Stock Values

Whenever I see someone using statistics to prove their point, I always think of the famous quote “statistics don’t lie, people do!” In fact, some people have learned to use statistics for lying so well that their audience never catches on. Fortunately, an attempt by Jared Polis to misrepresent data in the newsletter titled “Lessons Learned: Reflections on Colorado’s Public Education System & Important Information About Your School District” is much more obvious. His conclusion that “Higher funded students have higher test scores” is contradictory to the data you present and no matter how you manipulate the vertical scales on your graph, this fact will not change!

The NAEP reading and writing tests he quotes in the newsletter, are each on a 0 – 500 scale, thus the sum of the four such tests (4th grade reading, 4th grade math, 8th grade reading and 8th grade math) have a total range of 0 to 2000. Yet, this data clearly shows that there was only a 13 point difference (953 – 940) between the average combined NAEP scores of students from low income families, living in top and bottom five education-spending states. Also according to this data, the top five states spent an average of $11,704 per student on education in 2003-04, almost twice as much as the bottom five states, which spent $5,897 on average. Does Jared really want his newsletter readers to believe that less than two-thirds of 1% improvement in test scores is worth doubling the money spent per student in education?

In fact, the data he presents can be used as further proof of his poor logic, by noting that on average students from low income families living in the bottom five education-spending states actually did better on these NAEP tests than the national average. This, despite the national per student education spending being $8,287 in 2003-04 – a whopping 40% more than in the thrifty bottom five.

Perhaps, Jared, like most Americans, skipped his mathematics classes and thus never heard of correlation. Well, let me explain icorrelation in layman’s terms: If two sets of data have similar trends, they are said to be highly correlated. Two perfectly correlated sets of data have a correlation coefficient of 1, two sets of data that have nothing in common, have a correlation coefficient of 0 and those that have completely opposite trends will have a correlation coefficient of -1. With the NAEP data Jared present, correlation coefficient between the money spent and test scores is .8, but without the data for the top five states, the correlation coefficient changes direction and is -0.65. Thus, by excluding only one data point (an outlier in statistical terms), you can make the case that student test scores improve when per student funding decreases! Now, obviously, that would be the wrong conclusion just as Jared's is!

If you are honestly confused about where the money is truly needed and what factors really influence kids’ ability to learn, start by asking those who spend all their time with the kids – the teachers! I asked my sister – she teaches sixth graders at the Omar De Blair Edison Charter School and was nominated by the school for the prestigious Colorado Teacher of the Year award last year. Well, her school is better funded than most in Colorado, receiving $9,654 / pupil in 2005. Yet, teachers at this school, on average, get $35,170, more than $9,000 less than both the district and the state averages. And administrators at this school get on average $71,134, that’s more than twice what an average teacher receives! 6th grade class size is a rather reasonable average of 22.1 students per teacher, yet in the 7th grade, teachers at this school are expected to control an unruly average of 30.4 teenagers! No wonder than that by 7th grade only 18% of students taking CSAP tests at this school tested proficient or better at math, 30% at writing and 51% at reading. BTW, 40% of the students at this school were from low income families.

By comparison, Bennett Middle School, serving the farming community where I live, only received $7,648 / pupil in 2005, yet managed to pay its teachers $38,931 on average, which was over $1,800 more than the district average and hold the challenging 7th grade class size down to 17 students per teacher. As a result, 57% of 7th graders came out as proficient or better at math, 59% at writing and 72% at reading in CSAP tests. Oh, and the percentage of low income students at this school – 23%.

If you are thinking that all the difference between schools like Bennett and Blair are attributable to a smaller percentage of kids from low income families – it is not. Even if every one of those extra 17% of kids from low income families at Blair would have failed every CSAP test they took, Bennett would still come out ahead by a rather wide margin! So, how did Bennett do so much better than Blair, with less money? You don’t need to be a brain surgeon to figure this one out. They reduced overhead and concentrated on education; recruited and retained the best teachers and kept class sizes down to manageable levels. By doing so they created an atmosphere of learning at school, which matched that at home and allowed teachers to impart knowledge on the brains of the willing subjects.

Still think that we need to spend more money per pupil to solve America’s education problems? I hope not! So, now will be as good a time as any for Jared to admit that the data he was abusing was simply not statistically significant, ask for our forgiveness and start looking for better reasons to ask Colorado tax payers to support the various tax increase measures to benefit education. But let’s get one thing straight, first – while there are good objective reasons to pay teachers more and reduce class size, most other uses of education funds (including inflated administrator salaries) are wasteful. So, before we ask for any more money from taxpayers, let’s reduce the administrative waste and use the windfall to pay those underpaid and overworked servants that put their hearts and souls into teaching our kids!

More information on the National Assessment of Educational Progress (NAEP) testing
Colorado School Accountability Reports (CSAP) testing results