The publication of local school district ITBS scores by the Cedar Rapids Gazette this past fall has initiated a welcome public dialogue about the quality of our schools. While many parents were appalled at what seemed to be a poor performance by Cedar Rapids schoolchildren, local educators and administrators immediately responded that comparisons between schools are unfair, because of the significant role poverty plays in student performance. It is well established in the research literature that when comparing standardized test scores, poverty accounts for a great deal of the difference between schools. It should be obvious that to compare schools on the basis of one raw test score is unfair and simplistic.

However, there is a fairly straightforward solution to this problem. Using a statistical tool called linear regression we are able to analyze how two variables are related to each other. In this case, we are interested in the predictor variable (poverty, using percent of children receiving free and reduced lunch as an indicator) and outcome variable (ITBS score). When this is done using the 21 elementary schools in the Cedar Rapids School District, not surprisingly, we discover a high association between poverty and ITBS scores. In fact, almost 80% of a school’s test score can be predicted by their level of poverty, a number almost identical to that found nationwide by a variety of researchers. This leaves only 20% of the school’s test score unaccounted for.

Using regression analysis, we are able to predict what a school’s test scores are likely to be, based on their rate of poverty. We can calculate the difference between the predicted score and the score actually received by the school, allowing us to compare schools with the effect of poverty having been removed. The diagonal line on the graph represents the predicted ITBS score for a given rate of poverty, while the points represent the actual test score for each individual school. Allowing for some statistical leeway, schools would be expected to fall between the green lines. This model shows that eight of the schools fall into the predicted range, with seven scoring higher than expected and six scoring lower than the model would predict.

My purpose in doing this type of analysis is not to make comparisons among schools, or to try to portray some schools as "better" than others. My point is that it is entirely possible to determine quantitatively how much of a role poverty plays. It is difficult to accept administrators explanations that our schoolchildren fared poorly in comparison to the rest of the state, when Linn County is among the ten counties with the lowest rate of poverty in the state of Iowa. In order to answer this question, we would have to do a similar analysis, using all school districts in Iowa.

Unfortunately, drawing conclusions about the cause of these differences is very difficult for a variety of reasons. First, it is impossible to know whether comparisons are truly fair since we don’t know whether all students have been tested. If a school omits, for example, the special education population from testing, they will obviously have an unfair advantage over schools that require the entire student body to participate. This, too, could be evaluated, were the data available.

It is tempting to explain some of the variation between schools on differences in how and what is being taught. However, while individual schools do have their own personalities and uniqueness, there is very little variation in terms of the curriculum and methods of instruction throughout the district, so this is unlikely to play much of a part. There are many other possible factors that could explain some of the variation in test scores, but without access to the data, they cannot be evaluated in any objective manner.

Although the school district has lamented the newspaper’s publication of these test results, what is needed is more information, rather than less. The school district, and administrators at the State level, have the capability of doing a myriad of sophisticated analyses that could shed light on many of the questions asked by parents. Rather than facile explanations, we want objective data. The analysis presented here is the kind of information we should expect to see routinely from school officials.

Perhaps the most important issue is the mistaken belief that the schools have no control over the 80% of the score that can be explained by poverty. Correlation does not equal causation. There is very good reason to believe that our current "progressive" educational system, based on such notions as "constructivism", "child-centered teaching", and "developmentally appropriate practices" are actually regressive, widening the gap between disadvantaged and affluent children, and magnifying social injustice. In part II, these issues, and their possible solutions, will be addressed.