Fun Stuff > CHATTER
Things you know about AIDS
Verergoca:
Katie, you develop crushes on everyone, one more isnt going to hurt!
Also, while yes, the talk should, like most talks, should be taken with a grain of salt, but, given that the whole concept of TED (which is quite new to me, and i have to say, i lurvse it), is to make people think, to reach people that normally wouldnt raise an eyebrow.
To do that, yes, assumptions and leaps in logic must be taken. The thing is, if you dont do that, your getting nowhere anyway (and no publishing deal either!)
Also, i likses the James Watson talk moar, but that might be just me :)
dennis:
--- Quote from: pilsner on 21 Mar 2008, 09:07 ---To whit: HIV causes AIDS. HIV is transmitted sexually, through shared needles, blood transfusion, or during pregnancy/birth. Obviously an ABC (abstinence, faithfulness, condoms) program would cut down on HIV transmission to the extent that in the context of the local customs and culture it actually convinced people to have sex with fewer new partners. Her point of the effectiveness of this program in one country given the impact of coffee imports is somewhat mooted by the fact that, for instance, Togo may be set up with a government, clergy and local custom that makes a "faithfulness/condom" drive attractive to the locals whereas Ghana may not. Furthermore she consistently fails to mention what factors she excluded to derive causation from correlation.
--- End quote ---
These are good points, but she really brought up those arguments to show that the conclusion that ABC programs are effective isn't necessarily correct because the apparent success of the program may be due to other factors, such as coffee prices. To wit: that no one's proved that the drop in AIDS rate in Uganda was caused by the ABC program rather than correlated in the first place.
--- Quote ---The correlation of coffee prices to new HIV transmissions (which was far from a perfect match, observe the bump at the beginning in coffee prices that wasn't matched in HIV transmissions) might be explained by something other than causation. For instance, what if while coffee prices are high, there's more money for HIV testing, and more new cases are identified. Furthermore, her analysis omitted a major step -- she assumes that higher coffee prices means more trade, means more people from outside the country spreading HIV inside the country.
--- End quote ---
A perfect match in the coffee price curve and the AIDS rate curve isn't necessary as long as the changes are significantly interrelated. There are statistical methods to determine the quality of such a comparison match, and I don't think it's farfetched to assume that she's run the data and come up positive.
The bump you note is small, and also there's a capacitance to commodity price fluctuations affecting other systems that will smooth out a response curve. It might be too small to have a noticeable direct effect and may only show up as a higher-order response. Imagine, by analogy, that you're driving at a steady 55 mph and blip the throttle. You might see your tach jump and hear the engine rev up a little, but your speed won't change significantly--what's happened is that the drivetrain of your engine has smoothed out that bump in power, but that doesn't mean your speed is independent of engine output.
--- Quote --- That's a major leap of logic -- why should we assume that as coffee prices go up, the total quantity of coffee traded goes up? It's just as plausible that as coffee goes down, more coffee must be exported to make up for the shortfall in profits per unit of coffee. Without verifying which is the case, the causation link is nothing more than an unsubstantiated assumption.
--- End quote ---
No one is making that assumption. Exports are measured in value, not quantity since comparing a thousand coffee beans to a thousand sheets of paper isn't economically useful. She previously noted a positive correlation between amount of exports and rate of AIDS. As coffee prices go down, exports go down, AIDS rate goes down.
--- Quote ---I could go on but it would be rather pointless -- the presenter might have anticipated and explained away everything I'm saying and simply not had time to raise these issues in a 16 minute talk. My point is that with pop presentations which fail to approach a complex issue with anything resembling rigor, it pays to exercise a healthy level of skepticism.
--- End quote ---
That is assumed of such talks. She *is* an acedemic and is working on a paper documenting her research. However, this is beside her point that AIDS policy in sub-Saharan Africa should be revised.
I wouldn't consider this a "pop presentation", but as far as rigor is concerned, she was working off of data that everybody thought they already knew. If you think about it, she was actually promoting a rigorour reexamination of the data. Also, TED attendees aren't really the type to *not* be skeptical.
0bsessions:
--- Quote from: jhocking on 21 Mar 2008, 09:16 ---I went to highschool with her. We had a mouth class together.
--- End quote ---
This is how I read this, initially.
Clearly, I've not had an adequate amount of sleep.
Cam:
I believe this has already been addressed in depth:
http://forums.questionablecontent.net/index.php/topic,17409.0.html
pilsner:
--- Quote from: dennis on 24 Mar 2008, 04:18 ---These are good points, but she really brought up those arguments to show that the conclusion that ABC programs are effective isn't necessarily correct because the apparent success of the program may be due to other factors, such as coffee prices. To wit: that no one's proved that the drop in AIDS rate in Uganda was caused by the ABC program rather than correlated in the first place.
--- End quote ---
No, at 14:30 of the talk Oster states that the ABC campaign may have been 25-50% less effective than previously thought, ergo the 15 billion dollars in aid might be better spent other than on an education campaign. She never suggests that the ABC campaign was ineffective, just less effective than previously thought. Keeping in mind that, according to Oster, Uganda's infected rate dropped from 19% to 6%, this is a significant distinction. In any event, my point remains that even if we give Oster the benefit of the doubt and assume that only half of the 13% drop in infected is due to ABC, the impact of this analysis is lessened by the dependence of an education campaign on the culture in which its being implemented.
--- Quote ---A perfect match in the coffee price curve and the AIDS rate curve isn't necessary as long as the changes are significantly interrelated. There are statistical methods to determine the quality of such a comparison match, and I don't think it's farfetched to assume that she's run the data and come up positive.
The bump you note is small, and also there's a capacitance to commodity price fluctuations affecting other systems that will smooth out a response curve. It might be too small to have a noticeable direct effect and may only show up as a higher-order response. Imagine, by analogy, that you're driving at a steady 55 mph and blip the throttle. You might see your tach jump and hear the engine rev up a little, but your speed won't change significantly--what's happened is that the drivetrain of your engine has smoothed out that bump in power, but that doesn't mean your speed is independent of engine output.
--- End quote ---
I'm sure Oster's regression analysis came up positive. I'm also sure that you're aware how unreliable regression analyses are. You ever hear the old adage about two economists and three opinions? In any event, I'm merely noting the discrepancy to point out that one would have to assume that the linkage between export value and new infections only works when export value rises or drops precipitously for this linkage to work at all. This misses my main point, which you failed to engage, which is that the correlation between export value and new infections is better explained by the rising availability of testing than it is by a greater amount of export motivated mobility among Ugandans. The entire assumption that greater export value means more strangers means more spread of HIV bothers me something fierce -- a precipitous drop in export value might cause farmers to be uprooted and forced to migrate to urban centers which would entail (a) more exposure to HIV and (b) less ability to test for new infections because of all these suddenly appearing stranger ruralites. The correlation is interesting, granted, but the conclusions Oster is drawing from them are unwarranted and highly contingent on the economic and social situation in Uganda.
--- Quote ---No one is making that assumption. Exports are measured in value, not quantity since comparing a thousand coffee beans to a thousand sheets of paper isn't economically useful. She previously noted a positive correlation between amount of exports and rate of AIDS. As coffee prices go down, exports go down, AIDS rate goes down.
--- End quote ---
Here you are demonstrably wrong. Both the cumulative value and the quantity by volume or mass of exports are of interest to economists. Oster seems to agree with me given that she discusses both correlation with value and volume in table 12 of her working paper. Rising export value, which is what is correlated to new infections on the graph at 14:18 of the talk, is not what determines the amount of strangers circulating in Ugandan ports and towns spreading and catching HIV and thereby driving up the number of new infections. No, what would drive up new infections is an increase in the volume of coffee trafficked, which would require a larger number of people to be present loading, shipping, and accounting for the coffee, which in turn would bring more strangers to Ugandan ports and towns, etc. etc. Yet Oster draws the linkage between volume of exports in her talk and presents a graph comparing export value -- this is completely inadequate.
So why, we might ask, did Oster present a graph comparing value instead of volume? Table 12 of her working paper shows that according to her calculations, the value correlation explains about 60% of the drop in new infections whereas the volume correlation explains only 33% of the drop. Of course, mentioning this in the presentation would have lessened the impact....
--- Quote ---That is assumed of such talks. She *is* an acedemic and is working on a paper documenting her research. However, this is beside her point that AIDS policy in sub-Saharan Africa should be revised.
I wouldn't consider this a "pop presentation", but as far as rigor is concerned, she was working off of data that everybody thought they already knew. If you think about it, she was actually promoting a rigorour reexamination of the data. Also, TED attendees aren't really the type to *not* be skeptical.
--- End quote ---
I'm sure you know that I wasn't addressing my comments to TED participants, given that I posted them in the Questionable Content comic strip forums. Oster is an assistant professor at the very beginning of her career who stood up in front of a group largely composed of people who were not academics and who did not specialize in her field, and made a series of very speculative inferences that she would never have made with the same confidence at an economics conference, and which only present the most exaggerated and worst supported data from her accompanying working paper.
Yeah, I think “pop presentation” is dead on.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version