Should I Stay or Should I Go? Comparing Teacher Mobility in Florida's Charter and Traditional Public Schools

Abstract

Currently, we know very little about the mobility decisions of charter public school teachers and how these compare to the decisions made by traditional public school teachers. In addition, it is unclear whether the teachers who leave charter schools tend to be weaker or stronger than their peers. Using statewide administrative data, I begin to answer these questions by studying the magnitude and nature of teacher turnover in Florida's charter public schools compared with turnover in the state's traditional public schools. It appears that Florida's charter school administrators may be better able to recruit and retain teachers with high academic skills than their traditional counterparts. In addition, the mobility patterns exhibited by Florida's charter school teachers differ from those exhibited by traditional school teachers in important ways, including greater sensitivity to accountability measures and less sensitivity to salary considerations.

I thank Steve Mangum, Kathy Peck, and Jeff Sellers of the Florida Department of Education for their assistance in obtaining and interpreting the data. I also thank Jeffrey Brooks, Douglas Harris, Stacey Rutledge, JimWyckoff, participants at the 2006 Annual Meeting of the American Education Finance Association, and an anonymous reviewer for their valuable comments. Any remaining errors are solely my responsibility.

Notes

I thank Steve Mangum, Kathy Peck, and Jeff Sellers of the Florida Department of Education for their assistance in obtaining and interpreting the data. I also thank Jeffrey Brooks, Douglas Harris, Stacey Rutledge, JimWyckoff, participants at the 2006 Annual Meeting of the American Education Finance Association, and an anonymous reviewer for their valuable comments. Any remaining errors are solely my responsibility.

¹Administrators in districts that are particularly reliant on state funding may also find their ability to offer positions is constrained by delays in the passage of the state budget.

²Student performance on standardized tests is sometimes used as a proxy for student achievement, despite its limitations, but that approach is not taken in this study due to data limitations. The Florida Comprehensive Assessment Test is given only in certain subjects and at certain grade levels. In addition, a student's scores must be available for several years in order to use scores to calculate the value added by a particular teacher in a particular year. Given the relatively small number of charter schools in the state and the fact that my focus is on individual teachers, rather than schools, reliance on Florida Comprehensive Assessment Test scores would seriously constrain the sample used for analysis.

³The issue of whether the state should also become a chartering entity is currently under consideration. During the study period, only local districts and state universities could sponsor a charter school.

⁴There were no charter school leavers during this period who were coded as retirees by their administrators, so this exclusion only affects the analysis of mobility among traditional teachers.

^aIndividual teacher's average years of experience and salary during the years 1998 to 2003 are used because these are time-varying variables. Data are weighted to reflect the number of years that each teacher is in the database. For example, a teacher who was in the database for 3 years would carry a weight of three—one for each year teaching in Florida—whereas a teacher who was in the database for 2 years would carry a weight of two. Each teacher counts as one observation when calculating sample size for teacher attributes.

^bThe first number for charter school experience does not include the imputed years of experience, and the second number includes imputed experience. ^cThe average values for each of these variables are calculated for each school. Data are then weighted to reflect the number of years that each school was in the database. Each school counts as one observation when calculating sample size for school attributes.

*p < .10.

**p < .05. Statistical significance refers to differences between charter public school teachers and traditional public school teachers and their associated schools.

⁵Detailed breakouts are available from the author upon request.

⁶I thank Jeffrey Brooks for raising this point.

⁷In a study of Michigan's charter school teachers, D. C. Harris (2006) found that, on average, charter school administrators did pay a premium to graduates of selective universities.

⁸In 2004, Florida's teacher salaries ranked 29th in the nation (Muir, Nelson, & Baldaro, 2005). When teaching salaries are compared to average annual earnings in each state's private sector, the ranking drops to 33.

*p < .10.

**p < .05. Statistical significance refers to differences between other employment statuses and those remaining in the same school.

^aIndividual teacher's salary.

*p < .10.

**p < .05. Statistical significance refers to differences between other decisions and staying in the same school.

⁹The magnitude of the reported coefficients are difficult to interpret. Instead, I compute the partial effects for each variable at the mean values for charter school teachers. Following that, I calculate the change in the probability of moving or leaving using the equation

, where the change in x is defined in the text. In this example, thechangein x is a one-letter-grade increase. This result is then dividedby P(M _it | x = μ_charter) toestimateaneasily interpretable percentage change in the probability of the mobility decision of interest. I use the charter school teachermeanvalues for both the charter and traditional teacher when computing partial effects for each so that I can determine how the changes in each variable affect mobility rather than capturing differences in initial means.

*p < .10.

**p < .05. Statistical significance refers to differences between other decisions and staying in the same school.

¹⁰This effect size decreases as a teacher gains experience because the regression includes the quadratic term.

¹¹Men are more likely to switch to different professions, but women are more likely to leave the labor force altogether. Overall, men are slightly more likely to leave teaching. Traditional teachers of Color are more likely to move than White teachers. This is the opposite of the likely sign of the charter school regression coefficient but, as noted earlier, is consistent with previous studies of traditional public school teacher mobility.

*p < .10.

**p < .05. Statistical significance refers to differences between sending and receiving schools.

¹²D.N. Harris et al., (2006) found that Florida's traditional school principals tend to weight personal factors, such as caring, more heavily in hiring decisions than professional factors, such as undergraduate GPA.