Here's part of the Abstract:
Controlling for multiple potential confounders with logistic regression, both old-for-grade status and grade retention are independently associated with increased rates of behavior problems. Separate logistic regression analyses for blacks and whites showed that these findings pertained only to white children.
Conclusions. Whereas grade retention is associated with increased rates of behavior problems in children and adolescents, simply being older than others in one's class, without having experienced grade retention, is also associated with increased rates of behavior problems, most noticeably among adolescents. These data suggest that there may be latent adverse behavioral outcomes that result from delaying children's school entry.
Comparing rates of extreme BPI (Behavior Problems Index) scores with various combinations of old-for-grade and grade retention status reveals that both factors are associated with increased rates of reported behavior problems. Old-for-grade children who had been retained a grade or more had the highest rate of reported extreme BPI scores (19%); those who were old-for-grade without retention had a rate of 12%; of those who were not old-for-grade, children without a history of grade retention had the lowest rate of extreme BPI scores (7%)... Children who are old-for-grade without having been retained have a rate of elevated BPI scores that increases with age, with rates similar to the lowest-risk group (neither old-for-grade nor retained) when they are younger and approaching the rate for the highest-risk group (old-for-grade youth who had been retained) in the adolescent years.
From the Conclusions:
"At what age should children start first grade?" is a question that can have a lifelong impact on a child. This study suggests that being older than the majority of one's classmates can adversely affect children's behavior and that this effect is independent of having been retained a grade in school. These findings pertain only to white children, as neither old-for-grade status nor grade retention were independently associated with increased reported behavior problems in black children. Furthermore, the emergence of behavior problems among those who are old-for-grade without having experienced grade retention seems to be concentrated among adolescents.