Unfortunately, some new research seems to suggest that not only is this explanation wrong, but that there might not even be a true observation that needs explaining. Temrin, Nordlund, Rying and Tullberg looked at data in Sweden from 1965-2009 to answer the question of whether this apparent effect is the result of parents being more likely to kill stepchildren, or whether it is simply a result of stepfamilies having higher rates of violence which would skew the data.
Our results show that crime rates are higher among parents in stepfamilies than in families with two genetic parents,
both in the general population and among perpetrators of child homicide in our data set. This indicates that parents in
stepfamilies have characteristics that can increase the risk for child maltreatment, and we regard our results as support
for the hypothesis that the higher incidence of parental child homicide in stepfamilies is an effect of a negative
selection into stepfamilies of individuals who have a higher risk, on average, for committing child homicide. This
explanation for the higher incidence of child homicides in stepfamilies cannot easily be placed in the classical scheme
of ultimate and proximate explanations, but can rather be seen as a confounding or spurious factor that explains the
Their results appear to be a perfect falsification of the Cinderella effect: of 27 homicides committed in stepfamilies, 13 stepchildren were killed, 13 biological children, and 1 case of both.
Of course, the research here is still fairly young and needs to be replicated in other areas with different sample populations, but at the very least we have reason to think that the horse and carriage outside the party have just turned back into a mouse and pumpkin.