Hume was speaking in very strict terms. I'm sure that he accepted causality as a useful myth in an everyday context. For example, I doubt anyone would claim that we should stop researching medicines because we cannot establish a strict causation between them and the ailments they purportedly assuage.vivaxardas said:
Wasn't that grregg's point? The skin cream results were made up and so the subjects had no prior knowledge of what to expect, while they presumably knew something about gun control already.GuyN said:The catch with that explanation is that the most numerate (about 10% of the subjects) had no trouble with the value-free "skin cream" results, but the same subjects stuck on the value-loaded "gun control" results.
We could easily test grregg's hypothesis. Simply give the subjects similar made-up data about a politically-neutral subject about which they have prior knowledge. For example, most people know that aspirin has positive effects on both high blood pressure and pain. What if we gave them bogus data that contradicted both these findings? Would they show evidence of bias against the new bogus data? I'll bet they would.
I once picked up a coupon for a discount "Bowen massage" not knowing what it was. (I was suffering a bit of chronic pain at the time, so I was willing to give all sorts of things a try.) I showed up and it turned out to be a man who touches all your major muscle groups for a total of 2 minutes, while you spend 58 minutes lying on your belly in a dark room. He said that my muscles will "remember" the healing power of his touch. I knew that what he was doing was bullshit, but it worked! For about a day I felt a lot less pain. That's how powerful placebo can be. Now I understand why the homeopaths are still in business.Bloth said:Got to agree, a lot of people still want to believe in bullshit: Astrology, homeopathy, clairvoyancy, religion and conspiracy theories kind of prove it.
http://www.youtube.com/watch?v=_Irvuafg5GM
Well, here is Hume's view:Thothistox said:Hume was speaking in very strict terms. I'm sure that he accepted causality as a useful myth in an everyday context. For example, I doubt anyone would claim that we should stop researching medicines because we cannot establish a strict causation between them and the ailments they purportedly assuage.
Hume was an empiricist, who believed that all our concepts (he called them 'ideas') are derived from impressions given in experience. He discussed several concepts, one of them a concept of causation, and how we acquire this concept. It was a common belief that causation is a feature of the world, that things cause other things to happen, that stones break windows, and such, and that we can actually observe one event causing another. Also this concept involves a necessary connection between causes and effects, that causes necessitate their effects.
What Hume claimed is that we can never observe any necessary connections in the world, we only observe regularities, some events always follow other events. But there is nothing we observe in nature can ever tell us that something is a cause, and something is an effect. Consider two cases: (1) billiard ball A hits a billiard ball B, and B starts to move, and (2) ball A touches B, but B starts to move not because A hits it, but because there is a small motor inside. If the timing is just right, the second case would look exactly like the first. There will be no difference with impressions, and we won't be able to discern a causal difference - B moving because it was hit, or because it has a motor inside it.
So we do not get any impressions of necessity and necessary connection between events from the events themselves. Still, there should be some impressions we derive this concept, and he claims that this impression comes from our minds. After observing things happening many times, we form a habit of expecting this pattern to continue. This impression of necessity is a work of imagination.
In essence, Hume talks about our concepts, and not about causation itself. May be there are necessary connections between events in the world (realism about causation), but they are unobservable. For a realist about causation, causation is not just a useful fiction. In any case, it is fair to say that we can't observe any causation directly, and have to infer causal relations based on correlations we observe. That's what science is doing. The same observations may be interpreted in many ways, positing very different unobservable theoretical entities. But it is probably one of the areas we simply can't have a complete certainty.
Obviously, people shouldn't stop research just because philosophers have no frigging idea about a lot of things, especially concerning science. Science seems to work just fine and make our lives better. Still for a scientist it would be good to study some philosophy of science, in order to be aware of limitations and problems in an attempt to provide rational foundation for it.
By using the "being reasonable" phrase I meant a more colloquial meaning, i.e. being opposite to a whimsical, emotional person, unconstrained by logic and consequence. This has nothing to do with being numerable, and yet I got the feeling that the authors didn't really differentiate between the two. (Heck, they only 'discovered' that numerate people could be unreasonable when applying the data they were given if it conflicted with their preconceptions.)Veleda said:As far as I understand, they weren't measuring for being "reasonable." Being numerate shows that the subjects had the capability to understand the data and make an accurate assessment. It's a baseline so that you can see how they react to other factors.
The saying goes "there are lies, damned lies, and statistics".
Was this study really necessary, to point out something that we all know? That people will interpret things their way?
Was this study really necessary, to point out something that we all know? That people will interpret things their way?
I skimmed through the PDF trying to find the exact wording of the questions on the test but I couldn't seem to find it. Did it explicitly ask the test-takers, "What does this data say?" Or did it ask them a more general question like, "What should the city should do about gun control?"
If it was the latter case, then I think the test has real framing issues because it's no longer testing the people's ability to interpret the data but rather it's asking them to make a value judgment which might involve their prior knowledge of other statistical studies. And if they already know about real studies on the issue, that's probably a lot more convincing than an intentionally made-up data table which they probably assume is made-up for the sake of the test (the fake data probably doesn't cite a fake source from a real journal, so if people actually know of sources from real journals then that would seem more authoritative).
If it was the latter case, then I think the test has real framing issues because it's no longer testing the people's ability to interpret the data but rather it's asking them to make a value judgment which might involve their prior knowledge of other statistical studies. And if they already know about real studies on the issue, that's probably a lot more convincing than an intentionally made-up data table which they probably assume is made-up for the sake of the test (the fake data probably doesn't cite a fake source from a real journal, so if people actually know of sources from real journals then that would seem more authoritative).
The wording used appears to have been "Please indicate whether the experiment shows..." (at least in the "skin rash" case; the equivalent legend was not shown for the "gun control" case, but it would have been a gross blunder to use any other).WardDragon said:I skimmed through the PDF trying to find the exact wording of the questions on the test but I couldn't seem to find it. Did it explicitly ask the test-takers, "What does this data say?" Or did it ask them a more general question like, "What should the city should do about gun control?"
If it was the latter case, then I think the test has real framing issues because it's no longer testing the people's ability to interpret the data but rather it's asking them to make a value judgment which might involve their prior knowledge of other statistical studies. And if they already know about real studies on the issue, that's probably a lot more convincing than an intentionally made-up data table which they probably assume is made-up for the sake of the test (the fake data probably doesn't cite a fake source from a real journal, so if people actually know of sources from real journals then that would seem more authoritative).
I believe their intent was to encourage the subjects to interpret the "results" alone: what "the experiment shows". But they rather effectively, even if inadvertently, demonstrated that the subjects did no such thing.
"Why did the chicken cross the road?"
"Under carefully controlled experimental conditions, the chicken did what it damn well pleased." [attr. to B. F. Skinner]