Thursday, January 29, 2009

Federal corruption convictions per 100,000 residents over 18

Based on Department of Justice statistics of federal corruption convictions from 1998-2007, USA Today created a map to visually represent per capita rates of corruption by state. Other than a stained western Appalachian belt, no apparent patterns emerge, which tends to induce some skepticism into the value of the method of measurement.

Before discussing that, it's worth noting that children are not the targets of federal corruption investigations, yet having a relatively large number of them pads younger states in USA Today's analysis by spreading out the number of convictions over a larger population base. Older states do not fare particularly well. Thus, a ranking of the states by corruption convictions per 100,000 people aged 18 and older:

1. North Dakota10.70
2. Alaska10.19
3. Louisiana10.09
4. Mississippi9.76
5. Montana7.93
6. Kentucky7.44
7. Alabama7.13
8. South Dakota6.79
9. Delaware6.61
10. New Jersey6.33
11. Ohio6.27
12. Florida5.78
13. Pennsylvania5.75
14. Illinois5.19
15. West Virginia5.19
16. Tennessee5.16
17. Hawaii5.15
18. Virginia5.10
19. New York4.72
20. Connecticut4.13
21. Oklahoma3.92
22. Massachusetts3.73
23. Arkansas3.72
24. Missouri3.53
25. Maryland3.47
26. Wyoming3.44
27. Idaho3.41
28. Texas3.21
29. Rhode Island3.18
30. Arizona2.93
31. Michigan2.85
32. Wisconsin2.84
33. Vermont2.66
34. Indiana2.57
35. North Carolina2.56
36. Maine2.41
37. Nevada2.37
38. Georgia2.28
39. South Carolina2.15
40. Colorado2.07
41. New Mexico2.04
42. California2.01
43. Utah2.01
44. Washington1.99
45. Kansas1.81
46. Minnesota1.67
47. Iowa1.52
48. New Hampshire1.37
49. Oregon1.24
50. Nebraska0.90

See a visual representation of the data here. A spreadsheet of the data are here.

This corruption index shows no meaningful relationship with the usual suspects. Corruption does not reach statistical significance (p<.05) in correlating with estimated average IQ, the percentage of a state's population that is white, Presidential voting patterns in '04, economic inequality as measured by the gini coefficient, the violent crime rate, or median income. There are moderate relationships between the percentage of a state's population that is black (.30, p=.03) and the percentage of a state's population that is Hispanic (.33, p=.02), but these are not nearly as vigorous as the relationship between race and other kinds of crimes are.

Weather aside, if existence in bustling urban centers isn't a necessity, Montana and the Dakotas are generally regarded as decent places to live if, yet they are among the worst performers on the list. Germans and Scandanavians are usually not among the ancestries that first come to mind when thinking about governmental corruption. Are Minnesota and North Dakota really that different from one another? How about Montana and Wyoming?

The USA Today article quotes a North Dakotan suggesting this method of measurement might just as well indicate the level of vigilance against corruption as it does the prevalence of corruption itself:

Don Morrison, executive director of the non-partisan North Dakota Center for the Public Good, said it may be that North Dakotans are better at rooting out corruption when it occurs.

"Being a sparsely populated state, people know each other," he said. "We know our elected officials and so certainly to do what the governor of Illinois did is much more difficult here."
Population density doesn't correlate with corruption, either. But the first part of Morrison's comments may have merit. Last year, I created a criminal strictness index by state, computed by comparing the levels of violent and property crime to the incarceration rate. The simple idea being that if two states experience the same crime rates, but state A has a higher percentage of people behind bars than state B does, state A is a stricter state than state B is.

While these are federal convictions, they are tipped off and aided by state and local authorities, and political pressures by state and local officials presumably play some role as well. The strictness index correlates more strongly with corruption (.37, p=.01) than any other variable I looked at.

The Inductivist recently turned to the GSS to take a look at corruption by US geographic region. Amalgamating populations and convictions by state into the nine Census bureau designated areas and comparing them to the average corruption scores Inductivist found does not yield a statistically significant relationship (p=.16). I suspect that federal corruption convictions per capita present too much of a mixed bag to use as a proxy for the level of actual corruption in a state.

Thanks to Randall Parker for the initial heads-up on federal corruption conviction numbers.


Anonymous said...

The map shows corruption convictions rates, not the corruption rate. For example, North Dakota could have a order of magnitude lower rate of corruption than Illinois, but have much higher rate of detecting and convicting what little amount of corruption they have.

Anonymous said...

At a glance, it looks like the smaller states have higher deviations--- that suggests mere randomness.

Audacious Epigone said...


Right, given the relationship to the 'strictness index' (my measure of how tough states are on punishing criminals), that's what I think this method better measures--not actual corruption, but how diligently it is fought (to the extent that it measures anything of utility at all).

Anonymous said...

"Crookeder than a Lousiana Politician".................

Now that old saying makes sense.

Stopped Clock said...

It also covers only 10 years and so it's sort of vulnerable to the Brownian motion that can shake up an entire administration in a year and leave a statistical imprint that will last many years. When this topic came up on ½σ last month I vaguely remember someone mentioning that either the North Dakota or the Montana score was explained largely by just one city whose employees had been using tax money for personal expenses and were thrown out of office en masse. But the thread is gone now and the Google cache of it doesn't seem to provide any links to the original information.

Steve Sailer said...

Illinois has a political culture dedicated to training aspiring politicos in how not to get convicted and not cause colleagues to get convicted. I doubt if North Dakota has that cultural capital.

Audacious Epigone said...

Hah, Steve, that's a great point. Crafty people who are surrounded by other crafty people know how to avoid getting caught better than people who are forging ahead alone in their craftiness.

There are several potential influences here blowing in different directions. The list is interesting, but I don't see it as illustrating any clear pattern.

Steve Sailer said...

There should be some objective way to measure rake-offs, but I'm not sure what. If there was something that every municipality does, but it costs twice as much to do per capita in Chicago as in Minneapolis, say, that might give you a clue.

Anonymous said...

Might it be the % of the population that is Indian or is living on reservations? Any corruption there would be federal, where as a corruption in say, Des Moines City Council would be handled by the state?

Audacious Epigone said...


I'm not sure if convictions on reservations are included in state totals. Aren't they considered "domestic dependent nations" or something like that? It would be very helpful to have demographic breakdowns on the convictions.

But the correlation between per capita convictions and the NA% is .32--about the same as the black and Hispanic %s looked at individually, so if those convictions are included, it sounds plausible.

Anonymous said...

Have you checked the correlation with single party dominance? Califoria's government is utterly dominated by a single party. It seems to me that such a lopsided situation would make convictions more difficult. That leads to the question of which party members get convicted. If most of Californis'a corruption convictions are of republicans, or if most of Utah's convictions are of democrats, it seems almost certain that they'd have political underpinnings.

Another thought-
What are the per capita rates of employment in jobs that are succeptible to corruption charges? As massive as California's government is, I doubt it employs as many, per capita, as Connecticut's. Both have only one governor, for instance.

Tony said...

How about a rate of convictions per gov. $ spent in the state? This would partly capture the info re # of jobs subject to corruption, since a job has to have a salary, but it would also capture contractor jobs and other spending that isn't job-related directly.