Would you rather live next to a Democrat?

(Flickr / ClatieK)

(Flickr / ClatieK)

Do Democrats make better neighbors? According to a long article by Ken Stern in The Atlantic, the correct answer is first no, then yes, then it’s unclear, then it’s irrelevant.

The most important answer is that last one. Political party affiliation really isn’t a great indicator for kindness or generosity, despite the “emerging body of evidence” Stern refers to. There is a collection of studies and articles arguing Democrats and Republicans think and act differently, but the conclusions are flimsy.

For example, Stern claims to discover “a series of differences—albeit not statistically significant ones—showing that Democrats more frequently loan dishes or tools to their neighbors, help strangers carry their belongings, offer up their seat on a train or bus, allow strangers to cut ahead of them in line, and give food or money directly to someone in need.” (emphasis added)

But Stern ignores the fact that statistical significance is not just some irrelevant technical term. A conclusion that is not statistically significant means that it is a failed conclusion. In technical terms, there is no justification here to reject the “null hypothesis” (that Democrats are just as likely to engage in these practices as Republicans.) In practical terms, his conclusions above are ridiculous.

In a much-discussed book, Arthur Brooks argued that conservatives are much more likely to be generous with their money and time than liberals. But this conclusion, while stronger than the one above, does not avoid the “correlation is causation” fallacy. Breaking down the data, we see that conservatives are more generous because religious people are more likely to be generous and, well, religious people are more likely to be conservative.

While there certainly may be merits to further research in political determinism, we should take these results with a grain of salt. It seems that every year, another study will come out claiming one party is better than the other at some attractive quality.

For example liberal journalist Chris Mooney claims, “I show evidence suggesting that liberals tend to be right. So why is that? if you look beneath the surface of our politics, you find out we differ by psychological traits.” He has also written a book titled, The Republican Brain: The Science of Why They Deny Science and Reality.

But then Samuel Goldman in The American Conservative argues, “Although they pride themselves on being open-minded, liberals generally have far less contact with conservatives than conservatives do with liberals.” He claims that conservatives are, in fact, more open-minded.

There are seemingly reasonable explanations to be made for either case, but the resounding conclusion from all of this is that political science can often be masked conjecture. Think twice before dismissing your political opponent because he is supposedly hard-wired by science.

Originally published at Washington Times Communities.

6 responses to “Would you rather live next to a Democrat?

  1. Well done.
    I thought statistical significance had a much lower threshold when applied to human behavior, such as less than 40%. But this is a memory from a sociology class taken about 45 years ago. 🙂

    Likely not to pass the statistical test (even for correlation) is a statement I ran across in Christianity Today some time ago: “Those who read the Bible more often are likely to be more politically liberal than those who read the Bible less often.” How about that for a statistically vague claim?

  2. Well done! Is it correct that statistical significance requires crossing some threshold of standard deviations? I am thinking that, strictly speaking, causation can only be quantitatively assigned on some subjective, qualitative basis.

    I am all for what David Brooks calls epistemic modesty. I agree that there’s a lot of conjecture in popular articles, perhaps because potential permission to think ill of one’s neighbor draws a lot of page views.

    • Correct – usually, something is “statistically significant” only if there is a less than 5% chance that the same result could be observed even if the null hypothesis was true. (In this case, the “null hypothesis” is the assumption that Democrats and Republicans are no different in their practices of “neighborly” behaviors)

      It prevents against Type 1 errors from the sample (in this case, assuming that Democrats are more likely to wash dishes, watch pets, etc. when they’re really not.)

Comments are closed.