Causation vs. Correlation

by Martin Harris

Because human nature doesn’t change much (an Age-of-Enlightenment/Founding-Father belief challenged in our own times by Progressives who promise that, once they’re in charge, the perfectible society will result) a medieval proverb composed by long-dead European white guys still resonates: “the poor carpenter blames his tools.” Case in point: the various prescriptions of the Great Society, including the War on Poverty launched in the ‘60’s and based on the “root cause” theory of poverty causing everything from new highs in crime rates to new lows in public-education effectiveness. Various charts show that. $15 trillion of poverty-reduction-spending later, crime rates have shown zero to negative correlation with spending increases –think the growing crime rates of the 70’s and the shrinking crime rates of the 90’s– and K-12 productivity continues to decline as student achievement doesn’t improve while various spending initiatives –think smaller classes and pre-K, for two examples– result in ever-higher costs. The “tools” promised to work –spending on poverty –to the point where the majority of recipients have both multiple TV sets and no-cost cell phones; and class size reductions from near 30 to below 15, to the point where, now, the majority of students can barely read their graduation diplomas– are now blamed as inadequate, but of course would work perfectly if only more money were applied. Food Stamps (now a free credit card) are advertised to build enrollment and program-cost numbers, and Federal achievement tests showing 2/3 of students as incompetent in Reading and Math are dismissed by experts in education as irrelevant to “spontaneity-in-learning”, with creative spelling and the new math having displaced the rules of grammar and the multiplication tables. It’s a “blame-the-tools-but-demand-more-money” strategy which has worked amazingly well in recent decades (social-perfection, never achieved but always more expensive) but is now showing new strains.

The major “root cause” (a little Progressive lingo, there) is the increasingly widespread availability of rigorous studies: with unimpeachable statistical analysis, researchers of class size effectiveness (Hanushek et al) school choice (Hoxby et al) and education productivity (Richard Vedder et al) are now receiving widespread new attention and credibility. Government itself has furnished critical help: without the 40 years of Federal K-12 achievement tests (NAEP) and the recent Federal analyses of pre-K results as statistical foundations, advocates for both programs would still be casually claiming “excellence” even while demanding ever more money. To change the subject, advocates have shifted to focusing more intensively on poverty: the supposedly rigorous causal relationship between inadequate Reading and Math Proficiency and inadequate family income. “Income Gap Affects Test Scores” reads the header for a 28 June op-ed by Brandon superintendent-emeritus William Mathis in a local paper.

Or maybe not: with modest effort, your Humble Scribe could find in the 2012 Annual Report for the Addison Central Supervisory Union the inconvenient stats showing that Ripton (highest poverty-percentage Town in the ACSU per 2000 Census data) students were also at the top for Grade 8 Reading Proficiency and Proficiency-with-Distinction: 100%. The wealthiest (2000 Census) Town, Cornwall, students came in lower at 93%. Lowest-household-income town Middlebury students came in at 91%. In Grade 8 Math, second-wealthiest town, Weybridge, students scored 72% P+PwD, noticeably worse than middle-income towns Salisbury and Shoreham where students made 93%. Yes, sometimes, there may be rough inverse correlations between poverty and achievement, but there are so many exceptions that Mathis-style causality-claims by education careerists/experts who know better (after all, these scores are their stats) are questionable for both stated “fact” and un-stated ideology. They might want to look at motivation, both student and teacher, and their own politically-correct curriculum-content requirements.

* * * * *

Unlike the “smaller classes produce better achievement results” trope, seemingly logical but for 40 years refuted by test-score facts and therefore now being challenged in some quarters (a handful of States, for example, has adopted minimum-class-size standards over educator and teacher-union objections) the “poverty produces lower achievement results” trope, similarly first-glance logical but so far not seriously challenged, still receives general acceptance when recited. The ACSU Annual-Report results (arguably depicting a fairly typical non-urban regional school district with a range of socio-economic-status households and students) notwithstanding, it’s likely that few readers have examined the town-by-town test scores and noted the many instances wherein rich-town students did less well on average than poor-town students. The Report’s author dutifully printed all the stats but with some confidence that the internal anomalies refuting their basic argument would get (a little Senator Moynihan lingo, here) “benign neglect” from readers. They certainly weren’t pointed out by the Report’s authors. And just so for the Report’s exclusive use of NECAP (New England Common Assessment Program) test scores. That’s because they’re, by design, nicely higher than the Federal NAEP (National Assessment of Educational Progress) test scores, similarly administered annually to a statistically-selected sample of the same universe of Vermont and ACSU students.

Case in point: the Federal NAEP test scores, not in the ACSU Report, show Vermont Grade 4 Reading results (2009) at 42% Proficient-plus-Advanced while the ACSU/Vermont NAEP results show 70% Proficient-plus-PwD. When your annual per-pupil spending is half-again the national average ($15 thousand vs. $10 thousand) because your class size is a third smaller (10 vs. 15) you’d prefer to claim 70% on a four-State test, designed and sold to you as easier, rather than call unwanted attention to the 42% shown on a 50-State + DC nationwide test. That’s because 42% Proficient means (a little Grade 4 math here) 58% non-Proficient, unable to function at grade level. You can find some of the Federal NAEP scores on the State Ed Dept website, but the full detail is in the annually published National Digest of Educational Statistics, which your Congressman or Senator-of-choice will happily(?) get for you. Of course, even 70% isn’t good enough; which explains why Vermont has already tried first New Standards Reference Examination, has been using NECAP, and has now contracted for a new Common Core test, cutely labeled “Smarter Balanced”, to produce seemingly even-higher Proficiency results.

You could call it causation rather than correlation, or “blame-the-tools”, but: when the test vendor fails in his promise to produce the nearly-100% real-Proficiency “result”, the consumer (Vermont schools) votes with his feet. Just as school-choice parents would like to do, but can’t.