Articles to 2011-08-05

Zum Seitenende      Übersicht Artikel      Home & Impressum

First the link to this week’s complete list.

Isn’t economics wonderful? Those beautiful diagrams in Beale et al. would represent a distinct improvement to quite a few arts exhibitions. And isn’t it a noteworthy science of renown, that can warn about and predict a danger four years after it happened? Didn’t they also foresee and prevent the stock setback of 1987 and the dot-com crash? Looking at the current article we find it typical of the genre. A few rather primitive and simple-minded assumptions lead to a bit of easy high-school mathematics and everyone able to solve that believes himself to have gained deep and thorough insight. The one thing I have yet to see is assumptions with error bands and their propagation into the results. In the sciences just formulating a hypothesis is only the first step in planning research, that may or may not yield results. In economics the hypothesis alone suffices for an article in a prestigious peer-reviewed journal. And of course all asset classes and industrial sectors have the same probability distribution for failure and they are all completely independent from each other, so that one of them failing does not compromise the stability of any of the others. Isn’t life simple? I look forward to all the wonderfully effective market regulation arising out of this.

Before analysing Butler and Broockman we need to translate a few things from US-American into common English (this bit took me some time first):

That said, let us dig in. The first impression is an article without any graphic presentations at all. Fine for an essay on poetry or theology but strange in a quantitative context. The raw data in SOM table 2 are given in a format such that the only way to extract them for reanalysis is by manually retyping them, but, to be fair, they are given and they are complete, which is more than can be said for most other studies.

First, the main result is valid. White candidates more often answer Jake (60.8 %) than DeShawn (57.3 %), p < 0.05. Is this racism? Perhaps. The names were chosen for being maximally racially distinct. Apart from that, according to SOM table 1 Jake Mueller conveys no information while DeShawn Jackson is two to four times more likely to vote Democratic than Republican. (On top of that both Christian names suggest a working class background with little education to me, but I’m not American.) So the name does carry valid information besides race. Might that include the likelihood actually to vote, even seeing they bothered to ask?

Minority candidates are significantly less likely to answer at all (38.6 %) than white candidates (59.0 %), p < 0.00001. If they do, their preferential treatment is quite pronounced, Jake 34.1 %, DeShawn 43.5 %, p < 0.05. That said they still answer DeShawn less often than their white counterparts do, p < 0.0001.

Political candidates are not stupid, or at least their election staff isn’t. How probable is it, that somebody lacks the basic knowledge about how to vote but is committed enough to want to become involved in primaries? Candidates more often answer uncommitted writers (57.9 %) than affiliated ones (55.8 %), p < 0.2. Looking at the innocuous Jake alone this becomes a significant (p < 0.05) 60.5 % versus 55.9 %. So it seems they do smell a rat or a hoax. Note that this difference is actually bigger than that between Jake and DeShawn. Might "DeShawn Jackson" be another giveaway? The frequencies in SOM table 1 point that way. DeShawn is treated equally whether committed (55.8 %) or not (55.3 %), p = 0.83.

In conclusion, are there differences in treatment? Yes. Do they point to racial inequity? No, except perhaps unequal commitment of legislators towards their constituents. As a last remark I see that nearly all results I looked at are significant, while few in the original report are, which suggests a skewed view following an agenda.

"And this is [science]?" (Richard Phillips Feynman) (What he actually said was medicine. I wonder, do I have a peculiar stare too when confronted with voodoo?)

Zum Anfang      Übersicht Artikel      Home & Impressum

Creative Commons Attribution-Share Alike 3.0 Unported License Viewable With Any Browser Valid HTML 4.01! Valid CSS!