More SAT Charts!

Because you guys just can't get enough of them, I know it.

This one provides the performance by Racial/Ethnic group since 1996 in Math.


You have to marvel at the consistency over time. There is one clear trend - Asians are putting more-and-more distance between themselves and the rest of the field over time.

This chart shows who's been taking the test over the years.


There has been some change over time in the composition of the test takers - most obviously the proportion of white students bottomed out in 2003, and the percentage of all minorities has clearly grown. The numbers of "No Responses" peaked in 2003 as well, and the response rate has improved ever since - not sure what's going on there, but the College Board is obviously having some success in getting test takers to fill in their ethnicity.

A few days ago I posted on the consistency of the "gender gap" in Math among white test takers. Here are the gender gaps for each race over time:


Again, the consistency and the persistence is impressive, even among non-responders and "Other", though there are differences to note. The Asian gap has been clearly narrowing, as has the gap among African-Americans, though in both cases it has ticked up over the last two years. If we extrapolate these gaps into the future, the Asian gap will close in 47 years, that for blacks in 59 years, and the gap among whites should finally close up sometime around the year 5192. The flattening or turning up of the gap over the last few years that is evident in each of these lines could be indicative that the flattening could be permanent. My understanding is that the white gap used to be larger, so if this data went back further we might see a descending line for whites prior to 1996 then a flattening. We'll see what comes out for 2010.

More girls do take the SAT's than boys, but the distribution has been consistent over time, so we can't really blame the failure to close the math gap on an ever increasing proportion of female test takers:


Methodology note: For the comparative scores by race I used the total average for each year as a baseline. I computed the average score of the time period for the mean scores of all students, and then computed the deviation from this average for each year from the average standard deviation (s.d.). Then for each group, I computed their performance for each year as (Mg - Mt) / St, where Mg is the mean score for the group, Mt is the mean score for all test takers, and St the s.d. for all test takers for that year.