Authored by: asegar on Wednesday, June 17 2009 @ 01:50 PM GMT+4

The 1.37x the national average statistic for forcible rape is based on 6 cases. Although there may be other data indicating that Brattleboro has an above average incidence of forcible rape, this ratio by itself is not significantly different from the average national rate because of the small sample size.

---
-Adrian-

Authored by: Rolf on Wednesday, June 17 2009 @ 02:09 PM GMT+4

Is the sample size really equal to the number of rapes?

Does 6 incidents really mean, n = 6 ?

Rolf

---
Dreams Trump Video

Authored by: asegar on Wednesday, June 17 2009 @ 03:19 PM GMT+4

Yes it does - there were 6 offenses reported. You can download a spreadsheet from the FBI containing this 2003 information for every U.S. city with population > 10,000 here:

http://www.fbi.gov/filelink.html?file=/ucr/cius_03/xl/03tbl08.xls

---
-Adrian-

Authored by: Rolf on Wednesday, June 17 2009 @ 05:10 PM GMT+4

Actually, I think that n

is not the number of incidents of rape.

We are measuring incidents of rape.

n, in this case would be the number of years that were recorded.


Each year of information is a sampling of Brattleboro, I believe.

The amount that is measured is not the size of the sample, it is the size of the measurement.

Using another example, if we were simply measuring the number of calico cats in Brattleboro, and found that we had 125, we wouldn't say that the sample size was 125.

If we were going to study the habits of those calico cats, then the sample size WOULD be 125.

Or, if we measured the heights of men in Brattleboro, and found that the average was 58 inches, we wouldn't say that the sample size was 58.
The sample size would be the number of men who whose height we measured.

Still, you are right that when the number of incidents is less than 10, single incidents have a "disproportionate" impact on the way we see the data.

Going from 1 to 2 increases the incidents of rape by 100% for example.


1 rape is terrible, but the question is whether we have "more" than other communities, or more than the national average.

One possibility is that the data is from a single year, and if my conception of the sample unit is correct, than we have zero degrees of freedom, i.e. no way to make comparisons between averages for cities.

I can't find any indication of how many years the data was collected for, at least not yet.





---
Dreams Trump Video

Authored by: Lise on Wednesday, June 17 2009 @ 05:46 PM GMT+4

I have a request in to the VT Dept of Public Safety to see if we can get more recent stats, but reporting methods changed around 2003 and data for Brattleboro seems impossible to find after the 2002 crime year. If I get more recent stats, I'll post them here, as I found the rape numbers earlier in the decade to be a bit staggering too.

Wanted to mention also that per the BPD, those numbers represent reports of crime responded to and not necessarily convictions. That said, the 12 reported rapes in 2002 seems like a lot. Whether these were forcible rapes or not (is there any other kind?) is another question too. They're just listed as rapes on the state report.

Here's a link to the DPS page -- look for Crime Reports:

http://www.dps.state.vt.us/index.html

Authored by: Rolf on Thursday, June 18 2009 @ 06:55 AM GMT+4

Thanks Lise, for putting in that request for information.

---
Dreams Trump Video

Authored by: annikee on Wednesday, June 17 2009 @ 07:33 PM GMT+4

How are these things counted, as in no. of rapes? By conviction?

---
Down with Goldstein!

Authored by: asegar on Wednesday, June 17 2009 @ 10:12 PM GMT+4

We are comparing the incidence of forcible rape offenses in Brattleboro with the incidence of forcible rape offenses nationally. The latter, according to the FBI, was 3.64 per 10,000 people in 2003.

The FBI spreadsheet I referenced clearly implies that in 2003, there were 6 "offenses known to law enforcement" of the type "forcible rape" in Brattleboro, a town with a population of 12,035.

You can think of Brattleboro as a sample with size 12,035 of the population of the U.S.

Assuming that the incident of forcible rape is normally distributed, the 6 "offenses known to law enforcement" are not statistically significantly different from that occurring nationally. This is because the 4.4 incidents one would predict from the national U.S. incidence for a sample of size 12,035 is not statistically different from the 6 incidents reported.

---
-Adrian-

Authored by: Rolf on Friday, June 19 2009 @ 09:53 AM GMT+4

" You can think of Brattleboro as a sample with size 12,035 of the population of the U.S."

Well, I don't think that is correct either...

I admit to being rustier than I once was on stats, even rusty compared to what I knew a mere year ago.

But I think that the population size, 12,035 people in Brattleboro, is merely used to figure out the per capita incidence of rapes, and then, that is used for comparison sakes,( as you mentioned) to come up with an expected value per 10,000 people.

But the number of people in Brattleboro is not a "sample" of the US.

If it was a "sample" of the US population, we would be using just the Brattleboro data to make a statement about the national population. That's what a sample is, by definition.

The Brattleboro data is useful for telling us (maybe) something about Brattleboro. It is a "sub-sample".

So, the sample size is neither 6, nor 12,035.

I am still thinking that we have a single data point on this spreadsheet, from a single year. If we wanted an average rate for Brattleboro, we would need at least three years, and then we could start talking about whether our average is different from the national average over the same span of years. Right now, we have a single data point from a single year. I am thinking that our sample size is basically, 1.

I think you are right with only 6 incidents, it may not be significantly different from the national average, even though it is listed elsewhere as 1.37 times larger than the national average.

It still strikes me as interesting that our rate for that one year was listed as being 1.37 times as large as the national average, (created I assume by summing all towns single data points and dividing by the total number of towns)

AND other towns in Windham County of comparable size were listed as being .29 and .26 and .48 of the national average.

With a single year, and such low numbers, it is impossible to tell if this there is any significant difference. Having multiple years of data would let us know if there appeared to be a significant trend, or difference.

But it does seem like (to me) something worth pursuing.

Rolf




---
Dreams Trump Video

Authored by: asegar on Friday, June 19 2009 @ 10:38 PM GMT+4

It was a long time ago, but I used to teach statistics at the college level. I'll give this one last try.

The FBI statistics are for one year, 2003.

Yes, if we had more years available we would have a better basis for learning whether the reported incidence of forcible rape in Brattleboro is statistically different from the national average.

Even with one year's data, however, it's possible to say something about whether the incidence of forcible rape in Brattleboro is statistically different from the national average in that year, and that's what I've been writing about.

If you're still confused or have questions, give me a call.

---
-Adrian-

Authored by: Rolf on Saturday, June 20 2009 @ 07:10 AM GMT+4

Adrian,

Well that makes two of us who have taught stats to college students, perhaps in our distant past, and two of us who are rusty.

My rustiness is evident in that it took me this long to figure out that the sample size is equal to the number of towns / cities with populations of 10,000 or more. I thought the sample size was 1, as we only have one data point for Brattleoboro for the year 2003. Wrong.

You originally posted that the sample size was 6, and then changed that to say that the sample size was equal to the population size of Brattleboro.

So far, if anyone else in Brattleboro is reading this little series of posts, we both have added to the confusion. My guess is someone out there with a sharper grasp on stats, (or just more common sense than either of us) long ago recognized that the sample size for the national average is the number of towns that have a population of 10,000 or more.

Instead of my calling you, and having you "help me with my confusion" or (or perhaps, you calling me, and getting the same help hee hee) let's both admit that we goofed ?

Maybe?

Rolf



Your rustiness is evident in that you first thought

---
Dreams Trump Video

Authored by: Rolf on Saturday, June 20 2009 @ 07:17 AM GMT+4

PS,

Yes you are right there is way to tell whether a single data point for the year of 2003 is different from what would be expected (i.e.) the average.

That would be test we could perform, if we wanted to, as we have the data points for all the other towns.


We could calculate the standard deviation for the average, and then see if the data point for Brattleboro fell within an expected range of variation.

I agree it probably is not different from what would be expected from the average. But I haven't done that test, so I don't know.

Have you?

Sincerely,

Rolf





---
Dreams Trump Video

Authored by: JoanneN on Wednesday, June 17 2009 @ 04:08 PM GMT+4

Thanks for posting this information. I am surprised by the rape though. Than again does forcible rape include date rape? Both are incredibly wrong of course but my question is should you be worried about getting raped by someone while walking down the street. From my understanding as a whole most rapes are from people you know. At least that is what we were told when I volunteered at a woman's crisis center .

---
People who fight fire with fire usually end up
with ashes.
~Abigail Van Buren

To love a person is to learn the song that is in
their heart and to
si

Authored by: imhennessy on Wednesday, June 17 2009 @ 09:43 PM GMT+4

My own, uninformed, take on the issue of incidents of rape is that it seems to line up very well with a quick scan of the police log. Not that there are a lot of crimes which seem similar to rape reported, but the great preponderance of notes in the police log are DUIs or accidents involving some form of intoxication. It seems to me that problems with drugs and alcohol are most visible when they intersect cars, and that they are among the factors most likely to be connected to rape.

My own personal theory of rape (please do not take this as excusing rapists) is that, other than serial rapists, the majority seem to be crimes of immense stupidity. People, for a variety of reasons, are, or become, incapable of picking up on the fact that they are doing something terrible. I suspect that there is an escalation which begins with poor judgement and includes misinterpretation and violent reactions to disappointment or the idea that a mistake has been made. I think people, especially when drugs or alcohol are involved, react blindly to the guilt and shame of being exposed as having been in the wrong.

I'm not sure I communicated clearly my view. And I want to emphasize that I don't think this possible explanation excuses rape or assault, nor does it implicate the victim. I just think that most people, most of the time, do not consciously choose to hurt others. That does not diminish the hurt that is caused.

ivan