What is the average length of a superintendent鈥檚 tenure in the job?
You probably have a vague sense that it鈥檚 perhaps somewhere between three and six years. But it turns out that, on a national level, we simply cannot say for sure.
Now, the follow-up question I put to you is this: This seems like a critical piece of data for our field to have. Why is it so hard to get?
This reported essay is partly a plaint, and partly a plea that has developed out of my coverage the last few weeks about whether the stresses of the pandemic might cause more superintendents to leave, retire, or even get pushed out of the job.
Ultimately, I could only get anecdotal reports from hiring experts, researchers, former supes, and district administrators. (This tends to make we reporters who like certainty very uncomfortable.) All of them were generous in sharing what they鈥檝e heard. But all of them also said that it鈥檚 just very hard, based on what we know, to determine conclusively that there鈥檚 a spike in turnover.
It was probably a little naive to expect that such data would be easy to find. But it鈥檚 nevertheless discomforting not to know how to answer important questions at a time when many aspects of K-12 leadership are being tested.
For instance, could the pandemic affect female superintendents more, given nationwide child- and elder-care demands that have fallen more heavily on women? Might it affect the (slow) progress districts have made in hiring more diverse leaders?
I recently profiled a retiring California superintendent, Diann Kitamura, a Japanese-American leader who spoke about the frustrations of dealing with overlapping (and conflicting) guidance, the labor issues that have shaped in-person schooling discussions, and her own attempts to balance work and home commitments. But I couldn鈥檛 put Kitamura鈥檚 story in context beyond her own county.
The bottom line is this: As a field, we agree that district leaders do matter. Otherwise we wouldn鈥檛 bother to debate whether superintendent salaries are too high or too low (a perennial topic in hiring). And we wouldn鈥檛 obsess about the unusual sight of the leaders of the three largest school districts all leaving within just two months about each other, as we are currently.
So, what do we do?
The benefits and limitations of our current information
None of this is to say that we don鈥檛 know anything about superintendents. There are excellent research centers in the universities that study school leadership, and some high-quality professional journals that study the characteristics of effective district leaders, and how their work relates to the larger disciplines of organizational and management theory.
Top marks go to AASA, the School Superintendents Association, for doing yeoman鈥檚 work supplying Its collections are unquestionably the best we have at the moment.
The organization does this through two main sources: an annual salary survey that includes some demographic questions, and a decennial survey that picks up some longer-term trends and situates them in the historical development of the field of school leadership.
The AASA surveys get information from around 1,200 to 1,600 district leaders鈥攁round a 10 percent sample of the national force. That鈥檚 pretty darn good.
From these collections, we know some important trends. For example, over time, women are making up a greater proportion of superintendents: about 27 percent, according to the 2020 decennial survey, up from around 24 percent in 2010. More leaders of color are rising to the top job, too (about 9 percent in the 2020 survey compared to 6 in 2010).
But surveys like this have some built-in disadvantages. For one thing, one must be careful in interpreting results because each sample is different, and there are slight differences year over year in who鈥檚 answering the questions. It quickly becomes iffy to draw conclusions about some important sub-populations, because there are simply too few of them in the sample.
(In the 2019-20 annual collection, the modal superintendent鈥攖he one that occurs more than any other in the sample鈥攚as married, white, male, with some experience as a principal, and between two to three years into his current job.)
Finally, because the samples differ, we also can鈥檛 ask the same set of leaders, year over year, if they are still in that job a year later, two years later, and so on. Instead, we have to carefully put together some other data points to try to figure out turnover rates.
The annual collections ask how long respondents have been in their current job. (In 2020-21, about 47 percent were in years 1-5 and 27 percent in 6-10). And it asks how long they have been a superintendent overall (this is spread out a bit more across age ranges). But these don鈥檛 quite tell us how long a superintendent stays in any one posting.
There are other clues that turnover is increasing: A higher percentage of superintendents are now taking on their first gig in their early, rather than their late 40s鈥攂uoying the idea that superintendent tenures are getting shorter, said Christopher Tienken, an associate professor of education leadership, management, and policy at Seton Hall University, who works on the AASA collections as a researcher-in-residence. (On the other hand, a smaller proportion of superintendents in the 2020 decennial said they planned to leave in the next five years than in the 2010 decennial.)
What might we be able to answer with a longitudinal sample?
What could we learn if we did have year-to-year鈥攕ometimes called longitudinal鈥攄ata on the same sample of superintendents? We have some clues on this, thanks to work by Jason Grissom, a professor of public policy and education at Vanderbilt University, and his research colleagues who have looked at state-level data in Missouri and California.
They鈥檝e yielded some striking findings. In California, superintendent turnover was . In Missouri,
And the Broad Center, in 2018, looked at the tenure of big-city superintendents over a nearly 15-year period. It found that, at five and a half years, those leaders stayed almost twice as long as other estimates predicted.
The problem is that we don鈥檛 know if these findings can be extrapolated nationally, given how geographically and demographically different school districts are.
I asked Grissom what other questions we might be able to ask if we had a national, longitudinal set of data on superintendents. Among them, he said, we could try to answer:
- What are the predictors of superintendent turnover?
- What鈥檚 the variation in turnover based on superintendent characteristics, like race, gender, and experience?
- Do internal hires versus those coming from outside of a district stay longer?
- Do certain types of credentials or training experiences correlate with superintendents staying longer?
- What is the relationship between these factors and outcomes, like improved student achievement?
鈥淚f you鈥檙e a school district board trying to make a hiring decision, you need some direction,鈥� he noted. 鈥淩ight now we operate on craft knowledge and experience. Maybe it鈥檚 better to hire someone internally and hold onto them for the long term, because they鈥檝e built other relationships with people in the district.鈥�
It鈥檚 certainly worth asking why, exactly, we don鈥檛 have this data鈥攅specially when you consider the growth in this kind of analysis in other K-12 sectors. Beginning perhaps three decades ago, states began to set up longitudinal data systems to collect year-over-year information on students, teachers, and sometimes principals.
This has been a federal priority, too. The National Center for Education Statistics, a wing of the U.S. Department of Education, collects nationally representative data on teachers and principals every few years. And the U.S. Census Bureau administers the to find out why they stayed in their jobs, left the profession, changed positions, or remained in the same school.
But there is no comparable national collection for superintendents.
A smaller sample means data could be harder to collect
In part, that鈥檚 probably because putting one together would be, in one sense, more difficult.
The teacher and principal workforces are so large (over 3 million and 100,000, respectively), that researchers can develop a national, stratified sample of schools, and then survey teachers and principals within them, Grissom said.
For superintendents, though, one would ideally collect the information from all 13,000 districts; otherwise, you might not have the statistical power to analyze important data for subpopulations, like superintendents of color, and so on.
It鈥檚 not an impossible task to try to collect this data鈥� at least initially鈥攆rom every district, given public records and data-scraping tools, but it also isn鈥檛 simple. Basically, this comes down to a complicated question of getting the sampling frame correct.
And, Grissom said, superintendents are busy people. Even after dealing with the methodological challenges, getting them to answer a survey in the middle of uncertain times like the pandemic could be difficult. (Response rates on the federal teacher and principal surveys also have fallen in recent years.)
School leadership experts say the lack of federal investment in such a collection is probably also a political consequence of the fact that there are simply fewer superintendents compared to other actors in K-12 education.
鈥淭here are two very large national teachers鈥� associations; the principals鈥� associations are also large, and they tend to suck the air out of the room when you鈥檙e talking about lobbying or even grant funding to conduct large-scale research,鈥� said Tienken. 鈥淚t comes down in some ways to the size of the megaphone.鈥�
This is where I turn the question over to you for your insights. How do you think we might try to get at this data? Who might be interested in funding such a collection?
I can鈥檛 speak for AASA, but I suspect they鈥檇 welcome the opportunity to collaborate with researchers or funders on such a collection.
Any takers?
