A similar problem goes in the other inferential direction. For example, according to a blog entry at Talking Points Memo, CNN discussed a poll last week that showed that:
Texas voters that watched the Clinton-Obama debate supported Obama by a margin of 20 points, Texas voters who followed news about the debate but didn't watch it broke even, and Texas voters who paid no attention to the debate went for Hillary by 20 points.According to an "expert" guest on CNN, the poll's results show that:
...downscale voters look to the political process to "deliver" for them, and that's why they want specifics, and that's why they support Hillary. Upscale voters, on the other hand, want to "identify" with a candidate, and that's why they support Obama.Note that the original polling data said nothing about the socioeconomic status of those polled.
Take another example that hits closer to home: the continued drop in home resales and prices in January. Economists are keen to say things like, "home prices haven't dropped enough yet," and "people are waiting until the prices go down further." Of course there are many reasons why people aren't buying houses, right? And, more importantly, it turns out that home prices actually increased in some parts of the U.S. Another example of this explaining a macro phenomenon with individual behaviour comes whenever the stockmarket takes a plunge: "Mr. Analyst says that people are selling out of fear that oil prices will rise next week." That's a pretty amazing statement to make when you consider that millions of people are selling shares and there are still millions of others who are buying those very shares.
So, in my first set of examples, individual (micro) behaviours are used to refute the group (macro) phenomena. In the second set, the micro behaviours are used to explain macro phenonema. We're probably comfortable making such leaps from micro to macro (and the reverse) because it's quite hard to imagine that human behaviour is determined by many, many factors. In psychology, we are more comfortable predicting behaviour on the basis of probabilities, although we are nowhere as precise as we expect the weather bureau to be.