How is this showing AA is failing. This gives a single data point from 2005. It does not indicate a drop, no change, or increase in graduating percentages or total aggregate. All this says is that blacks graduate from college at a lower rate than whites. Also, seeing how the percentage of whites going to college also increased during the period shown in the first chart, how much this falls on AA is completely unsubstantiated.
Yeah, the kind of data they are showing and the way they are using it fails in every way to make the case that AA is ineffective. Even if we call that blip from 2009-2011 in admissions a "closing of the admissions gap", the graduation data is from 2005, meaning its from student who entered college between 1999-2001. Apples and oranges. Also, the admission rates include students in 2-year colleges, aka community colleges that don't use AA because they don't have standards and where most students don't ever go on to a 4-year school so of course they cannot graduate with a Bachelors. Minorities are more likely to go to a community college, so they are less likely to get a 4 year degree. No shit Sherlock.
Besides, trying to use longitudinal data is just not going to give you an interpretable picture. There are way too may things that vary across time (such as the economy) that have a big impact on who applies, how many apply, who goes, what major they take, and if they graduate.
IF the point is to argue that AA leads to a lower graduation rate among the groups targeted by AA, the better way is to merely point to the rather clear data that across schools and across years, and across gender and racial groups, students who get into a 4 year college with lower HSGPA and standardized test scores are significantly less likely to graduate. Those data are out there and the impact of these factors is large. Based only on this, it is really just a logical necessity that admitting more minorities with scores and GPA at the lower end of those admitted (which is the only people AA admits that wouldn't already be admitted) will lower the graduation rate of these groups. In turn, the non-minority students with higher scores who didn't get in as a result (it is a zero sum game), will necessarily be at the lower end of the scores and GPA among non-minorities. Thus, removing them removes the non-minorities that were least likely to graduate anyway, thus increases grad rates for these groups. In sum, so long as the non-racial admissions criteria predict grad rates (and they do), then it is a deductive certainty that AA admission policies will increase the grad rate gap between those groups AA is trying to "help" and other groups. Note that this is true, even if AA does increase the raw number of minority graduates (because it also increases the number of minority dropouts).