Unfortunately, there has never been much empirical evidence to support the notion that extemporaneous decision making involves less decision making. The Princeton philosopher John Fisher once said, “There is nothing wrong with imprecision in analysis, but for experimentalists it is dangerous because there is little effect to it.” Although the idea that “morality” is less a set of fixed options with ironclad dispositions than an “imperfect, emergent property,” as one study claims, may have some theoretical agreement with this view, there is no compelling evidence.

Today, more evidence comes down the chute from one study in Evidence-Based Education titled, “Free Choice and Recidivism in Lowering Student Recidivism.” The author used online data to analyze data on nearly 32,000 New York City public high school students who started in 2011-2012, had a disciplinary period in 2012-2013, and were identified as high school dropouts in 2014-2015.

Students were assigned between five and 24 days of support (“instructional time”) as well as 12 or 15 days of that support in a pre-incentive period of three weeks. The five-day support period was the same, timewise, for all courses taught in a curriculum but based on different math, English, science, and social studies content. The study then controlled for which courses offered the instruction, using a Public Policy + Economics model that included multiple variables for each student.

Results showed that approximately 50 percent of students lost the support during the intervention period and still dropped out. The students who lost support had “better disciplinary outcomes at a significant income level” and also exhibited fewer suspensions. However, the students who lost support at the highest income level (sometimes as high as $100,000) had significantly worse discipline than the others during the intervention.

The experiment included a free choice option. Although the choice was in a controlled environment, the search for facts meant that students would be biased toward fact rather than bias. The authors conclude that a proactive relationship needs to be set up between income and support during this period, or it is unlikely to provide students the benefits they deserve. This is unfortunate: Everyone would lose if students are dependent on their social network for support. They also recommend that more focus be given to an alternative which might include a reduced course load or smaller support cohort.

In my own work, I analyze empirical findings on “means-rated high school graduation rates” but also the costs and benefits of high schools. I want a system that gives students the most support and is personally comfortable, and supports individuals, families, and neighborhoods. It will take time to get there, but the science is changing to assist in this process.

The research I’ve highlighted today offers enough evidence to offer caution when recommending tough programs that impose great costs on schools. However, the data indicate that in addition to equity, there is another — and even more important — value to value-added based measures.

Psychologist Sylvia Noonan Fischer has shown that from 1975 to 2005, educational reforms had the effect of increasing social and racial segregation. This raises issues for policymakers who might suddenly be confronted with data about new disparities that the reform has created. There is good news: Various studies show that, unlike in Nash’s “redistributional symmetry,” the racial difference that we were able to detect was much smaller than we had expected.

This could be the hope for advocates of rigid, one-size-fits-all interventions. Of course, this must be carefully tested.

As John W. Wilson, Professor of Educational Policy and Management at the University of Pennsylvania wrote, “It is important to consider concerns like ‘good and evil,’ rather than ‘if it works well it will be evil’. Policies that work based on inherently evil intentions often lead to worse outcomes.”