Should the grade-level a student is in be based entirely on how old he is or at least partially on how skilled he is? This is the fundamental question underlying the debate over social promotion — the practice of moving students to the next grade regardless of whether they have acquired the minimal skills covered in the previous grade. Advocates of social promotion suggest that it is best to group students by age rather than by skill. Students who are held back a grade are separated from their age-peers and, the argument goes, this social disruption harms them academically. Opponents of social promotion favor requiring students to demonstrate minimal skills on a standardized test before they receive automatic promotion to the next grade.
Until now the bulk of the research favored social promotion. Most studies found that students who were retained tended to fare less well academically than demographically similar students who were promoted. The problem with this previous research is that it was never entirely clear whether retained students did worse because they were retained or because whatever caused them to be retained led to worse outcomes. This is especially a problem because these previous studies examined retention based on educator discretion. If teachers decide that one student should be retained while another demographically similar student should be promoted, they probably know something about those students that suggests that the promoted student has better prospects than the retained student. When researchers match students on recorded demographic factors they cannot observe or control statistically for what a teacher saw that led that teacher to promote one student while retaining the other.
But in a new study we conducted for the Manhattan Institute that avoids the pitfalls of earlier research, we find that holding low-performing students back helps them academically. We examined a policy in Florida that required third-grade student to perform at a certain level on the state’s reading test to receive an automatic promotion to fourth grade. Students who performed below the required level and repeated third grade made significantly greater academic progress than similar students who were promoted despite their lack of skills. The benefit of being retained grew so that by the end of the second year the retained students entered fifth grade knowing more than the promoted students did leaving fifth grade — this despite the fact that the retained students had not yet been exposed to the fifth grade material.
Of course, the key question is how do we know that we are comparing similar students when earlier researchers had so much difficulty making apple-to-apple comparisons? We are helped by the fact that retention decisions in Florida were based on the adoption of an objective, test requirement rather than educator discretion. This allowed us to pursue two strategies for making apple-to-apple comparisons. First, we could compare the academic progress made by low-performing students the first year the requirement was adopted to the progress of similarly low-performing students the year before the policy was put in place. These two groups hardly differed except in the year in which they happened to be born. Whether students in these two cohorts were promoted or retained did not vary according to unrecorded qualities that informed a teacher’s decision, but by whether they happened to be born in one year or another.
Second, we could compare the academic progress of students who were barely above the minimum test score (almost all of whom were promoted) to the progress of students who were barely below the required test score (most of whom were retained). Since there is some random error in testing, students barely above the testing threshold are hardly different in their academic ability from those barely below the threshold. The luck of answering one more question correctly might be all that distinguishes students in the two groups. So, the two groups would be very similar but one was likely to be retained while the other was not. Again, this allowed for a very nice apple-to-apple comparison.
Whichever way we looked at it, the result was the same: retained low-performing students made significantly greater academic progress than promoted low-performing students. Of course, this study does not definitively prove that test-based retention is beneficial. For one thing, researchers using similar methods analyzing a similar program in Chicago found that retained students fared no better or slightly worse than promoted students. There are important differences between how test-based promotion was implemented in Florida and Chicago that could explain these different results.
The point is that we have strong evidence from Florida that test-based promotion requirements can significantly enhance the achievement of low performing students. If those positive results continue and can be replicated in New York or Texas, where similar programs also exist, we may have to rethink the widespread idea that students have to be grouped in grade-levels by age rather than by skill level. Perhaps more students will benefit by being taught at a level appropriate to their skills. And perhaps school systems will be motivated to ensure that students acquire the required skills if they can’t simply pass students along regardless of their achievement.
— Jay P. Greene is endowed chair of education reform at the University of Arkansas and a senior fellow at the Manhattan Institute, where Marcus A. Winters is a senior research associate. They are authors of the bookEducation Myths.