I have only one data point, myself. No forced sex. On the rare occasion that I turn my wife down, she takes it gracefully.
[Andrew Dice CLay] Not nearly enough [Andrew Dice Clay]
Yes, there’s a spectrum. The law typically even acknowledges that and provides that a guy threatening “If you won’t have sex with me I’m going to leave you and file for divorce” isn’t typically considered rape or attempted rape, but saying “If you won’t have sex with me, I will hit you with this lead pipe and then go rape your sister” is.
And how do you factor in those that you would not be able to reach with this poll?
You’d have to start by getting some statistics for certain sample populations. “Among college students in California…” or something like that. You could partly include the unreported populations by asking about past events. For example, Ariel Castro’s victims couldn’t have been polled during their period of imprisonment, but they could be counted in retrospective data. You could also attempt to adjust the record for known abuse of victims who were killed by their abusers.
You’re right that this sample may not be applicable to the entire population, but sociology is an entire branch of science built around trying to draw generalized conclusions from limited, often self-reported data. We could at least set some upper and lower bounds for the number.
In thinking about the OP’s question, I suspect the numbers would be higher than many of us suspect. We know that something like 20% of women are sexually assaulted/abused at some time in their life. Many of these people are raped on a daily or weekly basis for years, by someone like a family member or spouse.
It’s probably safe to say that we’re nowhere near 30%. We’re no ducks.
The question isn’t “What percentage of U.S. sex is forced and involuntary?” How do you propose on either polling or gathering accurate statistics on such a sensitive subject in the Middle East or Asia?
OK… “Among college students in Kazakhstan…”
I mean, I just outlined for your how these things work. I’m not pretending we would talk to everyone or that we would get six decimal places of accuracy. Whatever we publish, someone will quibble with our methods and definitions. Still, I’m pretty sure we could get to +/-5% without even breaking a sweat, if someone was willing to fund such a study.
I answered that question later in the same paragraph you quoted. I said “Put 1,000 people into a room, make sure you have a wide range of ages, proportional representation of genders, ethnic groups, and economic status.” That’s how you make sure that your poll isn’t a biased sample of only the kinds of people who are easily reachable and fails to represent people who are harder to reach. I know it may sound counterintuitive but truly random samples are virtually impossible so the next best thing is a sample which has been manipulated to resemble what a hypothetical random sample would look like.
Here’s an oversimplified example. Suppose you take a poll on the internet and you get 8,000 replies but 7,400 of them are from people under 30. You figure that a good sample should have half under 30 and half over thirty. Clearly the sample you got was biased with too many young people. So you compensate by throwing out all but 600 of the 7,400 under 30 results. Now you have 600 people under 30 and you have 600 people over 30, so you’ve created a representative sample.
Naturally, there are limits to this technique. You’ll never be able to include in your poll people who have a pathological fear of talking to pollsters, for example. And you’ll never be able to include catatonic paraplegics. Luckily, those people are a tiny tiny fraction of the population and you can get pretty good results without them.
Imagine making a huge pot of soup. You’ve got carrots, beans, tomatoes, and chicken. You want to serve a bowl which represents what the entire pot is like. If you stuck your ladle in and pulled out a cup with all carrots and no chicken, that would not be a representative sample. So what do you do? You stir the pot. You scoop out a cup. And then, if necessary, you say “hmm that bowl has too many beans” so you put some of the beans back into the pot.
Naturally, there’s limits to this technique. You’ll never be able to include people who have a pathological fear of talking to pollsters, for example. And you’ll never include catatonic quadriplegics. But, luckily, such people are a tiny fraction of the population and you can get pretty good results without them.
Drakes grab ducks by the neck, ripping out feathers as they pound away. How exactly are they evaluating what’s voluntary? Cracked is cracked.
If I were to put a number on it, 20% sounds about right. I don’t think a lot of people are accounting for the sheer prevalence of child abuse and unwanted marital sex in the the world. Or how involuntary most prostitution is (I’m not talking about the Bunny Ranch prostitutes here) But the naivete in the thread is somewhat refreshing, actually.
Do you do that before or after you define your terms, such as “forced”? How, exactly, to you get your randomized “sample”? Telephone survey?
- Would a thousand people be enough to represent the wide range of ages, proportional representation of genders, ethnic groups, and economic statuses of the entire world?
- If so, would these thousand people be accessible for polling?
Am I being whooshed? What aren’t you understanding?
Let’s try looking up human trafficking on Wikipedia, which says:
Amazing, isn’t it? 2.5 million people who all fit your definition of “not accessible for polling” and yet we know the number, source, destination and dollar value withing a reasonable set of error limits.
The Wikipedia article could probably give you the number of these trafficked for sexual purposes. Figure the number of tricks per day, multiply that out.
Short answer, Yes.
The bigger the sample size, the lower the margin of error. However, it doesn’t pay to keep making the sample bigger and bigger and bigger because beyond a certain point the uncertainties caused by irregularities (such as tiny changes in the way the questions are worded) become more significant than the random fluctuations due to small sample size. When you’re dealing with a universe of hundreds of millions or even billions, a sample of 1,000 is pretty optimal. That gets your sampling error down to about +/- 4% and you aren’t likely to get your irregularities any smaller than that no matter how hard you try.
Conversely, if you WANT to screw up the results by intentionally playing with the irregularities, then you can increase the gap between reality and what the poll results seem to be showing. But I’m assuming here that our goal is to actually find the truth, instead of just making stuff up so we can put it into a campaign ad or something like that. Tweaking the questions is a good way to do it. Another way is to use some subtle selection process which is not recorded anywhere in your methodology, like excluding people who speak with an accent. Using dirty tricks like these, you can get results which are 20 or even 30 percentage points away from the truth. And increasing the sample size won’t make it more accurate.
I am not a statistician but I do have an MS in Mathematics and I used to teach college-level Probability & Statistics.
For the record, I laughed my ass off at this.
[QUOTE=Ann Hedonia]
It’s one of my favorite articles ever and I always jump at any excuse to post it.
[/QUOTE]
And this.
If only more rape threads were full of such hilarity.
My contribution: I once witnessed a turtle gang rape at a zoo. It was… awkward. And just as slow as you might expect.
2.5 million is a meaninglessly tiny number of people. Globally, it is one person out of 3,000, which would be represented as zero in a sample of 1,000. Certainly not 1/3, which was the starting point of the OP of this thread.
The UN has observed that there is a phenomenon, and one of its agencies responsible for such things made a reasonable effort to quantify it. Without having any access to data, except a sprinkling of anecdotes, they extrapolated a number based on actual incidences combined with presumptions based on a demographic principle that all people are are average.
How exactly do you “figure the number of tricks per day”? The same way the UN figured their 2.5M?
As to the “marital” example how much is truly I do not want to have sex with this person, and how much is oh god alright I’m tired but I love you and you do a lot for me so I’ll service you?