sundin13 said:
Basically, what your math tells you is "how many victimizations occurred over the last 81 years" (by assuming a constant victimization rate, which I'm fine with). Then, it seems you take this number and try to say "all of those victimizations happened to people who are currently alive". If you are using crimes from the past x amount of years, you don't know if the victims are still alive or not. Because of that, you would theoretically need to include everyone who has been alive over that period, increasing the population sample, which I did before, although that gives you the "percentage of people who have lived over the last 81 years who have been victimized over the last 81 years" (more or less), which isn't what you are trying to say. In order to say how many victims are alive today, you have to remove everyone who has died between the time their original victimization took place (which is present in your data set) and the present in order to say how many victims are currently alive (which is clearly not possible with the current data) The problems comes when you try to apply yearly data to "lifetime" data, which is unprovided by the BJS. I understand that the 1990 data doesn't include crimes happening in 2016, but a victim from 1990 may not be alive in 2016, so you can't include them as part of the "lifetime" total you are using for people alive today. As for my 1 to 10 figure, I didn't mean to imply we should use that number going forwards. I think we should do research that looks into the actual lifetime prevalence, so we don't have to use a ghetto number. What I meant, is that we should talk about "lifetime victims" by year of birth, not the year they were alive. This allows much better computation and comparison of trends. For example, if using a "all people who are alive" standard, you deal with people who are 100 years old, who may have been victimized 90 years ago, which doesn't say much about the present, but does say something about 90 years ago. On the other hand, we also have people who were born yesterday, who also don't tell us much about current trends. Because of that, the data gets confounded, so using a "by birthdate" system would clear things up a lot. Also, as I posted in an earlier edit (which I posted a little late so you probably missed it), BJS does include penetration with objects in its rape statistics. Also, its very worth noting that the BJS statistics are not rape statistics, but instead rape and sexual assault. Just because they don't include something under the definition of "rape" doesn't mean that it was excluded from these calculations. While its not always 100% clear what is included, the questionaire does clearly include coersion and unwanted sex, and the design of the survey is fairly personal so it wouldn't at all surprise me if many of the things that you are saying are not included were actually part of this survey and included under the "sexual assault" blanket, which is very broad. I did just notice that they actually do provide data on what percentage of "Rape and Sexual Assault" was "Completed Rape" though. 30%. So if you were looking at the "lifetime completed rape" statistics for all women who are alive today (assuming your average victimization rate remains constant), including future victimizations, the percentage would be about 5.5% (using the BJS definition of rape, not the expanded definition). However, even if you want to use the more full definition, a full 18% of their "Rape and Sexual Assault" numbers were verbal threats, which I personally would consider to be a completely separate crime. |
I totally understand your point about the 81 years, but my earlier counterpoint was that I had not actually adjusted the percentage to actual population. In fact, at the earliest cycle there were 70m people less already, though I factored the percentage for 160m. For instance, 1989 would have been 5.6% and 15 years prior would have been 6.4. Since the average can be assumed as 81, and we know that upper boundaries for life are near 95, that would indicate (loosely) that the majority of people would die between 65 and 100, with 30% being past that standard deviation, and the lower deviation being 15% dying between 50 (HA, I just made this loosely, but the actual number is in fact 15 years http://www.nber.org/papers/w14093).
So, about 30% die between 66 and 81, skewed closer to 81 than 66, and 15% die roughly from 51 to 66. That would mean that about 80% of people last until 66. This would take one cycle off of my figures, but again, seeing as how I didn't factor in population differences over the years, we would actually get:
4.7% + 5.6 + 6.4 + 7 = 4 cycles with 1 cycle being 15 years. Add to that the 6 years, which amounts to .33 extra cycles, which would add another 1.79%. I believe that moving forward, the number is closer to upper 200k per year which is off by about 100k during the period I used previously. This would affect the numbers by 1.5m. And with population increasing, we would probably be looking at 4% + 3.5% + 3.0% + 2.5%, so 13% or so for future lifetime totals if we start counting from now. 1 in 8 or so sounds about right to me, if I assume that the DOJ knows exactly how many go unreported (which I'm not confident of, but I'll accept).
Instead of 4% * 5.4 cycles
which would give us 21.7% + 1.79%.
We also need to factor in the 20% of deaths that happen before 66, but I'm not sure how to quantify it since the majority of rapes actually happen between 6-25. So people dying young doesn't much offset the number given that it's only 20% of deaths, with only about 5% happening in the highest risk ages. Just to be conservative, I will take off the entire 20% of our percentage which is:
23.5 *.8 = 18.8% I hope you can appreciate that this is extremely conservative in approach.
BUT....This number would factor in for deaths AND for population changes, even if conservatively. Still only 5% in total (18.8-21.4 <2.5%) to the number I found with about 4000 words less. Was the 4000 words worth a 2.4% difference given that the baseline being used was 20%? I'm not sure it was. I basically spent several hours finding a 1.2% change Conservative difference. Oh, I almost forgot that I was also using an incidence number from the lowest incidence numbers in time. Dude, I really think it's safely over 20%. Regardless of death or population or anything. In the future that will change. But I think it's very safe to say of all people currently alive, 20% is a workable number.
But anyway, regarding the second point, the methodological sheet mentioned the statements I made, which was published in 2006, so perhaps the questionairre has changed since then. I believe I copied and pasted the text in my big calculations post. But just in case, I'll repost it tomorrow, just to show good faith that I didn't make it up.
