Human Rights Workshop discussion continued..
Filling out precious designated workshop time with talk about posterior probabilities and likelihood ratios would give anyone a freakin’ headache. So, how about an example? Salado Puerto used an example of a kinship analysis in endogamic locals. Endogamic, endo meaning inside, can be used to describe humans that are characterized by their traditions of marrying within the tribe. If you think about a small island with people who don’t leave it and that they reproduce, you can gather the understanding that the people of this island marry and reproduce within the existing population. And yes, there are people on this earth who exist under these circumstances (no negative implication). These people have very similar DNA, so making a match would be not only difficult but the math would be forced. As if the investigator would be forcing the sample to fit someone’s DNA exactly. This forcing theory mixed with other factors, such as indicating a positive match because that’s what you or other people want to hear, or you have formed your own opinion and you make conclusions based on what you want the match to be or not, which may not be intentional, hopefully. Some things to consider: is this an adventitious match? What is the prior probability value? And do we use complementary analysis?
Salado Puerto says the experience of the investigator is also very important, that the probability is the threshold of coming to a conclusion; and then, make an ID combined with other methods. “Integration” is the term used for various analysis and a variety of professionals working together. A tip that we got from the workshop was to form Identification Committees because the best solution to finding a match is to use a combination of methods.
Lucky for you I have an elementary level of understanding posterior probabilities and likelihood ratios so I can just get to the point. The posterior probability is the probability of observed frequency, like how many times something happens within a sample, or the probability that Event A will happen just as long as Event B has happened. The likelihood ratio is the likelihood that your sample will fit into a known population. Either way, you are comparing things to find the best fit. If you want to make a positive match within an endogamic group of people, you would find difficulty since you are comparing people have similar DNA. You’re forcing the sample to fit into a smaller group so the individual that you are trying to match matches too closely with the others in the group. If this individual was compared to a worldwide sample, then they would be more identifiable, or easier to identify compared to individuals of other geographical and familial backgrounds. A positive outcome of this method in an endogamous group can eliminate the sample from the group by showing a low probability and low likelihood. Ruling out can be beneficial in the process of making the positive ID, too.