Money_Does Matter - school cost's in the United States

School Administrator, May, 1996 by Gerald W. Bracey

A typical statement on the link between money and achievement in schools can be found in Reinventing Education, a 1994 book by IBM CEO Louis V. Gerstner Jr., and several colleagues. He states: "During the decade of the 1980s expenditures for education increased by 34 percent (in real dollars), yet the only outcome measures, test scores, were by and large flat." This statement, although typical, is something of a curiosity to me because it is precisely the same statement in precisely the same words and syntax made by Denis Doyle in debate with me in 1992 (Doyle is the co-author of the book).

At the time, I showed Doyle the test scores for different grades for the Iowa Tests of Basic Skills. The scores rose through the mid-1950s to the mid-1960s, fell for a decade, then rebounded. The curves have indeed been flat for the last few years, but they are flat at all-time highs. Of course, if the increases in spending have been mostly for special education, one would not expect any change in test scores. But one finds it anyway. Doyle's resistance to incorporating this information into Reinventing Education is one of a number of incidents I had with conservative school critics that led me to write "The Right's Data-Proof Ideologues" in Education Week about a year ago.

District Correlations

Another common technique used to show that money and achievement are unrelated is to simply correlate the per-pupil expenditures of each district in a state with the districts' test scores. This seldom produces any sizable correlation. In a 1994 study for the Virginia Senate Finance Committee, the per-pupil expenditures for each of Virginia's 141 school districts showed that while a few of Virginia's districts spend considerably more than the rest, most districts look similar in terms of how much they spend on students. This means that spending is not going to be strongly correlated with anything.

An analogy might help understand this phenomenon, known to statisticians as "restriction of range." Suppose you are the dean of admissions at a college. Part of your job is to predict who will or won't be successful at your institution. Suppose further that one year, by some strange circumstance, all students applying for admission have the same SAT score. In this case the "restriction of range" is complete and because all students look the same, you can't tell from the SAT scores which of them will get good grades and which will flunk out. Virginia's school districts are not identical in their spending, but most of them are very similar and, therefore, the correlation between spending and achievement is guaranteed not to be very large.

One can get a further notion about this restriction of range and its impact on the relationship between money and achievement by imagining that spending fell to zero dollars. Surely, achievement would plummet (as it did in Prince Edward County, Va., which, for a time, closed its schools rather than integrate them). Similarly, if we spent $37,000 a year on each student, achievement would soar. The average teacher makes about $37,000 per year, so this figure would allow us to hire a teacher--a tutor really--for each child. The strong effects of tutoring on achievement are well known.