area (TEA, 2015a). The variable Number of schools represents the number of individual schools in that district or charter school organization (TEA, 2015b). Finally, the variable Students per teacher is the average number of students per full-time teacher in each charter organization or public school district (TEA, 2015b).

All of these variables can conceptually affect student outcomes – along with educational labor practices. Consequently, we have run a regression analysis to determine the actual effects of the starkly different managerial and labor environments of charter and public schools in the state of Texas.

Results

Table 1 presents group comparisons between the school district types. As theorized earlier, charter schools exhibit substantially higher teacher turnover – twice the rate of the traditional school districts. Consequently, we can say with some certainty that there are, indeed, differing labor practices and policies at work in these two different types of organizations, or, at the very least, vastly different labor environments.

Table 1.

Descriptive statistics – charter vs public.

	Public			Charter
	M	SD	n	M	SD	n	T
Math pass rate	78.6	9	998	66.2	17.1	148	9.75***
Science pass rate	80.8	8.79	998	69.3	15.83	148	9.18***
Writing pass rate	61.8	11.78	998	53.4	20.56	148	4.14***
Social studies pass rate	74	10.14	998	69.4	17.15	148	3.94***
Teacher turnover	18.2	8.85	998	36.7	15.15	148	8.81***
% Non-white	48.1	26.59	998	79.2	23.33	148	–14.69***
Poverty rate	58	18.48	998	69.3	24	148	–6.1***
Graduation rate	94.3	8.93	967	68	27.98	89	–17.85***
Total students	4756	13427	1026	885.3	1424	202	–4.77***
% ESL students	8.3	9.51	1026	14.7	18.67	202	8.98***
Average teacher salary	44,531	4105.2	1026	41,207	5293	200	9.972***
Percentage of teachers with an advanced degree	17.2	8.1	1026	18.7	11.3	200	–2.26**
Average teacher experience	12.7	.075	1026	5.1	.23	200	36.5***
Students per teacher	12.8	2.6	1026	15.4	4.2	200	–11.5***
Attendance rate	96.2	0.8	1026	94.7	4.5	193	9.4***

Charter schools are characterized with worse student outcomes across all five dependent variables. They also tend to be poorer, and to have larger non-white and English-as-a-second language student populations. Charter districts are considerably smaller, and their typical teacher salary is almost 10% lower than the average teacher salary in public school districts. The proportion of teachers with advanced degrees is marginally larger in charter school districts (18%, vs 17% in public school districts).

The average experience of charter school teachers is notably shorter than public schools. They also have a significantly higher number of students per teacher, and lower attendance rate.

Overall, there are significant systematic differences between the two types of school. However, as these district characteristics likely confound each other, a multiple regression analysis is required.

It is worth emphasizing that the standard deviations of all measures are drastically higher in the group of charter districts vs traditional public district, by definition suggesting extreme variability in quality and capacity of charter districts; conversely, traditional public districts exhibit much greater homogeneity across all measures. The topic of striking variability of charter schools vs public schools deserves special attention in future research.

Table 2 models the impact of turnover and charter status on all five dependent variables, including control variables. For all models, we performed White’s test to check for heteroskedasticity, which indicated that the variances are not homogenous. Although heteroskedasticity does not bias the coefficient estimates, but makes their standard errors incorrect, thus interfering with hypothesis testing, regressions with robust standard errors are generally an adequate remedy. Extreme outliers (districts with fewer than 250 students and districts with more than 40 schools are excluded from the analysis).

Table 2.

Regression results (robust standard errors in parentheses).

Variables	Graduation rate	Math pass rate	Science pass rate	Writing pass rate	Social studies pass rate
Teacher turnover	–0.0393	–0.289***	–0.286***	–0.313***	–0.221***
	(0.0282)	(0.0384)	(0.0359)	(0.0465)	(0.0419)
Charter school	–5.458	–8.806***	–7.751***	6.112	–0.0908
	(4.355)	(3.365)	(2.875)	(3.829)	(4.019)
Charter*Turnover (interaction term)	–0.126	0.146*	0.195**	0.0780	0.106
	(0.110)	(0.0809)	(0.0808)	(0.0875)	(0.106)
Percentage of teachers with an advanced degree	–0.0189	0.0357	–0.0493	0.0406	0.0571
	(0.0309)	(0.0441)	(0.0384)	(0.0497)	(0.0457)
Average teacher salary	3.78e–05	0.000148	0.000198*	–8.88e–05	0.000113
	(9.08e–05)	(0.000111)	(0.000111)	(0.000117)	(0.000144)
Poverty	–0.0295*	–0.190***	–0.219***	–0.318***	–0.238***
	(0.0159)	(0.0223)	(0.0205)	(0.0254)	(0.0293)
Percent white	0.0327***	0.0406***	0.0333**	0.00350	0.0395*
	(0.0112)	(0.0152)	(0.0143)	(0.0175)	(0.0208)
Total students	0.000401*	8.41e–05	0.000276	0.000249	0.000185
	(0.000207)	(0.000314)	(0.000246)	(0.000308)	(0.000290)
Attendance rate	2.358***	2.082***	1.162***	3.045***	1.647***
	(0.220)	(0.360)	(0.316)	(0.532)	(0.429)
Charter*Size (interaction term)	0.00108**	0.00144*	0.00249***	0.000605	0.00125*
	(0.000513)	(0.000749)	(0.000794)	(0.000609)	(0.000713)
Average teacher experience in years	–0.933	–0.505	1.215	0.160	–0.408
	(0.672)	(0.648)	(0.740)	(0.712)	(0.870)
Average teacher experience in years squared	0.0347	0.0228	–0.0553*	0.00376	0.00614
	(0.0251)	(0.0251)	(0.0288)	(0.0280)	(0.0340)
Suburban districts	0.169	–0.221	0.341	–0.983	0.921
	(0.417)	(0.564)	(0.513)	(0.610)	(0.621)
Students per teacher	–0.209	0.0550	–0.455**	0.125	–0.347
	(0.162)	(0.242)	(0.215)	(0.254)	(0.242)
Number of schools	–0.399**	–0.0867	–0.181	–0.223	–0.0796
	(0.162)	(0.248)	(0.194)	(0.245)	(0.228)
Constant	–122.7***	–112.3***	–22.98	–207.0***	–66.46
	(22.13)	(37.20)	(32.42)	(54.63)	(43.95)
Observations	855	954	948	952	925
R-squared	0.600	0.598	0.572	0.608	0.470

***p < 0.01, **p < 0.05, *p < 0.1.

As expected in Hypothesis 1, the direct effect of teacher turnover is negative, across all outcomes, with the exception of the 5-year graduation rate. These results are not surprising, and are in line with most organizational theories of turnover, as teachers leaving are expected to have both negative disruptive and compositional effects upon the educational workforce.

The findings also support Hypothesis 2, albeit very tenuously: the interaction term between charter status and turnover has a small, and borderline statistically significant, positive effect on math and science pass rates, but no statistically discernible effect on the other three dependent variables. This suggests that on the latter outcomes, turnover’s impact in charter schools is no worse than that of public schools, whereas the negative impact of turnover on math and science pass rates is reduced in charter schools, as hypothesized.

The direct effect of being a charter district is not of focal interest for the proposed hypotheses, but it is nevertheless notable. Per the above models, there are no discernible differences between charter and public districts on most variables, but again with two notable exceptions: charter districts are characterized with significantly lower pass rates in math and science. It is also worth noting that the performance of bigger charter schools is likely to be better, as the interaction term between charter status and size shows.

Other control variable effects worth noting concern teacher quality: the share of teachers with advanced degrees has no discernible effect, nor does average teacher experience and salary. The one notable exception are science pass rates, where both salary and experience show a positive effect, as is logical to expect, albeit the effects are small and borderline significant. Further, the model for science pass rates shows diminishing return on experience as approximated by the squared years of experience variable. Science pass rates are also the exception with another variable: students per teacher. Although in most models it had no statistically significant effects, an increase in the number of students per teacher reduces science pass rates, as is logical to expect.

Student attendance rates have an unsurprisingly positive effect upon student success, and poverty and ethnicity have an unsurprisingly negative effect across all models. Even when all of these variables are accounted for, the impact of charter status (whether or not a school belongs to a charter organization) is still strongly deleterious to student success in math and science exam pass rates.

Conclusion

The State of Texas has a mature and extensive charter school system that has been in existence since the late 1990s. These charter organizations operate alongside traditional public school districts, often within the same cities or counties. This allows us to cautiously compare the student outcomes between charter organizations and traditional public school districts, controlling for systematic differences in student and teacher populations, socioeconomics, etc. Once these variables (generally ‘student-side’ variables, such as socio-economic status, and ethnic background – the variables that schools cannot control) are accounted for, we are left with a number of ‘school-side’ variables, such as labor policy, teacher quality, and even the shape of the organization itself – variables over which schools can exercise considerable control.

We expected to find that charter school organizations, given their greater flexibility in employment and the firing of underperforming teachers, would have a fundamentally different experience of turnover than traditional public school districts. The more flexible labor environment found in charter organizations could allow them to more quickly terminate underperforming teachers, promote superior-performing teachers and, consequently, raise student achievement despite the expected overall increase in turnover.

This appears to be largely the case: charter school organizations do, indeed, have much higher levels of turnover, also as noticed in the literature by Torres (2014) and Lubienski (2003), and our estimates are consistent with theirs. Although the turnover itself cannot and does not have a positive effect on student achievement, per our results turnover is somewhat less harmful to student achievement in charter school organizations, albeit not universally, but only with respect to two specific outcomes, namely science and math pass rates, and not with regard to graduation rates, or writing and social studies pass rates. However, it may be worthwhile to single out these two outcomes: of all school subjects, math and science have the clearest standards of accomplishment – in other words, it is much easier to measure success in math and science, whereas standards for writing and social studies are inherently less formal. Accordingly, whereas some degree of performance variance – especially if it is related to the quality of the teachers – may be masked in assessing writing and social studies achievement, no such gray areas are possible in evaluating math and science accomplishment. This indirectly supports the notion that the effect is at least partially indeed owing to the likelihood that a significant amount of the turnover is accounted for by underperforming teachers.

These results broadly suggest that arguments advanced in the literature about the improvements in school outcomes that can be achieved given more flexibility in labor practices have some merit. However, the findings should not be overstated, and, indeed, must be approached cautiously. Although the negative impact of turnover on student achievement appears to be indeed softened – presumably because much of this turnover is accounted for by removing poorly performing teachers – the turnover rate itself is so high that it poses a serious question: is turnover a viable strategy to improve student achievement? The high rate of turnover may cancel out the benefits of less damaging impacts compared to public schools. Further, teacher pay in charter schools appears significantly lower, and if the dominant strategy is merely removing poor teachers rather than hiring higher-performing teachers (which would likely entail increases in pay), this may potentially reduce the benefits of flexibility and the resulting turnover further still.

That charter schools need to continually improve their organizational and labor practices is ultimately manifested by their worse performance than traditional public schools on some outcomes in this data set, after multiple controls. As this proverbial drag upon achievement does not appear to be coming from students or their major demographic variables, it seems likely that, organizationally speaking, charter schools in the state of Texas are currently in some way deficient compared to traditional public school districts – unless there are major unseen demographic variables affecting student achievement of which the wider educational literature is unaware. More research will be needed in this area to determine what, if any, organizational effects charter schools or their student’s circumstances are bringing to bear so negatively upon student achievement.

Lastly, the ‘at-will’ and anti-union labor environment perceived to exist in charter schools also coexists in a larger national context of de-unionization, even in public school districts. It would be a mistake to suggest that charter school organizations and public school districts have truly diametrically opposed labor policies, especially recently – in reality, the political movement which inspired the creation of charter schools has, in many ways, influenced the ongoing evolution of public school labor policy as well. Nevertheless, the distinctions are sufficiently clear to justify treating the two types of schools as institutionally significantly different, especially in labor policy.

In conclusion, the significant managerial and organizational differences between Texas charter school organizations and Texas public school districts suggest that greater flexibility in hiring and firing of teachers among charter schools may slightly improve student outcomes. The reader is cautioned that these results are relatively small compared to the profoundly negative effects upon student achievement caused by poverty, and, more worryingly, may be overshadowed by the existing gap in student success between charter organizations and public school districts in math and science pass rates.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Aaronson

Barrow

Sander

(2007) Teachers and student achievement in the Chicago public high schools. Journal of Labor Economics 25(1):95–135, https://doi.org/10.1086/508733.

Abelson

Baysinger

(1984) Optimal and dysfunctional turnover: Toward an organizational level model. The Academy of Management Review 9(2): 331–341, https://doi.org/10.2307/258446.

Ball

(2012) Global Education Inc.: New Policy Networks and the Neo-liberal Imaginary. Abingdon, UK; New York: Routledge.

Bankston

Ravitch

Petrilli

Renzulli

Paino

Bonastia

(2013) Charter schools. Contexts 12:16–25.

Carruthers

(2012) New schools, new students, new teachers: Evaluating the effectiveness of charter schools. Economics of Education Review 31(2):280–292, https://dx-doi-org.web.bisu.edu.cn/10.1016/j.econedurev.2011.06.001.

Clotfelter

Ladd

Vigdor

(2007) Teacher credentials and student achievement: Longitudinal analysis with student fixed effects. Economics of Education Review 26(6):673–682, https://doi.org/10.1016/j.econedurev.2007.10.002.

Collier

Millimet

(2009) Institutional arrangements in educational systems and student achievement: A cross-national analysis. Empirical Economics 37(2):329–381, https://dx-doi-org.web.bisu.edu.cn/10.1007/s00181-008-0235-9.

Dalton

Todor

(1979) Turnover turned over: An expanded and positive perspective. The Academy of Management Review 4(2):225–235, https://doi.org/10.2307/257776.

Dalton

Krackhardt

Porter

(1981) Functional turnover: An empirical assessment. Journal of Applied Psychology 66(6):716–721, doi: https://dx-doi-org.web.bisu.edu.cn/10.1037/0021-9010.66.6.716.

10.

Darling-Hammond

(2000) Teacher quality and student achievement. Education Policy Analysis Archives 8(1), https://dx-doi-org.web.bisu.edu.cn/10.14507/epaa.v8n1.2000.

11.

Day

(2007) Variations in the conditions for teachers’ professional learning and development: Sustaining commitment and effectiveness over a career. Oxford Review of Education 33(4):423–443, https://doi.org/10.1080/03054980701450746.

12.

Farrell

Wohlstetter

Smith

(2012) Charter management organizations: An emerging approach to scaling up what works. Educational Policy 26(4):499–532, https://doi.org/10.1177/0895904811417587.

13.

Figlio

(1997) Teacher salaries and teacher quality. Economics Letters 55(2):267–271, https://doi.org/10.1016/S0165-1765(97)00070-0.

14.

Guin

(2004) Chronic teacher turnover in urban elementary schools. Education Policy Analysis Archives 12(42), https://doi.org/10.14507/epaa.v12n42.2004.

15.

Hanushek

(2012) Low-performing teachers have high costs. Education Next 12(3):68.

16.

Hanushek

Rivkin

(2006) Teacher quality. In: Hanushek

Welch

(eds) Handbook of the Economics of Education, Vol. 2, Chap. 18. Amsterdam and Oxford: Elsevier, 1051–1078.

17.

Hanushek

Rivkin

(2010) Constrained job matching: Does teacher job search harm disadvantaged urban schools? NBER Working Paper No. 15816. Cambridge, MA: National Bureau of Economic Research.

18.

Hanushek

Kain

O’Brien

Rivkin

(2005) The market for teacher quality. NBER Working Paper No. 11154. National Bureau of Economic Research, Cambridge, MA.

19.

Harris

Sass

(2011) Teacher training, teacher quality and student achievement. Journal of Public Economics 95(7–8):798–812, https://dx-doi-org.web.bisu.edu.cn/10.1016/j.jpubeco.2010.11.009.

20.

Hur

(2013) Turnover, voluntary turnover, and organizational performance: Evidence from municipal police departments. Public Administration Quarterly 37(1):3–35.

21.

Krieg

(2006) Teacher quality and attrition. Economics of Education Review 25(1):13–27, https://doi.org/10.1016/j.econedurev.2004.09.004.

22.

Lake

Dusseault

Bowen

Demeritt

Hill

(2010) The national study of charter management organization (CMO) effectiveness: Report on interim findings. Center on Reinventing Public Education, Seattle, WA.

23.

Lubienski

(2003) Innovation in education markets: Theory and evidence on the impact of competition and choice in charter schools. American Educational Research Journal 40(2):395–443, https://doi.org/10.2307/3699394.

24.

Malloy

Wohlstetter

(2003) Working conditions in charter schools: What’s the appeal for teachers? Education and Urban Society 35(2):219–241.

25.

TWH

Feldman

(2010) Organizational tenure and job performance. Journal of Management 36(5):1220–1250.

26.

Park

Ofori-Dankwa

Bishop

(1994) Organizational and environmental determinants of functional and dysfunctional turnover: Practical and research implications. Human Relations 47(3):353–366.

27.

Podgursky

Ballou

(2001) Personnel policy in charter schools. Thomas B. Fordham Foundation, Washington, DC.

28.

Rivkin

Hanushek

Kain

(2005) Teachers, schools, and academic achievement. Econometrica 73(2):417–458, https://doi.org/10.2307/3598793.

29.

Ronfeldt

Loeb

Wyckoff

(2013) How teacher turnover harms student achievement. American Educational Research Journal 50(1):4–36.

30.

Schneider

(2016) School Choice: The End of Public Education? New York: Teachers College Press.

31.

Stuit

Smith

(2012) Explaining the gap in charter and traditional public school teacher turnover rates. Economics of Education Review 31(2):268–279, https://dx-doi-org.web.bisu.edu.cn/10.1016/j.econedurev.2011.09.007.

32.

TEA (2014a) 2014–2015 Student Attendance Accounting Handbook. Austin, TX: Texas Education Agency. Available at: http://www.galenaparkisd.com/handbooks/Student%20Attendance%20Handbook.pdf (accessed 15 February 2015).

33.

TEA (2014b) PEIMS: Data Standards 2013–2014, C054 ECONOMIC-DISADVANTAGE-CODE. Available at: http://ritter.tea.state.tx.us/peims/standards/weds/c054.html (accessed 15 February 2015).

34.

TEA (2015a) Snapshot 2013 Summary Tables Community Type. Available at: http://ritter.tea.state.tx.us/perfreport/snapshot/2013/commtype.html (accessed 15 February 2015).

35.

TEA (2015b) Snapshot 2013: Item Definitions. Available at: http://ritter.tea.state.tx.us/perfreport/snapshot/2013/itemdef.html (accessed 15 February 2015).

36.

Torres

(2014) Is this work sustainable? Teacher turnover and perceptions of workload in charter management organizations. Urban Education 51(8):891–914.

37.

USDA (2013) Child Nutrition Programs: Income Eligibility Guidelines. Department of Agriculture, Food and Nutrition Service, Washington, DC. Available at: http://www.fns.usda.gov/sites/default/files/IEG_Table-032913.pdf (accessed 14 November 2015).

38.

Vergari

(2007) The politics of charter schools. Educational Policy 21(1):15–39.

39.

Verger

Fontdevila

Zancajo

(2016) The Privatization of Education: A Political Economy of Global Education Reform. New York: Teachers College Press.

40.

Watlington

Shockley

Guglielmino

Felsher

(2010) The high cost of leaving: An analysis of the cost of teacher turnover. Journal of Education Finance 36:22–37.

41.

Welch

(1999) NCAA says it can show ‘direct link’ in athletes’ test scores and graduation rates. The Chronicle of Higher Education 45(46):1.