Monday, December 29, 2008

Bad Science: It's Not the Study. It's the Spin.

Dr Bill Hoyt (William H.) is an economics professor at UK and a former "Cassidy parent" when I was still the principal there. We consulted once on an earlier KERA report of his and I've spoken to his economics students, as well as his wife Dr Gayle Hoyt's class. Good folks. Solid researchers. Great parents.

Bill is the principal author of a new study on education spending in Kentucky along with Christopher Jepsen and Kenneth R. Troske. The study, Educational Spending: Kentucky v Other States, was sponsored by the Bluegrass Institute, the Friedman Foundation for Educational Choice (which promotes vouchers) along with the Center for Business and Economic Research at UK's Gatton School of Business, where the researchers work, and two of my daughters have attended school.

And as studies go, this one's a whole lot better than the last school choice study BGI sponsored.

But Bill, like all (?) economists, is a quantitative guy. So I was shocked to find that the front page of his new report comes with a commercial for school choice!

School Choice for Kentucky: Many agree with the concept. Some disagree. And some simply want more information. As the public debate continues to grow about how best to provide a quality education to all Kentucky children, it is important to know the facts about parent choice, and how parent choice programs have had an impact on communities, parents and students around the country. All of this analysis is done with one goal in mind: The best possible education for all of Kentucky’s children.

Short on facts. Long on point of view. I'm guessing Bill didn't write that.

Such a strong suggestion of bias is not generally found in quantitative studies. Typically logical positivists assert: “Whereof one cannot speak, thereof one must remain silent.”

Special interests have no such restrictions and may be expected to use good scholarship in support of their predetermined biases. Perhaps this is what happens when special interest groups sponsor a study.

As I read the second page of the report, I felt better about it. At least the biases are up front where the reader can grapple with them. And the Friedman Foundation even offers a challenge to those, like me, who would raise an eyebrow at the approach.

So if you’re skeptical about our research on school choice, this is our challenge to you: prove us wrong... We welcome any and all scientific critique of our work. But if you can’t find anything scientifically wrong with it, don’t complain that our findings can’t be true just because we’re not neutral.

Fair enough - but only if the Friedman Foundation restricts itself to the science and leaves unsupported commentary out. "Whereof one cannot speak, thereof one must remain silent.”

Being familiar with Bill's 1999 study, I'm not likely to complain about his science. The conclusions, however, are fair game for any thinking person. Numbers alone are not truth and special interests groups can twist numbers into a tale beyond recognition.

Take this statement from the report's executive summary:

In Kentucky state dollars make up a much larger share of a district’s educational budget than in other states, and this lack of control over funding could lead to less efficient uses of resources.
Is this an example of solid science being overtaken by biased conclusions? Lets break it down:

In Kentucky state dollars make up a much larger share of a district’s educational budget than in other states...
No argument. The data seems to support the conclusion. But then, there's this thin speculation:

and this lack of control over funding could lead to less efficient uses of resources.

What? Control over the sources of the funding?! What control over funding? School districts eat what the legislature feeds them whether by fork or by spoon. Is there some suggestion that school folks waste state money, but spend local dollars more wisely?

Stay tuned. There is a second report promised.

I'm dubious because I'm not sure the source of funding (state v local) matters one iota to school administrators when they plan their budgets. The only thing that matters is how much money - and how much improved educational service it will buy.

The Friedman Foundation's positivist bravado aside, this is a suspect conclusion which should have waited until they presented some basis for it. They should have remained silent.

But when the authors are free of the foundations, and stick to their science, I see nothing wrong with their findings - and none of it is surprising:

  • Despite the increase in educational spending that occurred with KERA, Kentucky still lags behind the average U.S. state in current expenditures per student. However, the gap between Kentucky and other states in per-student current expenditures has narrowed from $2,199 in 1987 to $1,092 in 2006.
  • Since KERA, Kentucky has surpassed all other states in the South-Central region in current expenditures per student. In 1987 Kentucky ranked 5th in current expenditures per students among the eight states in the region, but by 2006 Kentucky had the highest expenditures per student in the region.
  • KERA has led to a significant decline in differences in educational spending across regions in the Commonwealth. The gap in current expenditures per student between metropolitan and non-metropolitan districts fell from $600 in 1987 to $10 in 2006. Over this same time period districts in the Eastern part of the state went from having the lowest level of current expenditures per student to having the highest expenditures per student.
  • While KERA has led to greater uniformity in expenditures per student in the state, the sources of revenue continue to differ dramatically across areas of the Commonwealth. In 2006 districts in metropolitan areas received 40% of their revenue from local sources (property taxes) with the remaining revenue coming from the state (50%) and the federal
    government (10%). At the same time non-metropolitan districts received only 20% of their revenue from local sources with 66% of their revenue coming from the state and 13% coming from the federal government. For school districts in the Eastern part of Kentucky only 17% of their revenue is collected locally, while for districts in the Northern part of
    the state 42% of revenue comes from local sources.
  • Between 1987 and 2006 the share of revenue coming from local sources increased in Kentucky while the share of local revenue decreased on average in the U.S.. However, there still remains a significant difference in revenue sources between Kentucky and the average state. In 2006 the average percent of revenue from local sources in the U.S. was 42.8% while in Kentucky the percent of revenue from local sources was 31.1%. Among South Central States only districts in Alabama and Mississippi receive a smaller share of local revenue, and districts in Louisiana, Tennessee and Texas all receive a significantly larger share of local revenue.
  • How Kentucky allocates its money on education is quite similar to the average state in the country with approximately 50% of current expenditures going to instruction and 7% to administration. Compensation has a larger share of total expenditures in Kentucky with 70 – 80% of total expenditures going to salaries and benefits in contrast to 69 – 73 % for
    the average state over the period 1989 – 2006.
  • In 1987 Kentucky’s pupil-teacher ratio was 18.6 compared to the U.S. average of 17.4. By 1998 both the U.S. and Kentucky average was 16.5. However, since then the U.S average has decreased at a much faster rate; in 2006, the U.S. average was equal to 15.2 and the Kentucky average was 16.0.

And, in the strongest defense of KERA, proffered by "equity-minded" choice advocates, the researchers state:

Since 1990 differences in educational spending per pupil between urban and rural areas of the state have all but disappeared.

There is no better evidence of the intended impact of Rose v Council for Better Education on Kentucky schools than this. Prior to KERA (the General Assembly's response to the Rose case) spending disparities between Fayette County and Whitley County were 8:1.

Now for personal reasons, I love Fayette County students and want them to be the best in Kentucky. But in the eyes of the constitution - they are not worth 8 times more effort than a Whitley County student.

As long as Kentucky relies on a state property tax (collected locally) to fund its schools, the inequities due to disparate property values guarantee that "rich"counties will always "send" a higher percentage of their school tax dollars to the state when compared to "poor" counties. Thus, we expected to see a higher percentage of state dollars flowing to rural districts, and the converse in urban districts. The only way to fix it is to quit funding schools on a property tax.

The study's econometrics are helpful in gauging the efficiency of the SEEK funding formula, but there should have been no surprise about the findings.

The executive summary concludes:

At the same time there is a growing disparity in the sources of funding with urban districts now obtaining over 40% of their funding from local taxes while rural districts obtain only around 20% of their funding from local sources.

In the second report, the researchers plan to make a case. Expect a bunch of correlations. And remember that correlation does not constitute causation. But I'll bet a nickle right now that the special interest groups already have their conclusions outlined.

Here's the spin on the present study - first cautious reporting from the Herald-Leader:

KERA closes rural-urban school spending gap


Kentucky's spending on elementary and secondary education has grown and evened under KERA, but it still lags nationally, a new University of Kentucky study concludes.

The report suggests that while the 1990 Kentucky Educational Reform Act essentially has eliminated disparities in per-pupil spending between urban and rural school districts, it actually might have increased disparities in the sources of urban-rural educational funding.
"Suggests." "Might have." The much maligned "liberal press" reported it just the way the researchers wrote it.


But at Kentucky Progress, BGI buddy David Adams seems to have read a different report:
"funding has increased since KERA but hasn't quite made it into the classroom very well."
I'm not quite sure what David means, but it doesn't seem to be supported. The figure below shows the researcher's findings on instructional spending.


Perhaps David picked up the line from BGI Director of Policy and Communications Jim Waters.

On the BGI wiki (FreedomKentucky.org), the group clearly shows they are aware of the researcher's reservations - they even highlight the "possible" and "could" language. But I guess that wasn't definitive enough. Waters is quoted by the Friedman Foundation saying, "... it appears that not enough of that increased funding finds its way into classrooms of the commonwealth's public schools."

And I'm not sure why Waters spun his fabrications even further in the Georgetown News-Graphic. Perhaps his heart is two sizes too small.

What a great gig! Co-sponsor a study. Then, quote each other's misinterpretation of the results in different locations. And then, get it into the news before anyone else can look into your claims.

Unlike BGI, the Friedman Foundation chose a straightforward statement that defies an agenda-driven interpretation:

"As Kentucky reviews and refines its school-funding mechanism, it's important to have some perspective on the effect of its previous efforts," said Robert Enlow, executive director of the Friedman Foundation for Educational Choice. "We hope this KERA analysis will be of use to policymakers as they continue to improve their funding and schooling systems."

At the Bluegrass Policy blog, Dick Innes takes some literary license himself - twisting speculation into alleged fact.

For example, the researchers cautiously state their terms, like this:
"What evidence there is suggests that KERA may have....."

But the Bluegrass Institute, being on a mission, has no apparent interest in caution. Innes claims,
"the UK scholars indicate that the way tax supports for schools are divided among local, state and federal sources has acted to reduce local taxpayers’ control over their schools in an inequitable way."

Very definite in his mind. Then, Innes tells us a story. It might be true. It might not be true. But it's a good story. It's got hostages, and peculiar philosophies!

While high-wealth school districts still derive an appreciable amount of their total funding from local taxpayers, that isn’t true in low-wealth districts, which are generally located in rural parts of the state. As a consequence, by resisting tax increases when schools don’t perform, parents and taxpayers in wealthier districts have a somewhat more effective say over what happens in their schools than parents in Kentucky’s poor districts do. Thus Kentucky’s poor and rural school districts are more solidly held hostage to Frankfort and the sometimes peculiar education philosophies that emanate from the capital. Meanwhile, wealthier districts have been more resistant to Frankfort’s ideas and instead have kept a focus on rigorous coursework that better prepare kids for college and life.

Somebody ought to call to call Jon Draud and tell him that Shelly Berman and Stu Silberman are resisting the peculiar philosophies expressed in state law. I'll bet that'll be news to him.

But the UK scholars - while willing to offer weak speculation that "it is still possible that the lower level of local control over districts in rural areas of Kentucky could impact educational outcomes" - are appropriately reserved in noting the limitations of that speculation.

While capital and administrative spending did increase more than instructional spending in percentage terms, some caution should be taken in drawing conclusions from these changes as much less is spent on these functions, making small absolute changes in spending seem more significant in terms of percentage changes.
Problems with the state's chart of accounts continues to frustrate researchers who wish to precisely determine how education dollars are spent in Kentucky.

Say a superintendent hires a curriculum coordinator to help several schools improve. Are they district-based (as many were listed in the early 90's)? Or are they school-based (a later trend)? The job doesn't change - but how a district accounts for it will deternmine whether the position is "administrative" or "instructional." Innes has written about the problems with MUNIS and the chart of accounts since a 2006 OEA report. He and the UK researchers lament the limited amount of clarity as lack of transparency - with justification.

But caution? Apparently BGI didn't pay for caution. They're building a case for school choice and seem to lack the Friedman Foundation's restraint. The game plan seems clear. Anything that can be made to sound like bad news for the public schools - is good news for school choice. Otherwise, it's hard to connect the dots between SEEK funding and choice.

Let's not forget, however, that expanded school choice could be delivered federally if President-elect Obama presses one of his central campaign issues.

But let's wait for Dr Hoyt's second report to see where the UK researchers go with this line of thinking - so as not to jump to conclusions.

That would be bad science.

1 comment:

Anonymous said...

The reports urban areas are Census Bureau Metropolitan Areas.

That means Grant, Larue, and Trigg County, among others, count as urban.

I understand the academic argument, but it doesn't pass the smell test.

Jefferson and Fayette County's still spend substantially more than state average, and the districts closest to Cincinnati come in close behind them.