I would like to study how the death of a respondent's parent affects his personality (or whatever DV). I have an age of respondent, and also an age when respondent lost his parent. The problem is that these records are available only 23 years back. That is if somebody lost his parents 25 years ago this information is not available. That means that the two variables (age and age-at-parent's-death. APD) are highly correlated - because nobody can lose his parents at age that exceeds his own, and nobody can lose them earlier than 23 years back (because there is no this info in the dataset).
If I want to study if the loss of a parent at early age affects him more than when he gets older, is it any reliable way to do it with existing dataset?
Here is scatterplot of two variables (age, and APD):
