[Updated with a new statistic for chapter 3, I had to check the raw data for comparable stats]
[ Updated 2nd time with information I happened to find interesting in the marriage report I checked for stats]
em·pir·i·cal
1: originating in or based on observation or experience
2: relying on experience or observation alone often without due regard for system and theory
3: capable of being verified or disproved by observation or experiment
4: of or relating to empiricism
So, what to say about a rather silly book Do-Gooders by Mona Charen (hmm, my Amazon Associates link function hasn't been working, this is annoying, not that I think anyone would want to buy this thing, but I like linking to the edition I read). This is my first experience reading a book from the right wing rage machine (the left has one too, but they are more prone to documentaries). I don't think anyone is expecting much factual accuracy from this kind of work so I'll only go into things moderately. What does jump out at me though is this quote from the book, "Conservatives tend to underestimate the worth of their own principles-freedom, self0reliance, tough mindedness, empiricism- and seem to accept the idea that liberal values-compassion, soft-heartedness (soft-headedness?), and equality- are superior." I can accept most of this as what I would expect from a political polemic, fair game for this genera. This book displays all these qualities and attributes the qualities with liberals it claims to. With one exception, there is nothing empirical about this book. It quotes a lot of numbers but it does not seriously rely on observation, it makes a great deal out of system and theory rather than experience and observation, and it doesn't follow any kind of systematic method for collection or use of data. The author jumps around between rates and absolute numbers with no regard for which is appropriate. The author pays no attention to base rates, quoting statistics to prove her point without acknowledging underlying trends. Most evidence presented is in the form or quotations and anecdotes, the fundamental attribution error (ascribing observations to human action rather than to situational factors) is rampant.
Now, from this sort of book I wouldn't expect empirics, except empiricism is singled out as a primary virtue. I'll go briefly through the book to point out the real howlers.
Chapter One - Judge Not
This chapter basically goes over the rise, and decline, of the crime rate in the US attributing crime primarily to liberals insistence on rehabilitation and light sentencing and the fall in crime rates to increased punishments. The big issue this ignores is that both the rise, and later fall, in crime rates were coordinated across all developed nations, none of which had the same tough on crime approach the US used. While US crime rates were worse in many aspects before the sharp rise, they remain so in many aspects. There are some anecdotes of particularly stupid things liberals said, but if I desired to do so, I'm sure I could find similar anecdotes from conservatives. This proves nothing. Crime is not an area of specialty for me and is overall not completely understood, so I've got little more to say. But if base rates and widely ignored correlations are not taken into account to provide an explanation, empirical methods are not being used. The main problem in this chapter is spurious correlation and the fundamental attribution error, base rates are ignored, causality is attributed because two things are happening roughly at the same time but precise dates are not established meaning that the comparison is meaningless, and all the incendiary rhetoric and quotes amounts to no more than meaningless anecdotes.
Chapter 2 Stoking Fear and Hatred in the Name of Racial Sensitivity
This chapter is mainly polemic quotes, I don't really have anything to say about this. With one exception, the author states that "about 180,000 ballots in Florida were not counted due to voter error: either undervoting, or overvoting. It is impossible to know how many of the "spoiled" ballots were cast by African Americans since we have a secret ballot."
Anyone that knows empirical methods will know this is bullshit. We do not have to know the provenance of individual ballots cast to establish how many ballots were cast by African Americans (within a certain degree of statistical error). If we know the demographics of given counties and if we know other relevant characteristics of the voting districts in question, and we know the participation rates for the districts, then we can use statistical methods to answer this. It will take a lot of slogging through boring data for graduate assistants but this is well within the realm of the possible (assuming data is collected for the relevant counties, not being very interested in electoral politics I don't know 100% that this data exists, but I believe it does). If more votes are being rejected consistently in African American counties than can be explained by other factors than it is possible to establish causality. I don't know if this is the case, but anyone that understands statistics will know that the secret ballot should have no impact on establishing the veracity of this claim.
Chapter 3 The Promise of Compassion
This is basically a long chapter on the culture of dependency argument. Here I'm going to break out some statistics, since I have them. A lot of ink is spilled over rising illegitimacy rates, illegitimacy increased among all groups in society not just the poor or ethnic groups. These rates varied with no relation to changes in welfare benefits. There is no causal relation between welfare and illegitimacy (welfare did allow more women to keep their children despite being single, however, simply giving them up for adoption is no sure way to have them adopted so this is a naive suggestion). There are legitimate concerns regarding the decline in marriage rates and the effect this has in children, but the dates and distribution of the problem isn't right for a welfare based explanation.
There are base rate problems, she claims between 1963 and 1973 welfare caseloads increased 230%. Population increased, the economy was substantially worse in the 1970s, and family structure and demographics have both changed. The raw number means nothing without associating it with other changes. Poverty also declined remarkably during this period and labor force participation also rose. What does the increased caseload have to do with anything?
She talks a bit about welfare reform. The main takeaway here is that work incentives were a good thing but the actual fear of liberals largely came to pass. There is evidence that the most poor and detached women that had been long term welfare recipients simply dropped off the rolls and did not start working, this pool has likely increased. The poverty gap (basically how much income is below the poverty line) among the most deprived groups has increased. On the other hand, employment has increased and welfare rolls decreased dramatically and spending declined from $24 billion in 1988 to $13 billion in 1999. But what isn't mentioned is the massive increase in work supports that accompanied welfare reform, spending rose from $11 billion in 1988 to $66.7 billion in 1999 and non-cash work supports increased from $9.5 billion in 1993 to $18 billion in 1999. Claims about jobs are also misleading, once the expansion was over unemployment rose and labor force participation fell, the lack of rise in welfare rolls had less to do with lack of need than it had to do with new restrictions. There is evidence that women use welfare more strategically, in the low wage sector of the economy jobs tend to be unstable so women delay going on welfare, despite lack of income, until joblessness is prolonged. They want to save their 60 months for when the really need it. This does not indicate reduction in need but that women value security over present income, they want welfare as insurance and are willing to undergo deprivation to retain a safety net.
Perhaps the best example of deceptive, and non-empirical use of statistical data is when she notes that the percentage of African American children raised by married parents increased from 34.8% in 1996 to 38.9% in 2001. What's notable about this is the implication that the percentage was rising before welfare reform and declined after. I couldn't find exactly comparable statistics, but the birth rate among unmarried women was already falling before 1996. According to "Nonmarital Childbearing in the United States 1940-1999, National Vital Statistics Reports, the birth rate among unmarried women had begun to fall in 1994 (and probably even earlier due to changes in reporting data in the early 1990s resulting in an overstatement of births to unmarried parents in the first few years of the decade relative to the rest). While this is change is a good thing, it is unlikely to have anything to do with welfare reform, unless a very high degree of political information and sophistication is assumed for unmarried poor women (also the number of people on welfare isn't large enough for it to be likely that welfare policy changes would be significant in national level statistics).*
Chapter 4 Rewarding the Worst Families
There are tons of problems with the foster care and adoption systems. It would be nice if there were enough wiling parents to adopt all the children being cared for by bad parents. The author does nothing to establish that her alternative is realistic. Trying to offer services to dysfunctional families is an attempt to make the best of a bad situation. Blaming liberals for this is just absurd. Not much more to say about this.
Chapter 5 The "Grate" Society
Yep, homelessness is bad. No, blaming this on deinstitutionalization is not really accurate. It is true that we let people out of institutions without putting adequate systems in place. But institutions were very destructive to the individuals involved and extremely costly. This chapter is mostly anecdotes and doesn't really present real alternatives. Yes this is a problem, but some empirical evidence rather than anecdotes would have helped. But, of course, "there has never been a more humane country in the whole history of humanity," so I guess the problems with both the earlier and current system are OK then.
Chapter 6 The Liberal War on Rigor and Patriotism in America's Classrooms
Yeah, I like standardized testing to. Other than that, this chapter is silly.
*Incidentally, reading this report was somewhat eye opening on numbers. I realized that some of what I had understood about welfare and teen pregnancy came from reports written in the 1980s. There is a high degree of cyclicality in these numbers and rates rose significantly in the 1980s, meaning that some of what I though was true based on these papers may not be (then again, those writing these reports may have had better data sets, the start of the series in 1940 is likely a trough without data going further back it's hard to say what a base rate would be). This couldn't be due to welfare since the 80s weren't great years for welfare increases but it does indicate rates were somewhat higher than I believed relative to early periods (rates also fell in the 60s and part of the 70s before rising again, total % of births decreased because births to married mothers declined more sharply, what this indicates is that planned pregnancies fell while unplanned remained stable or increased). Of course, 1940 is very likely to have had an unusually low rate of births of all kinds, given the war going on and the lingering effects of the depression. I'd like to have numbers going back to 1900 for better comparability.
[Update: This was interesting, pregnancy rates feel sharply among non-white women between 1990 and 1995 from 175 per 1000 to 152 in 1995 (women 18 - 44). Among all women, the decline was from 102 to 96. The abortion rate also fell sharply between 1980 and 1995. I have no real theoretical observations from any of this, other than I hadn't realized the high cyclicality of birth rates. Very interesting, and certainly something that doesn't seem to match with welfare state policies. Marriage rates also increased among women who conceived first child before marriage in 1990-1994 relative to 1980-84. Seriously contradicts the view about welfare reform having something to do with marriage, this seems to be varying independently.
No comments:
Post a Comment