I just read something interesting. We have all heard an expert witness give the odds of a match. Suppose the odds of a match are 1/1,000,000; if the case was in New York City and there was no other evidence besides the DNA match (i.e. the match occured by a database of convicted prisoners' DNA), that would mean the odds that the man being accused was guilty would only be 1/18!

Certainly non-intuitive!

I don't think many people have or could be charged on DNA evidence alone - the cops need a sapina or something to get it in the first place and they need pretty good proof to get one of those.

BTW if there were 23 randomly selected people in a room the mathematical chance of two of them sharing the same birthday is 1/2

I've heard the odds are closer to 1:1,000,000,000 -- this means that only five or six people in the world would have DNA that is close (and those would be blood relatives). I've often wondered how OJ's attorneys managed to get the jury to ignore DNA evidence -- especially when the only other possible match was his own family (ie.e, son, father, etc.).

Even at 1:1,000,000 combine that with motive and opportunity -- once again your at pretty good odds the right guy is convicted.

LNT: I am pretty sure it's also not true. How did one arrive at the 1/18 result?

Odds being 1/1,000,000 for a match means that if you pick two people at random, the odds that their DNA would match is , well, 1 in one million :) (pretty straightforward, right?). how big the population is does not come into account.

On the other side, if you want to say that the odds that there are two people in NYC (with a population of 10 mil) with a match are much greater, I agree with you. (let's assume here we talk about unrelated people). But then, what are the chances that one of those two people is the victim of a crime (one) and then the other one is a suspect (two).

(post 200, who-hoo :) )

No one is ever convicted solely on DNA evidence. In fact, the main use of DNA evidence is to eliminate possible suspects as much as it is to make sure you got the right guy.

AFI's comment makes more sense there. But if it comes up a match, then the guy is likely screwed. But I also agree with the equation that T2Bruno came up with in that regard....

Usually, DNA evidence is the nail in the coffin or the way to narrow down an investigation, depending on the matter at hand. You either have everything but need more forensic evidence to advance the case, or you need to start checking your list of possible suspects. As Aldeth pointed out, more often than not, it's a way to exclude suspects rather than prove someone guilty.

Exactly NonSequitur. To use a very simply example - one that is not even criminal - take a paternity test for example. If a woman claims that one of three men can be the possible father, a simple DNA test will definitively eliminate two of the three.

Also, the results are never 100% conclusive, but at times they are so close to 100% certainty that there really can be no debate. Take the OJ Simpson case. The DNA results indicated that there was a 1 in 10 billion chance of the DNA belonging to someone else. Seeing as how there aren't even 10 billion people on the entire planet, it logically followed that there was no one on the planet other than OJ from which the sample came from.

Thought this might help some of you

The focus of most criminal investigations is on linking evidence from the crime scene to suspects, and for more than a century, science has played an increasingly important role in this process. Fingerprinting was applied to criminal investigations beginning in the 1880's. Shortly after the principle of ABO blood typing was reported in 1900, its relevance to forensic investigations became apparent. In the 1960's human leukocyte antigen (HLA) typing became the premier serologic tool for personal identification, although in practice, it was useful for only a small percentage of samples.

Finally, the 1980's ushered in the age of DNA testing, which permits investigators to perform almost unbelieveable feats of identification. With current techniques, it is possible for a single person to be differentiated from all the people that have ever lived using DNA from a single hair root.

You've got the odds thing wrong.
First, the DNA test is not perfect and there's a certain chance that the results will be wrong. This is easily calculated and small.
Second, the original DNA sample is most likely damaged, so you only have most of the DNA to test against. This is like comparing a partial fingerprint to a complete one.
This means that, while the odds of you having the same DNA with someone are astronomical, the odds of your DNA that has been sitting on the sidewalk all night having the same parts remain that you share with a complete stranger are significantly greater. These odds may well be somewhere between 1/1,000,000 and 1/1,000,000,000, depending on how old your sample is and what conditions it was left in.
Also, if the odds are 1/1,000,000 and there happened to be 18,000,000 in NY on the day the crime was committed, that means that around 17 other people should share these DNA strands. Of course, if these other 17 people never met the victim, have no criminal record, and were all at work miles away when the crime was committed, it's not looking good for you.
Although, the others are right that DNA is only one part of a much larger case.