A researcher in Louisiana notes that John White and his fellow advocates for vouchers were overjoyed by the latest study by voucher advocate Paul Peterson of Harvard and Matthew Chingos of the Brookings Institution.
Here, they hoped, was proof that the Louisiana voucher program would boost college enrollment rates for African American students!
But, Noel Hammatt of Baton Rouge points out that they ignored some inconvenient facts.
The voucher-receiving group of black students had a 24% increase in college enrollment, but the voucher-not-receiving group had a 20% increase in college enrollment. Hmm. Not so amazing, after all.
And the African American students in the study, in comparison to Hispanic students, were more likely to be children of college-educated parents and only children.
Overall, the study found NO significant effects. (See my own post here.)
But this what happens when politicians politicize research and researchers give them ammunition to do it.
Diane Ravitch's blog 
Thanks for your reactions to our study, in both this post and a previous one. I wanted to quickly correct an error in this post. The 20% increase is the effect of being in the treatment group (offered a voucher) as compared to being in the control group (not offered a voucher). This effect includes the impact of using the voucher among students who used it (as 81% of African-American students did). When we scale up the 20% impact to reflect the fact that 81%, not 100%, of African-American students used the voucher, we calculate that the impact of using the voucher was 24%. Because the impact of the “offer” includes the impact of using the voucher, it doesn’t really make sense to compare the 20% and 24% estimates in the way that Hammatt does (and others have).
Describing the impact of a voucher offer in the way that we do in the paper is standard in social science research (where it’s called the “intent to treat” estimate), but it has been confusing in the popular discussion of the study because of the ambiguity about whether the offer effect includes the use effect (which it does). I hope this clarification is helpful.
I wonder if this statement from your report increases the ambiguity ambiguity for most readers. “A voucher offer is shown to have increased the overall (part-time and full-time) enrollment rate of African Americans by 7.1 percentage points, an increase of 20 percent. If the offered scholarship was actually used to attend private school, the impact on African American college enrollment is estimated to be 8.7 percentage points, a 24 percent increase.” Having received your clarification, I of course acknowledge my mistaken (and it appears to be a fairly common mistake) understanding of this aspect of your report.
Without the data it is difficult for readers to ascertain the accurate meaning and possible conflicts between statements in different portions of the report. In fact, I point this out in my response to the report, when I say “There is little consistency in terminology, requiring the reader to often re-examine the document to determine what the authors are attempting to claim.”
I do note that this is appears to be the only one of numerous concerns raised in my response to your report that you are questioning. I invite readers to read your entire report, along with my response to it. Carefully examine the evidence, and then answer the questions I have raised, and others you may have, for yourself. It appears that the full report fails to mention what we only know about from having read the earlier studies by Peterson et al. on this voucher program. First, that it was acknowledged in the earlier study that those receiving offers of vouchers had a slightly higher income level (and of course, I noted that there was no discussion of income in this latest report, other than a mention that all students were from “low-income” families) than the control group. I am also wondering if you would care to explain the lack of apparent interest in why the students in the “voucher offer” group (for African-Americans only) as opposed to the control group for African-Americans had 33% greater likelihood of having parents with a college degree and also had a 33% greater likelihood of being single a single child. Remember, there were not variations of this magnitude between the “voucher offer” and the “control” groups for Hispanic students.
I look forward to your response! And thank you for helping us better understand this report.