I'd like to point out that the methods employed here to survey the wages of prostitutes leave plenty of confounding variables in play. "Analyzing wage informati

I'd like to point out that the methods employed here to survey the wages of prostitutes leave plenty of confounding variables in play. "Analyzing wage information on more than 40,000 escorts in the U.S. and Canada collected from a web site, we find strong support for EK." 1. Remember from statistics class what the problem with conducting polls via random telephone calls to thousands of different homes? Not everybody has a phone, and the reasons are nonrandom. Well, not every prostitute has the internet...or advertises on the internet...etc. 2. Just "a" website? Only one? Well, this isn't _necessarily_ a problem-- so long as the conductors of the study can provide reason to believe the population of prostitutes advertised on this particular website is demographically no different than the total population of (be it "high-end" or "all") prostitutes into whom they attempt to gain insight. Yeah, good luck with that. 3. Are we assuming that the listed hourly rates of the prostitutes on the website exactly correspond with what ends up being charged? This can swing results in several different directions: perhaps some bartering occurs on occasion, maybe special deals can be had ("Two-for-one, ehh? Oh, but only for the 26- to 30-year-olds? Well damn."), maybe they round off the time spent "on the job". In addition, the _averaged_ 'hourly rate' a prostitute charges often decreases as the total time billed for the evening increases ($50 for 15 minutes vs. $100 for 45 minutes, for example), which would make it difficult to guess at true wages from one number alone ($/hr), given that the math will inevitably be done differently from prostitute to prostitute (as well as from pimp to pimp, if one is involved). And let's not even start down the road of analyzing the "a la carte menu"...just know that each girl could have a different menu with different prices, or even no menu at all. Again, good luck to the study conductors in dealing with all these confounding variables. 4. Etc. Point: Though I haven't read anything more than the abstract, I nonetheless highly doubt that this study was done in the proper statistical manner. "Good enough" never really is. If there are confounding variables, they need to be eliminated. Otherwise, no conclusions can be reliably drawn.