People seemed to like my post explaining how changing demographics are masking improvements on the NAEP reading test. One comment that’s been voiced a couple times, by Kevin Drum and others, is that I used the Long-Term Trend NAEP as opposed to what everyone else was talking about, the main NAEP reading scores that were released last week. As I explained in the post, I used the Long-Term trend in this case because it has undergone fewer changes over the years and has been in use for longer. But, my point is not just limited to this version; it also applies to the main NAEP. Here are the score increases for fourth-grade reading from 1992 to 2009, by race/ethnicity:
Overall: +4
Whites: +6
Blacks: +13
Hispanics: +8
Asian/ Pacific Islanders: +19
American Indian/ Alaska native: -7
Only one group, American Indians and Alaska natives, gained less than what we gained as a nation. They made up just one percent of the sample in both 1992 and 2009, so it’s not likely they significantly dropped the average. What changed is the percentage of students who were identified as white versus those identified as Hispanic. In 1992, test-takers were 72 percent white. By 2009, that fell to just 54 percent. Hispanics, on the other hand, went from seven to 21 percent of NAEP test-takers. Each of these groups actually made bigger gains than the nation as a whole did, but because Hispanics score lower, on average, than whites do, our overall total made only modest gains (and, just to cover all my bases, the story is the same for eighth-graders). In other words, the main lesson should be that all groups made gains and gaps are closing. Demographics are hiding those things.
Click Image To Enlarge
5/11/2012 | U.S. News & World Report highlights ideas from "Moving On Up: How Tuition Tax Breaks Increasingly Favor the Upper-Middle Class." 
5/11/2012 | In Diverse, Steve Burd comments on the Obama administration's plans to address student debt. 
5/11/2012 | Kevin Carey talks to Salon about the government's role in rising student debt.
{ 6 comments }
Or to be precise, the bottom 10 and 25 percentile are down since 2002, with the other three groups being flat. If you measure against 2003, of the five groupings, one has declined, one is flat, one went up two points, and one went up by one point.
But I’m also curious when you’ll apply you statistical adjustments to D.C. where they will diminish the gains on 4th grade reading. And your adjustments still won’t change the pattern where D.C.’s gains in 8th grade especially have been limited to the upper percentile since Rhee and the upper percentiles since NCLB, will they?
Chad,
It would be a stretch to use any NAEP results to show that test-driven accountability is working. But NAEP scores make an extremely strong case against NCLB. Can you adjust scores demographically to disprove the standard appraisal that growth slowed after NCLB?
I’m just asking in regard to 4th grade, but in regard to 8th grade reading I don’t see any wiggle room. Any way you cut it, the declines and stagnation of 8th grade reading scores is a powerful (irrefutable?) argument against NCLB.
And if you aren’t interested in evidence for or against data-driven accountability, what is the purpose of your posts? Do you just love the innate beauty of the numbers? If so, I can buy that. Learning is its own reward and teaching and learning have an intrinsic value.
But now that people are contemplating an extension of the worst of data-driven accountability in the name of reform, I’d think you’d want to give equal time to 8th grade scores.
By the way, the bottom 10 percentile in 8th grade reading has gone down since NCLB. So when you say “the same is true” you should say precisely what is that same thing that’s true.
John, first of all, I’ve been very careful NOT to make claims about whether test-based accountability is working or not based on NAEP results. I think that’s a stretch any way you look at it.
Second, in reference to my point about changing demographics making the overall scores look worse than they really are, that applies to 8th-grade reading as well. Table A-17 shows the race/ethnicity breakdowns for 1998 and 2009 (whites fell from 68 to 57 percent, Hispanics jumped from 12 to 20 percent) and income (free- and reduced-price lunch students jumped from 30 to 43). Meanwhile, overall scores have gone up a point, but all race/ethnicity groups have made at least that much progress. The same is true for the bottom 10th percentile.
you wrote, “and, just to cover all my bases, the story is the same for eighth-graders).”
So what was that story for 8th grade reading – the declines and the stagnation was just less bad? So the decline in the bottom ten percentile was just less bad when adjusted?
John, as I wrote in my post, the story is the same for eighth-graders.
Duncan himself has said the elementary test scores do not an education make. If gains can’t even be sustained into 8th grade, much less high school and beyond, how can you say all this test-driven accountability is working? When can we expect your analysis of 8th grade scores, especially among the low performers?
Comments on this entry are closed.
{ 1 trackback }