I am trying to work out a scale for assessing a person’s risk for a particular behavior (like rob a bank). The likelihood of this behavior occurring or not occurring will be based on scores for a number of factors (impulsivity; insecurity; respect for authority; greediness etc.).
A score for each factor is determined as follows: After answering certain questions relating to a factor, people receive a raw score, based on their answers. This score is then converted to a normalized score by comparing the raw score with results from a sample population. So - if someone has a normalized score of 0.64980796071382 for "greediness", he or she is in the 65th percentile . If s/he scores 0.3580796071382 for "respect for authority", s/he is in the 35th percentile.
Let's say I already know that a higher score for "greediness" and a lower score for "respect for authority" means a greater chance that someone will rob a bank. How would I create a scale that combines both scores and comes up with a single percentile rank? What if I did this for several other combinations of two factors? How would I take the scores for each combination of factors and add them together to come up with one final percentile score?
Also, I do not necessarily have access to the information related to the sample used to normalize the percentiles.