Craigslist adult section => less dead hookers?

[Jolly_Penguin]

https://thinkprogress.org/craigslist-erotic-services-platform-3eab46092717/

The September 2017 study, authored by West Virginia University and Baylor University economics and information systems experts, analyzes rates of female homicides in various cities before and after Craigslist opened an erotic services section on its website. The authors found a shocking 17 percent decrease in homicides with female victims after Craigslist erotic services were introduced.

The data does not provide a single clear explanation as to why female homicide rates drop so steeply, Scott Cunningham, one of the paper’s authors, told ThinkProgress. It’s possible, for example, that when Craigslist opens erotic services ads, some women in abusive domestic situations decide to become sex workers, move out, and so escape violent homicide at the hands of their spouses or boyfriends.

The most likely explanation, though, Cunningham says, is that sex workers simply make up a huge percentage of female homicide victims. When sex workers are safer, female homicide rates drop significantly.

Not sure if this study was included in our last thread on this topic.

 

[Jimmy Higgins]

I’m having a hard time finding actual data supporting their conclusion. I can’t even find data supporting a drop in homicides.

 

[Jarhyn]

https://thinkprogress.org/craigslist-erotic-services-platform-3eab46092717/

The September 2017 study, authored by West Virginia University and Baylor University economics and information systems experts, analyzes rates of female homicides in various cities before and after Craigslist opened an erotic services section on its website. The authors found a shocking 17 percent decrease in homicides with female victims after Craigslist erotic services were introduced.

The data does not provide a single clear explanation as to why female homicide rates drop so steeply, Scott Cunningham, one of the paper’s authors, told ThinkProgress. It’s possible, for example, that when Craigslist opens erotic services ads, some women in abusive domestic situations decide to become sex workers, move out, and so escape violent homicide at the hands of their spouses or boyfriends.

The most likely explanation, though, Cunningham says, is that sex workers simply make up a huge percentage of female homicide victims. When sex workers are safer, female homicide rates drop significantly.

Not sure if this study was included in our last thread on this topic.

Well, in the spirit of my response to Underseer's last thread, I fully expect that by Page 2, we have exactly the people I would suspect bloviating about how this is really a good thing because the word of Femgod states in chapter 1 verse 5 of the book of Pink that never should the government do anything that "promotes" prostitution even if it leads to fewer dead women. Or something.

 

[ronburgundy]

I’m having a hard time finding actual data supporting their conclusion. I can’t even find data supporting a drop in homicides.

Page 42 of the linked research paper shows the data in the most interpretable format. The zero line on the Y axis refers to there being zero difference in the rates between cities with craigslist erotic services and those without. The gray bars around each point is the measurement error. There is no difference between cites and no change over time during the 50 months leading up to some cities adding the erotic services. Starting at 10 months after some cities added erotic services to craigslist, there was the first significant difference where homocide rates were lower compared to the control cities. The difference got bigger and bigger in a rather linear fashion every 10 month period until it leveled off at about 70 months. The same is true of rape rates. Also, they used multi-level modeling, so they controlled for city effects to ensure that it wasn't just some peculiarity of a few of the cities in either group that lead to the overall difference.

They also do analyses showing that when comparing these same cities on male homicides, female homicides by acquaintances (e.g., boyfriends and husbands rather than "johns" and stranger rapists) there is no difference or significant change in the difference over time. That is important, because it rules out most of the alternative explanations which are already limited to things that happened to change in the one group of cities but not the others at the same time as the change in "erotic services".

 

[Jimmy Higgins]

It feels like data fiddling. They chart between 2002 and 2010. The murder rate in the study drops in 2006, but the murder rate in the US also dropped as a whole at the same point.

The murder rate plateaued during W’s first term. And the study doesn’t predate the plateau, so it seems impossible to derive any usefulness from this study as it is impossible to determine whether this data shows causation from Craigslist or was just the general trend over the murder rate dropping again.

What draws a flag for me is that there is no straight up data, ie no inclusion of murders in a city before and after. They just put all the numbers into some analysis and come out with their results.

 

[steve_bank]

There has also been a cooer elation between teen sexual abuse and rape with the rise of video games and misogynistic mystic, all dem hos and bitches.

One video game had points for rape.

It is probably social media in general. What is the correlations between rape and social media?

A guy in his seventies in my facility actually thought young women were attracted to him on SoCal media. He lost money until he wised up. Social media as riddled with scams, crime, and abuse.

I am shocked, shocked I tell you that there is crime related to social media...

 

[ZiprHead]

I thought CL banished the adult advertising late last year.

 

[Don2 (Don1 Revised)]

I thought CL banished the adult advertising late last year.

Somewhere around there, yes.

 

[Loren Pechtel]

Found what looks like a draft of the study:

https://cear.gsu.edu/files/gravity_...ne-Erotic-Services-Advertising-and-Murder.pdf

 

[Jimmy Higgins]

Found what looks like a draft of the study:

https://cear.gsu.edu/files/gravity_...ne-Erotic-Services-Advertising-and-Murder.pdf

The study is linked in the OP article.

 

[ronburgundy]

It feels like data fiddling.

No, it's use of proper statistical controls and comparison groups. They seem to be following textbook scientific methods for proper methods of a quasi-experimental multiple time-series design, which is what is required to test a hypothesis about the pre-post impact of a specific event in a naturalistic context. Their methods are about the best one can do to test a causal theory outside of completely randomized lab experiments.

They chart between 2002 and 2010. The murder rate in the study drops in 2006, but the murder rate in the US also dropped as a whole at the same point. The murder rate plateaued during W’s first term. And the study doesn’t predate the plateau, so it seems impossible to derive any usefulness from this study as it is impossible to determine whether this data shows causation from Craigslist or was just the general trend over the murder rate dropping again.

They actually examine homicide rates for 13 years, which for every city includes several years prior to each city implemented "erotic services" (or not) and several years after the implementation. So, nothing that happens in any single year is relevant. The analyses show that every 10 month range years beyond the first 10 months after implementation had significantly lower rates of female homicides by non-acquaintances than every 10-month range in the 4-5 years prior to the implementation. And since different cities implemented the "erotic services" in different years and the specific year did not change the effect, then nothing that happened with national level stats in a given year can account for the results. And the fact that there is no corresponding drop in male homicides or female homicides by acquaintances further rule out anything that happening in these specific cities that would lower homicide rates in general.

What draws a flag for me is that there is no straight up data, ie no inclusion of murders in a city before and after. They just put all the numbers into some analysis and come out with their results.

That isn't a red flag, it's a sign that they actually understand good science and know what the lay public does not, which is that such raw numbers have zero scientific meaning and don't even inform us about correlations and covariance. The only thing that provides any information are relative comparisons. that is basic intro to research methods stuff. You won't find that kind of raw data in any scientific article. What your asking for is equal to expecting a drug study to show you all the individual medical measurements for each individual in the study.

 

[Don2 (Don1 Revised)]

Are the numbers of prostitutes, including casual prostitutes, also going up? I think it might be expected.

Are raw numbers, not ratios, going down? Then maybe this is good.

What about since craiglist sex went away? Did ratios stay below their high point or increase?

 

[steve_bank]

Correlation is not necessarily causation.

Craig's List would have been only a small part of the issue. I can google 'seattle escorts' and I'll get beaucoup hits. Go to a site, pick a women, submit a credit card for validation, and you are on your way. There are plenty of apparent independents on social media. Full body massage is a typical offer.

Late night TV ads, call this number for girls looking for dares.

I expect street walking hookers are largely a thing of the past.

 

[Jolly_Penguin]

Correlation is not necessarily causation.

True and always important to keep in mind. But it does give some evidence.

Craig's List would have been only a small part of the issue. I can google 'seattle escorts' and I'll get beaucoup hits. Go to a site, pick a women, submit a credit card for validation, and you are on your way. There are plenty of apparent independents on social media. Full body massage is a typical offer.

That's just from your side of the fence though, and most of those other options are highly organize and require money. A typical entry level prostitute, working on her own or with a friend or two, isn't going to have a high profile escort serve that can process credit cards or be savy with social media. If you read the linked to study, or speak with a few who work in the industry, you will realize that sits like Craigslist or Backpage, etc that offer cheap or free adds is a convenient way for them to get off the street and get some semblance of screening, and they can do it themselves instead of rely on an "agent"/pimp who will get them work but also may abuse them. Many, and probably the vast majority of women advertising on sites like Craignslist/Backpage were working without "management". It isn't great screening still, but it may make the difference between a dead and live prostitute in many cases.

I expect street walking hookers are largely a thing of the past.

They aren't as common as they used to be, but they are still plentiful. You need only go on a drive in certain parts of town and you can see it for yourself.

And also, say the study has found a truth, and that these sites lower the frequency of murder of women. Could we expect that to change any minds on if it should be legal or not? I actually think few minds would change, even if they knew this to be true.

 

[ronburgundy]

Correlation is not necessarily causation.

But that overly simplistic and highly overused mantra is not a reasoned argument.

This is not a simple correlation (see my prior posts). It is very complex multiple time-series design with proper comparisons and control variables that greatly lower the probability of most alternative explanations. The most plausible and parsimonious interpretation of their data seems to be a causal effect of those "erotic services" on rape and homicide rates. Note that even with a double-blind randomize experiment, all that is actually observed is covariance of events where the context makes alternative explanations implausible. Causation cannot be observed. It is always inferred from covariance where there are factors that limit the plausibility of other sources for that covariance. Their design is second only to a fully randomized lab experiment in its ability to support causal inferences.

Craig's List would have been only a small part of the issue. I can google 'seattle escorts' and I'll get beaucoup hits. Go to a site, pick a women, submit a credit card for validation, and you are on your way. There are plenty of apparent independents on social media. Full body massage is a typical offer.

Late night TV ads, call this number for girls looking for dares.

So, your rejecting their conclusions that have substantial empirical support in favor on your own conclusions based on pure speculation?
Do you have any empirical basis to conclude that both sex workers and those looking for them were using use late night ads and (usually pricey) "escort services" in the same way and to the same extent that they used craigslist back in 1998 to 2009?
Jimmy gave some good reasons to doubt your assumptions from the perspective of the prostitues, and just off the top of my head, here are some reasons from the perspective of the customers. It is unlikely that most "johns" are fine with using a credit card for their purchase. Second, many probably do not want to going through a formal company where others besides the single person they will be getting serviced by are mediating the process and "taking their order" so to speak. Although not always the case, craigslist has more of a 1-on-1 private interaction type of vibe to it. The very fact that craigslist is so incredibly popular and widely used for every type of interpersonal or business exchange suggest that other formats are not providing the same thing in the same way that draws users to craigslist. Besides, even if craigslist is only one way to achieve a similar result, that would just mean that the effect they observed is only a portion of the complete reduction in homicides and rapes of sex workers that legal use of online interfaces could provide.

Their evidence shows a particular pattern of non-linear change where the inflection point occurs shortly following a particular event that is independent of time/year (as thus all other variables time to time/year), because the event occurred at different times in different areas and only shows the pattern of covariance with female homicides and rapes in that same area, not other areas and not with similar crimes (male homicide and female homicide by acquaintance).

The theory sex workers ability to use these mediums rather than more dangerous alternatives reduced their rapes and homicides explains all of this complex pattern of data. Unless you can present an equally plausible alternative account that predicts that same complex pattern of data, it is irrational not to accept that the study provides evidence supporting a causal reduction of this crimes against women due to the availability of erotic services on craigslist.

Maybe I'm being a bit hard on you, but I get really tired of the "correlation does not imply causation" canard that is so often wrongly applied to complex multivariate designs, which is usually done to blindly dismiss evidence that supports a conclusion they don't like. I'm all in favor of being skeptical and critical of research on any topic. It's what I do best and do it for a living and a hobby. But you have to at least deal with the actual data presented in the context of the specific methods used, and be able to come up with specific flaws and alternative explanations for the data.

 

[Jolly_Penguin]

Maybe I'm being a bit hard on you, but I get really tired of the "correlation does not imply causation" canard that is so often wrongly applied to complex multivariate designs

Maybe the better mantra would be "Correlation is mere evidence and not conclusive"?

 

[steve_bank]

Too many variables to draw a conclusions that isolates craigslist as primary cause.

I have a statistics background and have done data correlations. I have experienced finding correlation not being causation more than once. I understand times series and techniques like auto regressive moving averages. ARMA.

There is an old book How To Lie With Statistics. You cab draw most any conclusion you want using statistic's applied to non numerical issues and it not being outright fraud.

 

[Jarhyn]

Too many variables to draw a conclusions that isolates craigslist as primary cause.

I have a statistics background and have done data correlations. I have experienced finding correlation not being causation more than once. I understand times series and techniques like auto regressive moving averages. ARMA.

There is an old book How To Lie With Statistics. You cab draw most any conclusion you want using statistic's applied to non numerical issues and it not being outright fraud.

And if you want to claim that is at play here, please provide the analysis that would support that skepticism. It's all good and fine to say "data fiddling can happen" and "correlation is not causation" but to actually impugn a piece of data, well, you have to actually impugn the data. So either trust, or verify. You don't get to ethically distrust without verifying.

 

[Jimmy Higgins]

Too many variables to draw a conclusions that isolates craigslist as primary cause.

I have a statistics background and have done data correlations. I have experienced finding correlation not being causation more than once. I understand times series and techniques like auto regressive moving averages. ARMA.

There is an old book How To Lie With Statistics. You cab draw most any conclusion you want using statistic's applied to non numerical issues and it not being outright fraud.

And if you want to claim that is at play here, please provide the analysis that would support that skepticism. It's all good and fine to say "data fiddling can happen" and "correlation is not causation" but to actually impugn a piece of data, well, you have to actually impugn the data. So either trust, or verify. You don't get to ethically distrust without verifying.

Steve is just making an observation that I as well made that the conclusion of the study seems to be based solely on fiddling with numbers. There is no raw data they present to help demonstrate their point. In 2006, there was a general decrease in the murder rate in the US after having plateau'd for a number of years. Take a look at this BJS report from 2013 (Report Page 4).

The straight up statistics show a decrease for male (~15%) and female (~11%) within the same period as the cited report for the OP. I'll take straight up stats over ones shoved through a statistical grinder.

It is possible that legalizing such things can lead to lower abuse and murder, I'm not hanging my hat anywhere with that argument. What I'm saying is that the murder rate dropped for male and female within the same range as reported for the report, which implies a generalized reduction in crime.

 

[Jarhyn]

Too many variables to draw a conclusions that isolates craigslist as primary cause.

I have a statistics background and have done data correlations. I have experienced finding correlation not being causation more than once. I understand times series and techniques like auto regressive moving averages. ARMA.

There is an old book How To Lie With Statistics. You cab draw most any conclusion you want using statistic's applied to non numerical issues and it not being outright fraud.

And if you want to claim that is at play here, please provide the analysis that would support that skepticism. It's all good and fine to say "data fiddling can happen" and "correlation is not causation" but to actually impugn a piece of data, well, you have to actually impugn the data. So either trust, or verify. You don't get to ethically distrust without verifying.

Steve is just making an observation that I as well made that the conclusion of the study seems to be based solely on fiddling with numbers. There is no raw data they present to help demonstrate their point. In 2006, there was a general decrease in the murder rate in the US after having plateau'd for a number of years. Take a look at this BJS report from 2013 (Report Page 4).

The straight up statistics show a decrease for male (~15%) and female (~11%) within the same period as the cited report for the OP. I'll take straight up stats over ones shoved through a statistical grinder.

It is possible that legalizing such things can lead to lower abuse and murder, I'm not hanging my hat anywhere with that argument. What I'm saying is that the murder rate dropped for male and female within the same range as reported for the report, which implies a generalized reduction in crime.

So, I'm going to go with Ron here and point out that you still have to validate this claim and you have not. You are potentially one of maybe five people on these whole forums capable of finding a real fault in the study if one exists, and framing the fault in a way others can see it. So either DO that, or quit shitting on the floor.