BBC article around the time

The cause of the extremely high California prison spending is that it seems that it is extremely expensive to imprison people in California, at least 2.5X what Texas spends, and even more than what New York spends. I.e., its not that CA imprisons so many people, relatively speaking, as that does not seem out of line with its population and demographics.

On the contrary, the US underclass is probably more Hispanic than black these days. The California ghettos have been getting very Hispanic for a couple of decades now, leading to different sorts of ethnic frictions.

Gentlemen,

One explanation for all this is California crime trends. The improving trend in crime rates from @1990/1992 began reversing from 2001-02. That mini-trend seems to have been squelched by 2004, and as of 2005 at least was declining again.

http://www.disastercenter.com/crime/cacrime.htm

Interestingly, overall per-capita California crime rates are back to 1960 levels, and the crime explosion of the 1960’s-70’s has been entirely reversed. I don’t know if this is solely due to public policy (policing and incarceration) but it seems likely that this played a significant part. A huge public policy success overall.

As for comparisons with other states – California has a relatively young population, which means more college students and more criminals, and everything around here is much more expensive. CA public higher education spending per capita is also several times that of other states. There are few places that spend so much per public college student.

Thanks for the pointer, Kieran. At a glance, it looks as though my hypothesis is partly correct, but not the whole picture. White men with no high school diploma are indeed incarcerated at an elevated rate—over 11 percent—by age 30-34, but that’s still nowhere near the nearly-60 percent figure for black men. The obvious question: what are all those non-criminal uneducated white men doing, and why aren’t more black men doing the same thing?

This is not about black men.

The tables are in the paper linked in the post, dan.

In general, in a time-series, use a baseline that shows the data not the zero point. If the zero point reasonably occurs in plotting the data, fine. But don’t spend a lot of empty vertical space trying to reach down to the zero point at the cost of hiding what is going on in the data line itself. (The book, How to Lie With Statistics, is wrong on this point.)

How do you determine this from the quote you selected? Or are you using other statistics not quoted to make this statement?

How much of the increase in incarceration rate is due to “three strikes and you’re out” and other more stringent incarceration policies?

Has anybody tried to estimate the distribution by socioeconomic class of the benefits of a marginal dollar spent on the prison system vs. the higher education system? It would be very tricky to estimate the deterrent effect of money spent on the prison system, of course, but I’d think that the benefits of higher education dollars would be so skewed towards the middle and upper classes, and the benefits of reduced crime towards the lower classes, that even assuming very modest deterrent effects, the prison dollars would still be more redistributive.

A paper by Pettit and Western provides some estimates, notably the astonishing finding that in the cohort born between 1965 and 1969, thirty percent of black men without a college education—and sixty percent of black men without a high school degree—had been incarcerated by 1999. Recent cohorts of black men “are more likely to have prison records (22.4 percent) than military records (17.4 percent) or bachelor’s degrees (12.5 percent).”

Two thoughts:

(1) The rate for black men seems to be declining—presumably that means that the major sentencing crackdown of the 1980s and 1990s had at least some effect. The obvious question is whether the decline is slow but steady, indicating that the overall strategy is working, or whether it’s flattened, indicating that new strategies are in order. What do the figures say?

(2) I’d like to see the incarceration rates for black men without high school diplomas compared with the corresponding rates for white men. Given the relative ease with which high school diplomas can be obtained in today’s public school system, I would expect that the strong correlation between failure to obtain one and incarceration represents a single implicit “career choice” more than a lack of options.

Still, there are way too many people incarcerated in California. Many of them for trivial things.

I agree, but my point was in reference to this particular graph and the statement that it was a bad graph just because it didn’t go through 0 and therefore gave an impression of the value ‘quintupling’, which presumably is unlikely for anyone reading the above graph (especially when the start/end values of $5.2 and $10 for that particular trend are explicitly added onto the graph).

Running through zero on a graph can be necessary in some cases, and if its sufficiently close to the base of the graph it should be added anyway for simplicity, but if you have to add 50% to the height of your graph just to get the origin to be 0, then its unlikely to be a good idea. If you went back in time with the series and had values of 2 to 12 on the scale, then expanding it back to 0 might be reasonable, but either way the values shown on the graph are very clear as it is, even to a cursory glance.