This was posted as a rebuttle to Caerons research on www.space.com under there FreeSpace forum, you have to register to view that forum though:
“Intersting, interesting...Well now...it appears that I’ve found the hole in the statistical argument used.
Basically, I took each separate grouping and calculated the 95% confidence intervals on the sample of each group. Here are the statistics on it:
Hetero
One Sample, Mean: t
Details Data Input Results
- Confidence Interval: Two sided Confidence Level % 95.0% Sample Mean 3.86
- Numerical Variable Sample Size, n 11 Standard Error 0.567
- Source: Summary Measures Sample Mean, XBar 3.86 t +/- 2.228
Sample SD, s 1.88 df 10
95% Interval: m
- From: 2.597
- To: 5.123
Homo
One Sample, Mean: t
Details Data Input Results
- Confidence Interval: Two sided Confidence Level % 95.0% Sample Mean 5.03
- Numerical Variable Sample Size, n 11 Standard Error 0.289
- Source: Summary Measures Sample Mean, XBar 5.03 t +/- 2.228
Sample SD, s 0.96 df 10
95% Interval: m
- From: 4.385
- To: 5.675
Basically what I found is that there is a 95% probability that the true mean abuses of Straight abusers is between 2.597 and 5.123.
The 95% confidence interval of Homosexual abusers is between 4.385 and 5.675.
Now, taking the raw statistics at face value where “Homosexual” abusers abuse the same sex and “Straight” abusers abuse the opposite sex (which I don’t agree with that assumption to begin with), this means that the argument that Homosexuals abuse more than Heteros is bogus if using the data set provided.
Additionally, there is a very large standard deviation of the straight data set, which implies that there either is not enough data in the set, or that the set has problems.
I love busting arguments...LMAO...”
Can someone qualified with statistics please analyse this rebuttle and let me know if it’s correct?
Thanks,
Facts are mean-spirited
The primary fault with your "logical" statistical model/argument is that you presume sexual orientation to be a scientifically objectively proved factor. I read the study to be based upon the sexual act alone regardless of what sexual orientation the reported predators may claim to "feel" they are... In essence, male on male sex is by default a homosexual act AND implies by default an objectively proved state of homosexual orientation at a minimum while the predatory act was perpetrated...
SO -assuming ALL cases reported are normally distributed (hence proportionally equal) then one would expect to see an equal number of male on -> female, female on -> male, female on -> female, and male on -> male sexual predations. This study correctly points out the obvious, which is (when normalized based upon the accepted "sexual orientation" distribution within society), that homosexual males present the greatest predation danger to young boys... In summary it is dangerous and quite stupid to consider homosexual coaches, babysitters, boy scout leaders, etcetera to be on par with heterosexual boy scout leaders etcetera etcetera....
I can’t tell what the data is a score of. The t-test is based on the assumptions of the binomial distribution and the normal curve, and assumes continuous variables. If you are dealing with a binary score (eg, “molest/didn’t molest) then you are using the wrong test. You need a nonparametric measure. Also, your sample size is really too small for a t-test. Try a chi square.