Changes

One issue that I don't see addressed explicitly enough above (though I may have missed it, in which case apologies) is that of bias. I'll pick out just two sources of bias that (as far as I can see) completly undermine the article:

One issue that I don't see addressed explicitly enough above (though I may have missed it, in which case apologies) is that of bias. I'll pick out just two sources of bias that (as far as I can see) completly undermine the article:

# What is the source of information on breast cancer cases among "Hollywood types"? It comes from the media, and is therefore heavily influenced by the Hollywood publicity machine and the priorities of the newspapers, TV channels, etc. The control group (the rest of the world, other female residents of California, whatever) is not subject to that sort of bias. Nobody reports a story about "32-year-old housewife dies of breast cancer". The only way to do this comparison is to ignore all information that comes from the media, and use an equivalent data source - which I guess means official epidemiological data - for both Hollywood and control groups.

# What is the source of information on breast cancer cases among "Hollywood types"? It comes from the media, and is therefore heavily influenced by the Hollywood publicity machine and the priorities of the newspapers, TV channels, etc. The control group (the rest of the world, other female residents of California, whatever) is not subject to the same influences. Nobody reports a story about "32-year-old housewife dies of breast cancer". The only way to do this comparison is to ignore all information that comes from the media, and use an equivalent data source - which I guess means official epidemiological data - for both Hollywood and control groups.

# The selection of cases in the "Hollywood" group is subject to serious ''selection bias'', because a woman of roughly the right age and roughly the right profession is assigned to the group after (and perhaps, one suspects, ''because of'') the diagnosis. You must ''first'' define the population (the denominator) and ''then'' look to see how many of those who are in that population have the condition (the numerator).

# The selection of cases in the "Hollywood" group is subject to serious ''selection bias'', because a woman of roughly the right age and roughly the right profession is assigned to the group after (and perhaps, one suspects, ''because of'') the diagnosis. You must ''first'' define the population (the denominator) and ''then'' look to see how many of those who are in that population have the condition (the numerator).

I'd be very interested to see (a) the hypothetical causal link between a high incidence of breast cancer and "Hollywood" and (b) some proper data, properly gathered and analysed. Insistence that "it's obvious", and labeling those who ask for evidence as "liberals", is no substitute for a solid hypothesis, hard data, and a proper analysis. [[User:Humblpi|Humblpi]] 06:05, 8 May 2008 (EDT)

I'd be very interested to see (a) the hypothetical causal link between a high incidence of breast cancer and "Hollywood" and (b) some proper data, properly gathered and analysed. Insistence that "it's obvious", and labeling those who ask for evidence as "liberals", is no substitute for a solid hypothesis, hard data, and a proper analysis. [[User:Humblpi|Humblpi]] 06:05, 8 May 2008 (EDT)