On a fairly regular basis, I'm faced with a parent who wants help interpreting some standardized test scores. Often, the test scores are in the form of a percentile rank. Many people know what percentile rank means, but in case you're not entirely sure, it is a measure of how well someone has done in comparison to the rest of their group. For instance, if a child achieves a percentile rank of 60 on a certain portion of the ERB or ISEE, it means that he or she has done better than 60% of his or her peers. It does not indicate that the child got 60% of the questions correct. In fact, it doesn't say anything at all about the percent of questions that the child answered correctly.



Most of the parents I work with know quite well what percentile rank means. But, they often act on the numbers as if they are percentages on a classroom test. By this I mean that they will look at a ranking of 75 percentile, as compared to independent school students (generally, an elite group) and act as if that's an inferior score.

I suspect many parents make a subconscious connection between percentile rank and percent and therefore don't see a 75 for what it is because that number would normally indicate something unsatisfactory.

In a way, this phenomenon is good for me as a tutor- when parents worry, I get new clients. Nevertheless, I'm not happy about it. It results in unrealistic expectations that cause both students and parents great anxiety. After all, it's not possible for very many people to be in the 95th percentile, nor is it a measure of failure to be below it.

I believe that appropriate changes in the way scores are reported could go a long way towards alleviating unnecessary anxiety (without lowering standards). One approach might be to eliminate percentile rank and replace it with a more fortuitously named ranking system such as stanines. Stanines are a method of breaking down scores into nine groups, with the top group having a rank of nine and the bottom group having a rank of one. Stanines give roughly the same information as percentile rank and are much harder to confuse with percentage. One disadvantage is that few people understand stanines, so although the ERBs and ISEEs already report scores as stanines (in addition to reporting them as percentiles), most people overlook them. Another disadvantage of stanines is that because the groups are relatively large and all stanines are integers, two scores in a single stanine are sometimes farther apart than two scores in adjacent stanines.

Alternatively, actual percentages could be added to score reports. This would have the effect of shifting the focus to the individual student's performance and away from comparison with the group. Of course, when it comes to school admissions, comparison with the group is most important from a practical point of view, so a percentage would not be a sensible replacement for percentile rank, merely an addition to add context.

Neither one of the suggestions I make is fully satisfactory, although I think they move in the right direction. I'm not a statistician and therefore don't really have the tools to come up with the best way to report scores. My goal here is to point out a problem and hopefully open a discussion.