House M.D., March 14th 2011 episode: Taub uncovers information implying that a patient has plans to set off bombs at school. He and Masters discuss what to do with the information. The last line says it all.
-----------------
Masters: What are you gonna do?
Taub: What do you think I should do?
Masters: So you can do the opposite? There are kids all over the country doing dumb, potentially violent things, but the percentage of them who would actually kill anybody is minuscule.
Taub: So I shouldn't do anything?
Masters: Mm... while the odds are low, the fallout could be huge. Tens or even hundreds of lives.
Taub: So I should call the cops.
Masters: Of course, over-identification with the subject could skew the calculation towards being over-protective. Or, alternatively...
Taub: You suck at this.
Masters: Hate the statistics, not the statistician.
----------------
http://www.tv.com/house/bombshells/episode/1375867/trivia.html
Everything adds up pretty well in the discussion between the 2, while Masters is correct with his number analysis and whatnot, it almost seems to infer that he is just thinking solely on the statistic of kids in the country who act and do the dumb things that they say they are going to do. Yes the statistic for who acts may be minuscule but there are many lurking variables that need to be taken into effect if you are truly going to analyze what should be done. Something as simple as the personality type of the patient or something like family life could affect the outcome more than school, but I digress. The analysis based solely on numbers points to doing nothing. Most likely the chance that this patient will act will be low, if not even 3 or more z scores away based upon a normal distribution curve. Being human, numbers are not all we rely on though, so in the end someone needs to either consult this person, which I am sure happens or authorities need to be informed. Great show by the way.
ReplyDeleteSamantha Wagner
ReplyDeleteI really like this post because it is a truly modern way of showing the power and fault to statistics. Masters is correct in the statement of that kids do talk and boast about events of violence but statistically few every act upon it. It is also true that when kids do a dramatic act of violence, such as a bombing there are many casualties. In order to make a more accurate judgment many more variables need to be taken into account, and the most important and least predictable is the human element.
This post is interesting. I kept reading "Masters: Of course, over-identification with the subject could skew the calculation towards being over-protective. Or, alternatively..." This made me think because what I believe he is referring to is the fact that the percentage of the event happening is low; however, I feel that he is stating that the data is misleading because it only represents actual occurrences and not thoughts of this occurrence. What I mean to say is that I believe Masters is stating that there are far more people that think about this topic than the data suggests.
ReplyDeleteMatt Gates
ReplyDeleteThe post is very good because it shows a different way statistics is used in the real world. Masters states that there is a very low chance statistically that the person would actually do it. What Masters is saying even though the chance is low there might be more things behind the information like how many people REALLY think about doing something like this. Many more people probably think about this situation but they are not counted in the data or they are counted in the data but really would never say anything or do anything.