In Reply to: Re: Suicide Statistics posted by Thorwald on May 12, 2005 at 12:19:02:
Suicide rates in the U.S. are based on 10-year population surveys. Yearly fluxuations in the base population estimate require regression formulas that factor death and birth rates. I would imagine WHO does a yearly world population estimate using a similar time frame & methodology. Because suicide rates are extremely small, calculating them on a yearly basis shows very fluxuation. Suicide rates are generally calculated for a period of time like every two to three years using rate of population change estimates based on a 10-year census. Suicide rates are also calculated by age groups, with adolescents and the elderly having the highest rates. Comparing WHO suicide estimates to TFI former membership doesn't take into account that TFI former membership doesn't include the high rates among the elderly.
I don't think the methodological issue is so much one of "per year", but how you go about defining the parameters of your base population from which you derive your rate. (To find the "per 100,000", you must multiply the base population rate by 100,000.) If I understand your problem correctly, you want to estimate your base population as "all people who have left TFI to date".
1) TFI membership has stabilized within a fairly narrow range over 31 years. In 1973 when I was a member, the estimated membership was around 10,000. It's approximately the same now at around 12,000. It might have been higher in the 1980s, but that doesn't matter. There's not a lot of deviation between T1 (1973) and T2 (2004).
I can drop 1968-1972 from the equation because that four year period represents growth from zero to 10,000 members. This is outer tail of the total distribution over time, and it skews the estimation.
2) Betcause there's not a great deal of deviation in the membership population between T1 & T2, an average 1,600 leaving each year can be applied to the last 31 years. This average takes into account a higher proportion leaving when total membership was higher as well as a lower proportion of departures when total membership was lower. At 1,600 per year for 31 years (1971-2004), there would be 49,600 former members.
Using the same suicide estimation formula found in my previous post, there would be 22 suicides per 100,000 former Family members. That's still higher than the WHO average, which covers ALL age groups, including the elderly who have the second highest rate.
If you want to me to follow up with you on this at xfamily, please let me know. I think you've got a good idea for an article. I am a statistician, and I talked this problem over with an epidemiologist who calculates this kind of stuff all the time. We enjoy figuring out how to do fuzzy population estimates with funky data.