How We Calculated State Testing Costs

My Education Week “Truth About Testing Costs” commentary outlines why the rhetoric about the high budgetary costs of testing is not only overblown, but in many ways counter-productive, because it stifles investment in the very sorts of high-quality assessments that most educators deeply desire.

My inquiry found that states actually spend very little of their per-pupil spending on standardized testing costs:

Credit: Education Week

I’ve received a number of good comments and questions about how I calculated these figures and made decisions about which costs to include.

I was primarily interested in exploring the validity of claims that funds spent on testing were a root cause of many district and state financial woes — i.e., whether if we eliminated testing costs we’d save teachers’ jobs, art classes, sports, school nurses, librarians, small classes, and more. So, I stuck to the budgetary costs because I wanted to focus on costs that were solely related to statewide standardized testing and would be eliminated if testing was removed. I considered not just costs from federal NCLB mandated testing, but also from state testing programs such as high school exit or end of course exams.

There is some background research on this topic, including the 2010 Stanford Center for Opportunity Policy in Education study that I cited in my commentary, a 2002 paper by Caroline Hoxby, a 2001 Education Commission of the States report, and our own 2006 Margins of Error report. All estimate testing costs as a very small component, less than one percent, of state education spending.

I also consulted with the two assessment consortia, PARCC and Smarter/Balanced, which have done their own studies to try to determine state assessment spending. For example, if you read all the way to page 189 of the Smarter/Balanced Race to the Top funding application, you’ll see estimates for their member states. While these figures differ from mine, they point to the same conclusion: that expenditures are generally much less than one percent. (My figures differ since I looked at all state tests across all enrolled students, not just the expenditures for federally-mandated tests and testing grades.)

To calculate our data, my colleague Mary Nguyen and I consulted state budgets, data sources, and also contacted state departments of education. For example, in California, we looked at the 2009-10 state budget and then also added in additional federal funds that went towards testing purposes. For California, these expenditures included not just testing contracts, administration and reporting, but also $17 million that went towards “funds to reimburse districts and charter schools for costs associated with administering the STAR, CELDT, and CAHSEE that are above and beyond the CDE’s statewide contracts.” In a normal year the state’s costs would have been several dollars per student higher, but California’s budget problems that year led to cuts of $17 million from the assessment budget (see memo detailing budget cuts to the assessment program).

As I noted in the commentary, state budget figures could underestimate various expenditures related to testing, such as preparation materials or personnel. And, some may argue that other expenditures, such as the additional funds that California budgeted for its California High School Exit Exam Intensive Instruction and Services program, are really testing-related costs. For the sake of argument, I noted in the op-ed that you could double the amount spent on testing in California to $28 per student and you’d still be at only .03 percent of state per student expenditures.

A few persons have asked about additional district-level expenditures and other assessment-related programs. We found one study from the NCLB-era, Testing in Colorado: Time, Cost, and Purpose (2007), that looked at not just statewide testing costs, but also the costs of district-level interim and diagnostic assessments, college entry exams like the ACT, and the NAEP. This report found that “at the state and local levels, Colorado spends an estimated $54.59 per year per pupil,” (out of total spending of over $9,000 per pupil that year.) While it covers a wide range of different types of assessments, this study says it did not include test preparation time. (This would be extraordinarily difficult to quantify because what counts as “test prep” is highly subjective and extremely variable across schools.)

Finally, there are definitely additional opportunity costs in terms of teacher/student time. But, per the point of this inquiry, it’s important to remember that if we eliminated testing, this wouldn’t be money we could spend on other things because we would still pay teachers for that time. We reviewed a 2008 paper for the National Research Council on the costs of implementing not only testing, but also standards and accountability systems. As part of this study, the authors estimated teacher time spent proctoring and administering tests in three states. The paper found costs ranging from an additional $24 to $35 per student. As many commenters noted, this instructional time is important. While the costs of time don’t show up in budgets (and can’t be recovered to stave off cuts), we know that each instructional minute matters. And, thus, the quality of the assessments and what we gain from the time spent on assessment are paramount.

While each of these sources provides a different slice on the issue, the overall conclusion is still the same: testing is a relatively small part of state educational spending. There are though, significant costs to misguided rhetoric that pushes states to cut corners and develop low-quality assessment programs.

Tagged as: Assessment, Assessment Consortia, Open Platforms, Policy Dilemmas, Race to the Top, State Education Budgets, Stimulus Funding, Technology, Testing