Thursday, April 10, 2008

Strictness index: Which states are the toughest? The most lenient?

That 'strictness index' is a way of gauging how tough or lenient states are towards their criminals by looking at the net effects of how they apply criminal justice--namely how many people they put in prison--rather than by what statutes might say about how a particular crime is to be dealt with.

It is figured by taking incarceration per 100,000 residents (then multiplying it by a factor of 100 for ease of presentation) and dividing it by the 'crime index' (described below). If a state has 500 people locked up and its crime index is 8,000, then its strictness index is 6.25 ((500*100)/8000). If another state locks up the same proportion of people, 500, but its crime index is only 6,000, it has a higher strictness index, at 8.33. The second state is thus tougher on its criminals than the more lenient first state is.

The crime index is figured by adding violent crime and property crime (both per capita) together, weighing violent crimes nearly 18 times as heavily as property crimes. Violent crime includes murder and non-negligent manslaughter, forcible rape, robbery, and aggravated assault. Property crime includes burglary, larceny, and motor vehicle theft.

The reasoning behind the weighting involves a couple of things. In 2001, the FBI reported that just under half (49%) of all inmates in state prisons were there for having committed violent crimes, while one-fifth (20%) were being held for property crimes. Thus there are about 2.45 times as many thugs behind bars for violent crimes as there are for property crimes. Nationally, there are more than seven (7.31) property crimes committed for each violent crime committed. Property crimes are more frequent, but unsurprisingly, most severe punishment is meted out to perpetrators of violent crime.

From this, I estimate that violent crimes are about 18 (2.45*7.31) times as serious in the eyes of the public as property crimes are, and thus weight the two factors comprising the crime index accordingly. This doesn't mean that the crime index is 18-parts violent and only one-part property, because there are so many more property crimes than violent crimes that take place. It is, on average, 68% violent and 32% property.

A shortcoming is the absence of data on drug-related offences. Nearly one-quarter of inmates are in prison for such crimes. Unfortunately, I can't find numbers broken down at the state level for them.

The incarceration rates are from Pew's massive report entitled One in 100: Behind Bars in America 2008. Despite the year mentioned in the title, the data used for the strictness index are from '05. Data on violent and property crime, which come directly from the FBI, are also from '05. Keep in mind, we're looking at the percentage of people behind bars relative to the percentage of people committing crimes. So a state can have lots of people in prison and still not be very strict, if it has lots and lots of crime.

The strictness index, by state (colored according to how each voted in the '04 Presidential election):

1. South Dakota12.63

2. Mississippi

3. Idaho10.74
4. Virginia9.85
5. Kentucky9.85
6. North Dakota9.61
7. Wyoming9.48
8. Wisconsin9.35
9. Georgia8.36
10. Louisiana7.94
11. New Hampshire7.67
12. Alabama7.66
13. Connecticut7.28
14. Vermont7.16
15. Texas7.06
16. Oklahoma6.99
17. Indiana6.88
18. Colorado6.53
19. Montana6.43
20. Maine6.17
21. New Jersey6.13
22. Pennsylvania6.06
23. West Virginia5.90
24. Utah5.87
25. Arizona5.76
26. Delaware5.68
27. Ohio5.61
28. Oregon5.57
29. Kansas5.43
30. Missouri5.36
31. California5.35
32. Iowa5.12
33. Michigan5.11
34. Nevada5.00
35. Florida5.00
36. Arkansas4.98
37. North Carolina4.98
38. Nebraska4.92
39. New York4.78
40. Hawaii4.77
41. Alaska4.72
42. New Mexico4.68
43. South Carolina4.62
44. Rhode Island4.34
45. Washington4.19
46. Tennessee4.12
47. Maryland3.94
48. Illinois3.91
49. Minnesota3.57
50. Massachusetts3.38

Not surprisingly, conservative states tend to take a tougher stance on crime, liberal states a more lenient one (although the libertarian-left states of Maine, Vermont, and New Hampshire are relatively harsh). The correlation between strictness and Bush's share of the '04 vote is .47 (p-value of zero).

Geographically, the Northwestern interior and the South are the most stringent (with the notable exceptions of Arkansas, Tennessee, and the Carolinas), while the West coast and parts of the Midwest are the most permissive, but no clear patterns emerge (see a visual representation here--click on each of the ranges at the bottom left of the map to better distinguish individual states from one another).

Whining about how 'whiterpeople' presume their cultural and political proscriptions are best because places with lots of whiterpeople are the best places to live and whiterpeople as a group do quite well is a reoccuring theme of mine. Bridget Moynahan might be able to care and provide for a child out of wedlock, but an unskilled black woman with a double-digit IQ living in the urban core cannot. Piety might seem silly to Richard Dawkins, but for the left half of the bell curve it can be a saving grace. A generous welfare state with libertine social policies may function well enough when all its inhabitants are Dutch, but things get trickier when Somalis and Moroccans start showing up in significant numbers.

And opposing mandatory sentencing in favor of judicial discretion, probation, and rehabilitation for serious criminals might not lead to degeneration in the streets of Cambridge, but in Tucson that outcome is more than just a theoretical possibility.

Or are the results even that contingent upon demographics of the state in question? The data suggests otherwise. The crime and strictness indices correlate positively at .52 (p-value of zero). States that are more willing to throw people into the slammer experience less crime even without adjusting for estimated IQ, affluence, average educational attainment, and the like.

How can that be if Mississippi is so strict and Massachusetts so lenient? Obviously you're at greater risk of being murdered in the former than you are in the latter. But Mississippi experiences relatively little crime for being a Southern state. The surprisingly lenient places in the South mentioned previously--Arkansas, Tennessee, and the Carolinas--all suffer from more crime than Mississippi does.

While Bay Staters can feel good about having a few less people in their prisons than North Dakotans can (356 to 359, respectively), they have to accept that the violent crime rate in Massachusetts is almost five times as high as it is in North Dakota! Nearby Connecticut incarcerates people at a rate more than 50% higher than Massachusetts does, yet it experiences less crime.

Similar relationships exist in other places. Take tranquil states like Wyoming and Nebraska, for example. The former incarcerates 690 people per 100,000 compared to the latter's 421. But despite (or perhaps in part because of) Nebraska's more diminutive prison population, its crime index is almost 20% higher than Wyoming's is.

Based on the proportions of their populations behind bars, you'd think Indiana suffered more from crime problems than does its more sophisticated neighbor, Illinois. Not so.

Even Minnesota, home to the fictitious Lake Woebegon (and in many ways its embodiment at the national level) experiences crime rates about 12% higher than the perpetually impoverished and 'backwards' state of West Virginia does. The coalminers throw more people in jail, though (443 to 300).

The US' 'unique' demographic situation (28% NAM) among developed nations leads me to believe that we do not, as a country, have "astronomical" rates of imprisonment, nor do our individual states that are less hesitant than others to lock people up. I agree with Ron Guhname instead:
The adjective I would use is appropriate.


Steve Sailer said...

Very interesting.

Have the FBI crime statistics been validated against other measures of crime, such as the annual crime victimization survey or the murder rate by state? I'm wondering how standard is the methodology across states, or if the FBI is just publishing whatever numbers they are fed by local authorities.

Steve Sailer said...

It would be interesting to correlate strictness with the crime rate.

Audacious Epigone said...


The murder rate and my 'crime index' correlate at .76 (the murder rate correlates with the total violent crime rate--that is, without taking property crime into account--similarly, at .76). The p-value is zero in both cases, but those relationships still aren't quite as strong as I would've expected. As murder is putatively the most likely to be reported (and personal assault and various property crimes among those least likely to be reported), it's usually looked at as the most reliable measure of criminality. Still, I wouldn't expect murder and other crime to track at exactly the same rates--John Lott's More Guns, Less Crime would be in trouble if they did. So I'm fairly comfortable with the numbers.

Here's the FBI's official discussion of the two methods, the UCR (which I used) and the NCVS. The numbers are fed by the state's on a monthly basis, and the FBI states that it then checks them for validity by looking at variances and other data, for whatever that's worth.

One oddity about the victimization surveys is that rates are sometimes stated as per household rather than per capita.

But I can't find NCVS state-level data, either in the downloadable spreadsheets the DOJ provides, or its 143-page annual report on the results of the surveys. Any ideas on where to find them? My hunch is that the largest variances exist in states with a higher percentage of their populations living in impoverished urban areas, where crimes are less likely to be reported (East St. Louis, for example).

Audacious Epigone said...

Re: correlating strictness with the crime rate, I did so with my crime index (which consists primarily of violent crime rates), and they correlate positively at .52, which I found surprising. But lenient states tend to have more crime than tough states do, although it's hard to see as much until you look by regions of the country.

Audacious Epigone said...

Here's the link to that NCVS report.

LemmusLemmus said...

The last time I heard about this, state-level NCVS data were generally not available to the public. To use them, you need to a) be an academic researcher, b) have your research project approved and c) come to Philadelphia (?) to analyze the data in-house.

Audacious Epigone said...


That very well could be. I'm not having any luck, anyway. Glad someone from Germany could answer a question about accessing data from US law enforcement that I didn't know the answer to :) Thanks.

Steve Sailer said...

How big is the annual NCVS sample? 50k? So that would be 1k per state, which sounds kind of small. You'd probably want to aggregate it over a decade or so. 10k per state sounds good.

But the feds won't let us look at the data anyway, so it's all pretty theoretical.

LemmusLemmus said...

The size of the annual NCVS is about 90 000, so that's more like 2 000 per state - on average. That would be a fine sample size for high-frequency events like voting, but for stuff like rape you get huge confidence intervals. For overall crime, my guess is that you get useful numbers; I have not done any calculations on this.

Note that what I described in my earlier post is the state of things after the great liberalization concerning the NCVS data around 2001 or so. Certainly the best use of one of the mightiest datasets in the social sciences, financed by the taxpayer.

Audacious Epigone said...

Echoing Lemmus, nationally the forcible rate rape is in the low 30s per 100,000 people, so at 2,000 per state, that'd average out to less than a single person per state. So state level estimates couldn't be reliable at that sample size.

Sleep said...

Wow. I am very impressed with conservative northern New England. Being a native, I am so tired of the stereotype that we're just an extension of Massachusetts and of seeing statistics and voting results prove the stereotype right.

One thing that's always puzzled me is the stark difference in political outlook between Minnesota and the Dakotas. People tell me Minnesota's so liberal because of all the Scandinavians, but why then are the Dakotas not merely conservative, but often extremely so? I've heard various explanations but none of them really make sense by themselves, and they contradict each other, so they can't all be contributing factors.

Can Brazil count as a developed country? If you look at a map of income in Brazil by geographivcal area, you can see the southern areas (which are, of course, mostly white) are as rich as some of the poorer EU countries. It's only because of the huge expanse of poor areas that Brazil's economy is classified as Third World.

Sleep said...

Average annual income in Brazil:
Amazon: $4248
Northeast: $3243
Interior: $8616
Southeast: $9126
Far South: $7793
These numbers are subject to variations in the exchange rates. If you factor in the cost of living being low, the per capita income of the southern areas would probably rise to the level of some of the poorest US states.

Audacious Epigone said...


I was thinking just about the same thing, honestly, only from a partisanly conservative perspective: The South, because of its relatively poor performance among whites and of course its significant black populations, tends to make 'red' states look worse than blue ones. If a few of the Dakotas, Montana, Colorado, Wyoming, Idaho, Nebraska, Iowa, and Kansas become safely 'blue', it'll really start looking 'bad'.

Also, where are you getting those Brazilian stats from? Very interesting. I'd appreciate a link if the source is online and you have one.

Sleep said...

I just used Wikipedia (the "Regions_of_Brazil" page) and the currency converter. But holy crap! The recent drop in the value of the US dollar has caused per capita incomes across the world to soar to previously unimaginable levels when quoted in US dollars. I have to retract my statement about southern Brazil being on level with the low-rung EU states, because those EU states have doubled their incomes in just a few years thanks to the change in the exchange rates. (Which is why I should have avoided using literal GDP's and used PPP all along. Even the Heritage Foundation has switched to PPP now.)

Click here to see the list of countries by GDP.

Audacious Epigone said...


Thanks, I'm interested in that generally, how dollar depreciation has effected nominal wealth. When I get the time later in the week I'm going to look at it in detail.

Anthony said...

La Griffe ran some numbers which showed that the incarceration ratio between blacks and whites was lowest in states which had the highest proportion of its white population in prison, which he used as a proxy for strictness. It might be interesting to run the correlation between strictness index and black/white imprisonment ratio.

Audacious Epigone said...


Thanks for the reminder. That is a good idea. I'll take a look at it.

Audacious Epigone said...


The two correlate in the expected direction (as La Griffe's B/W ratio decreases, the strictness index created here increases, but it is weak--only .13, with a p=.37). But just by eyeballing and removing three outliers (CT, HA, and WI), the correlation jumps to a noticeable .27 (p=.07)). I'm not sure what the variances in these three states stems from.

Anonymous said...

hi i was wondering if you could do a update or if you feel nothing has changed thanks =)

Anonymous said...

This list is not quite accurate. I have a state felony back in 1985 and I cannot even apply for a pardon or an expungement in New Mexico for a nonviolent taking of hotel services the crime valued at less than $800.00. I never served time in prison but yet this state treats all felons who did serve time in prison the same way as the federal government.