WILL Blog | Flanders: Responds to critics of MPS, MPCP crime study

New Report Shows Milwaukee Voucher Students Less Likely to Commit Crime

Opponents of Parental Choice Program mischaracterize report, offer nothing but hyperbole

Whenever an academic study is released with findings that show the benefits of the Milwaukee Parental Choice Program (MPCP), opponents come out of the woodwork with a predictable stream of inaccurate criticisms.

A recent study by Corey DeAngelis and Dr. Patrick Wolf found that long-term students at MPCP schools are less likely to commit crimes than comparable students at Milwaukee Public Schools.  Sadly, in criticizing the study, Milwaukee Public Schools and state Rep. Sondy Pope (D-Cross Plains), an opponent of school choice, mischaracterize the findings of the study, the meaning of inferred causality, and common understandings of “scientific progress.”

Mischaracterizes the findings of the study: According to MPS, “The authors admit that they cannot conclude that the voucher program affected criminality . . . .”  While MPS would be correct if they only examine students who attend private schools with a voucher for a few years, the key findings of DeAngelis and Wolf are among students who attended MPCP schools for a full 12 years.

MPS claims that it is improper for the scholars to examine students who attend private schools over the long term, but offer little justification for why this is incorrect.  The potential benefits of private schooling—in exposing students to a new school system, morality based on religion, and the soft skills needed for building social capital—may take several years to manifest.  To say that the study need only focus on the instantaneous effects is no more justifiable than saying a study of a drug cannot examine its effects years down the road.

Mischaracterizes the meaning of inferred causality:  MPS attempts to discredit the study for utilizing inferred causality.  “…[T]hey treat the relationship as “causal” any way in part because they have “theoretical reasons to expect” that private schools reduce criminal behavior and because no prior research exists that tells them differently.”  While it is true that the DeAngelis study does not provide absolute proof of A causing B – a near impossibility when studying children — the sophisticated statistical methods utilized was in fact created to approximate causal relationships.[1]

As I wrote this morning, the authors matched students in the Milwaukee Parental Choice Program (MPCP) with similar students in Milwaukee Public Schools (MPS).  This sophisticated matching method represents the cutting-edge in quantitative analysis, and comes as close as possible to the “gold standard” of experimentation.  DeAngelis and Wolf compared students in each group with Wisconsin criminal records data bases to determine if there are differences in the likelihood that each group of students committed a crime in the years following high school.

The “apples to apples” comparison that matching creates allows the creation of a control group very similar to the treatment group, and the effects observed have been shown to be very close to those that are found with true experimentation.[2]

Some might argue that a better control group would be those that stayed in MPS for 12 years.  However, the matching process allows the authors to compare outcomes between students who were equally likely to complete 12 years in the choice program rather than with a set of MPS students who might vary greatly in their likelihood to have completed those years.  Therefore, it would be “apples” to “oranges” to compare the outcomes of the set of graduates at both types of schools.

Mischaracterizes “scientific progress.”  As Rep. Pope wrote, “Strong theoretical reasons’ do not amount to evidence and the absence of previous research on a topic does not validate weak research.”  Putting aside the fact that School Choice Wisconsin has previously conducted research that, at minimum, is consistent with the findings here, Rep. Pope does not quite understand how scientific progress is generally made.

Scholars produce findings, and then the research community has an opportunity to critique those findings.  Any problems identified serve as the basis for future research.  And the process repeats itself.

Neither MPS nor Representative Pope offer legitimate directions for future research, nor do they identify any serious methodological errors in the existing study.  Until such criticisms are levied, and a study conducted that finds differing results, the work of DeAngelis and Wolf should be taken as the preeminent study on this topic.

In a state where 1 in 8 African American males is imprisoned, schools that can have a strong impact on reducing crime among this group ought to be lauded.  Instead, cognitive dissonance among school choice opponents leads them to dismiss a solid study.


[1]Rosenbaum, Paul R.; Rubin, Donald B. 1983. “The Central Role of the Propensity Score in Observational Studies for Causal Effects”. Biometrika 70: 41–55
[2] Bifulco, Robert,  Ladd, Helen 2007. “School Choice, Racial Segregation, and Test-Score Gaps: Evidence from North Carolina’s Charter School Program.”  Journal of Policy Analysis and Management 26: 31-56.
Share This