Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 2 down vote favorite
1

One can argue that the evidential value of a database hit increases when the database gets larger. Indeed, having a unique hit in a database simply means that all the other persons in the database are innocent. Therefore, if the database is larger we have the extra information that many more people are innocent. Clearly, this increases the weight of the evidence against the suspect.

  • Is this reasoning correct?
  • Is the probability of a unique match the same, irrespective of the size of the dataset?

This is how I see it: if the database contains everybody, then a unique hits tells us with certainty who the donor of the DNA prole is. So it really is true that a larger database is stronger evidence. But then I started thinking: suppose the database consists of 16,000,000 proles, and suppose that the DNA prole we are looking for has frequency 1/1,000,000 (one in a million). Is it still reasonable to believe that the underlying model is correct?

share|improve this question
One might temper this sort of reasoning, which is based solely on counts, by considering the differences between (i) a huge database that contains no immediate members of a person's family and (ii) a database that contains only members of a person's family. These extreme examples further suggest that the amount of "evidential value" may depend on whether the result is or is not a hit and on what conclusions are being drawn as a result. These examples also show that the question requires more information about the probability model one has in mind for matching. – whuber♦ May 14 '12 at 15:48
In your last paragraph are you implicitly comparing two models where you have a sample of 16 million and a population of (how many?): A), where an event occurs one in a million and B) where an event is unique in the population? And you want to decide whether A or B is more plausible, given an observation of one event out of 16 million? – Peter Ellis May 17 '12 at 1:02
Is there a chance of a spurious match? Can a large enough database give you two matches and what'd you do then? – curious_cat Mar 11 at 12:26

2 Answers

up vote 0 down vote

Keep in mind that the case against a suspect is stronger in a database match than in a so called cold" case, but contrary to a popular idea, this is not because likelihood ratio increases when compared to a cold case, but because posterior odds are higher in the database case.

share|improve this answer
up vote 0 down vote

Not sure what do you mean with DNA prole.
But a DNA profile with a frequency of 1/1,000,000 is a very bad profile.
Forensic DNA profiles are based in highly polymorphic regions of the DNA.
Not one region: at least 13 different regions.
Therefore, if you find a match in your database you can assume at least that the suspect or a close relative (including monocygotic twins) was the perpetrator, and that's independent of the size of your database.
Of course, the bigger the database the higher the chances of finding a match.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.