Tuesday, February 12, 2008

Improving Socio-Economic Status

This is a response to Ken DeRosa in D-Ed Reckoning, in which he argues, "Let's stop wasting time with these misguided schemes (to improve learning by alleviating poverty) and focus our efforts elsewhere."

As a foundation for public policy, the research in this post is surprisingly slim (and in places dubious) while the argumentation is not especially tight.

DeRosa asks, essentially, "can student achievement be improved by artificially increasing the child's SES?"

I have no idea what 'artifically' means in this context, since I have no idea what a 'natural' or 'real' SES would be for a given person. All SESs are artificial constructs: the exist as a result of the financial and other allocations provided to them by society.

The proposition being advanced in his post is, essentially, that improvements in a child's SES will not result in increased educational outcome. That is the only way he can conclude "It's all well and good to attempt to ameliorate the plight of the poor.... Just, don't expect that it's going to improve student achievement or improve real SES in the long run across generations."

In particular, DeRosa appears to be opposing a particular class of improvements of SES: "to increase the family income of poor families." The rest is left to "hope".

It is unreasonable to suppose that such a plan would work, and to my knowledge, anti-poverty advocates do not support such an approach. The effects of poverty are persistent. Throwing money at them after the fact and then saying that money fails to fix the problem is like steering after the Titanic has hit the iceberg and then saying steering makes no difference.

Poverty agencies all know that financial support is a necessary, but not a sufficient, condition for the alleviation of poverty. What conditions are also required will be discussed below. But it should be clear that the failure of money alone to solve the problems does not mean that the problems are not caused, at least in part, by a lack of money.

DeRosa continues, "At least that's the theory. We've been testing this theory for forty years now by providing massive injections of financial assistance to the poor. The gains in academic achievement, however, have proven to be elusive."

The use of the editorial 'we' here is misleading. In many jurisdictions, the gains have been impressive. In countries like Canada, Finland, Denmark, and others, something approaching economic equality has been achieved. And the educational consequences, to judge by the PISA test results, have been impressive.

The "massive injections of financial assistance" to the poor in the United states have obviously been insufficient. In this society, the additional measures appear not to have been undertaken. One wonders about housing standards, health care, and educational services.

It is also possible that the amount of money spent is, in fact, itself insufficient. DeRosa may argue that "this is a ridiculous argument," however, one would presume that there is a minimal cost to educating all children in a society, and if that cost is not met, then some children will not be educated. This argument is not ridiculous at all - it is merely not one that may be resolved easily or conclusively in a short discussion.

But all of these considerations aside: DeRosa's main argument is that differences in educational outcomes are due to genetic factors, and therefore money spent to change environmental factors is money wasted. At least, that's what the introduction of the Minnesota twin studies suggests. And thus he says, "The results have been consistent. About three quarters of the variance in IQ and student achievement is attributable to genetic factors. While the variance attributable to familial factors is about zero."

This result is subject to numerous criticisms:

First, it is not clear that IQ tests are a reliable measurement of educational outcomes. The tests are intended to identify innate, or native, intelligence, not actual learning achieved. Thus, it is not surprising that most of the findings would be explained by innate or native factors.

Second, the reasoning behind the attribution of genetics as cause is flawed. As summarized in the Wikipedia page we are linked to, "the similarities between twins are due to genes, not environment, since the differences between twins reared apart must be due totally to the environment." This is a non sequiter. There are numerous possible causes of similarity other than genetic factors.

Third, even the studies alluded to indicate that genetics play only a minority role. The Harris paper cited, for example, states that "Heritability generally accounts for 40% to 50% of the variance in personality characteristics." (p. 459)

Fourth, the examples provided aren't even looking for differences in learning outcomes. As the Wikipedia article summarizes, "Of interest to researchers are prevalence of psychopathology, substance abuse, divorce, leadership, and other traits and behaviors related to mental and physical health, relationships, and religiosity."

Fifth, the argument equivocates between types of influence. We began, above, talking about socio-economic status (SES). But these studies purport to show, show, as DeRosa states, "It was found that the contributions to the correlation between twins in g by... all dimensions of the Family Environmental Scale... were all zero to within two decimal places."

Of course, even that is a ridiculous conclusion, and DeRosa is quick to admit it: "Which is not to say that abusive parents and ghetto life aren't going to have a detrimental effect." Of course they are.

The more likely explanation for the experimental results in these very small and very localized studies is that the families were not different in any way that mattered. And in particular, none of the separated twins was raised as a malnourished poor black kid in the ghetto (for one thing, experimental ethics would have prohibited it, as it would have in such conditions constitute abuse of the experimental subject).

Sixth, the apparently 'genetic' differences very likely have other causes. The data cited from the Minnesota Transracial Adoption Study is purported to show that placing poor black kids into the homes of rich white people didn't change their educational outcomes (as misleadingly measured via IQ tests).

But what we know for certain is that none of these poor black kids was born in the white family. The child's entire prenatal history - including any possibility for malnutrition, cigarette smoking, drug abuse, pollution, and a variety of other environmental factors, may have played into the child's educational potential. For some kids, the ship hits the iceberg before they are even born. But this doesn't mean that the deficiencies are genetic. It just means that no amount of money after birth will alleviate the impact of poverty before birth.

DeRosa concludes "that low-SES does not cause or or significantly contribute to low student achievement. Further, student achievement will not be significantly improved by trying to artificially increase a child's SES."

This is simply not supported by the evidence he offers. At best, what he's shown is that some of the damage caused by poverty is permanent, and that other parts of the damage caused by poverty stem from the environment outside the home, and not conditions inside the home. But poverty advocates know that as well, which is why they generally resist 'blame-the-parent' programs.

He also concludes, "genetics is a large fly in the ointment that has a more significant effect on outcomes than the environmental factors in any event." This again is not shown.

Again - as I have argued before - these outcomes have complex causes, and these complex causes are often misrepresented in simple variable-effect surveys. Let me illustrate with one example.

A major component in my own education was access to classic works of literature as a child. But in order for me to benefit from these books, two concurrent things must take place: first, the books must be present in the home, and second, my cultural environment must favour reading.

Now suppose there are no books in the home, but the environment contains plenty of books - there's a school library, say. Then, given a cultural environment that favours reading, I get the same benefit.

Now from this it looks like it doesn't matter whether or not the parents have books in the home. And this is the outcome of the twin studies. But now suppose there are no books in the home and no books in the environment. Even if the culture supports reading, I cannot benefit. But I *would* have benefited had the resources been available. Unfortunately, this counterfactual is never measured.

Poverty creates ripple effects that bounce back and forth through a child's life. The child may have to deal with prenatal or infant malnourishment, substandard living conditions and health conditions, poor access to resources both in home and at school, community attitudes that enforce a norm by discouraging achievement, chronic health and social issues caused by pollution, crime, and other factors, systemic discrinination based on race, appearance, accent, and other factors.

DeRosa wraps up, "It's all well and good to attempt to ameliororate the plight of the poor. We do quite a bit already, perhaps too much."

The evidence doesn't bear that out. The evidence suggests that much more effort is made to find 'magic bullet' solutions - like small schools, phonics, charter schools, whatever - anything, everything EXCEPT to acknowledge the role of poverty in an unequal society in educational outcomes.

Instead of trying so hard to make a child's poverty a non-issue in their education, let me suggest a more productive, research-based, and enlightened strategy: feed them.

Then, maybe, if there's any money left over, given them food for thought - access to reading and learning material, tools to manipulate and create with, a space for them to be themselves, and an environment that values learning, creativity, and achievement.

Oh, that's not a magic pill solution either. It's still only part of the solution. The rest of the solution involves the broader initiatives we see in the countries that score well in the PISA evaluations - government interest and investment in education, public or affordable health care, broad equality of income (whether government mandated or won through union action), enforcement of housing and other health and cleanliness standards, and positive and accessible role models in media. To name a few.

4 comments:

  1. Nice post, Stephen, but I have a few disagreements.

    All SESs are artificial constructs: they exist as a result of the financial and other allocations provided to them by society.

    If but this were true, you’d have a firmer leg to stand on. Except perhaps at the extremes, most income is earned, usually by supplying a service or knowledge, the price of which is generally determined by the scarcity of the supply and the amount demanded.

    In particular, you appear to be opposing a particular class of improvements of SES: "to increase the family income of poor families." The rest is left to "hope".

    I don’t oppose these improvements ; what I’m saying is that there is no or little evidence that such improvements will be successful. And, at least some evidence demonstrating that doing this hasn’t worked under some reasonably controlled conditions.

    But it should be clear that the failure of money alone to solve the problems does not mean that the problems are not caused, at least in part, by a lack of money.

    You are forgetting, perhaps conveniently, the large amount of non-cash benefits that are provided which amount to far more than the cash benefits. In fact, most of these benefits already address the underlying problems you allude to supra. And let’s also be clear that you have no evidence supporting this casual theory. It remains an opinion and I’d go so far to say an unsubstantiated opinion.

    You continue, "At least that's the theory. We've been testing this theory for forty years now by providing massive injections of financial assistance to the poor. The gains in academic achievement, however, have proven to be elusive."

    In many jurisdictions, the gains have been impressive. In countries like Canada, Finland, Denmark, and others, something approaching economic equality has been achieved. And the educational consequences, to judge by the PISA test results, have been impressive.

    Do you have a cite for these PISA gains? And a definition for impressive? Cross country comparisons of poverty are problematic. There might be more economic equality in Canada, Finland, Denmark but there is far more income in the U.S, a third more on average. And income, for the poor in U.S. excludes all non-cash benefits, all after-tax benefits, the income from cohabitors providing a skewed picture of the real income of the poor.

    The "massive injections of financial assistance" to the poor in the United states have obviously been insufficient.

    You’re basing this conclusion on what? The failure to achieve results? That seems circular to me. Is there some country that provides more than the U.S.? That is doubtful. Was there success? Is there reason to believe that success can be replicated in the U.S? Are the conditions the same or similar? Are there any cites?

    One wonders about housing standards, health care, and educational services.

    Why don’t we stop wondering and start providing some support for these theories. I see at Marginal Revolution you’ve been citing the 50 million people don’t have health insurance meme (actually it’s 47 million), but those stats have been discredited. The real number is below 10 million.

    however, one would presume that there is a minimal cost to educating all children in a society, and if that cost is not met, then some children will not be educated.

    The correlation between education spending (at today’s levels) and student achievement is virtual zero. Ssshhhhh. Don’t tell anyone, but urban schools are often some of the best funded schools in the state.

    your main argument is that differences in educational outcomes are due to genetic factors, and therefore money spent to change environmental factors is money wasted.

    That’s not what I’m saying I said “About three quarters of the variance in IQ and student achievement is attributable to genetic factors. While the variance attributable to familial factors is about zero.” If ¾ of the variance is attributable to genetic factors, then ¼ are attributable to non-genetic, or environmental factors, but not necessarily factors related to SES.

    First, it is not clear that IQ tests are a reliable measurement of educational outcomes.

    It’s more clear than you think. (Unrelated children reared together (N=156) in 71 different families included in the Texas Adoption Project were compared for similarities on intelligence and achievement tests. The purpose was to see if a distinction between the two types of tests based on their heritabilities could be sustained. Results indicated no substantial differences in correlations for the two types of tests, and hence little or no support for the notion of an aptitude-achievement distinction based on differential heritabilities.)The aptitude-achievement test distinction: A study of unrelated children reared together.

    Furthermore, the studies I cited conducted all sorts of academic achievement tests, not just IQ tests. Look at the other columns in the table, the last one looks like it’s getting cut off though.

    There are numerous possible causes of similarity other than genetic factors.

    Like? The studies didn’t find that 100% of the variation was due to genetics, but .70. Some studies have found slightly more; some slightly less.

    Third, even the studies alluded to indicate that genetics play only a minority role. The Harris paper cited, for example, states that "Heritability generally accounts for 40% to 50% of the variance in personality characteristics." (p. 459)

    Even this, which is one of the lower estimates, is a higher correlation than the SES one. I sense some confirmation bias.

    Fourth, the examples provided aren't even looking for differences in learning outcomes. As the Wikipedia article summarizes, "Of interest to researchers are prevalence of psychopathology, substance abuse, divorce, leadership, and other traits and behaviors related to mental and physical health, relationships, and religiosity."

    There were many things researched in these studies and many articles were generated. Some are relevant to the present discussion, others not.

    Of course, even that is a ridiculous conclusion, and you are quick to admit it: "Which is not to say that abusive parents and ghetto life aren't going to have a detrimental effect." Of course they are.

    There are other environmental factors than those measured on the Family Environmental Scale. Some relevant environmental that do matter are lead poisoning and severe malnutrition. Two factors not commonly found among the poor in the U.S.

    And in particular, none of the separated twins was raised as a malnourished poor black kid in the ghetto (for one thing, experimental ethics would have prohibited it, as it would have in such conditions constitute abuse of the experimental subject).

    First of all, the poor are far more likely to be supernutriated than malnourished. You’ve been reading too much Dickens. Second, the studies show that at least for the children with two black biological parents, the measured outcomes were closer to the typical performance of a low-SES environment than the high-SES white environment that the child grew up in.

    But what we know for certain is that none of these poor black kids was *born* in the white family. The child's entire prenatal history - including any possibility for malnutrition, cigarette smoking, drug abuse, pollution, and a variety of other environmental factors, may have played into the child's educational potential.

    This sounds like a fighting retreat to me. Certainly the lower incidence of breast-feeding among black mothers plays a role. This argument, however, does not bear close scrutiny since black IQ in the top decile of SES remains below white IQ in the lowest decile.

    You conclude "that low-SES does not cause or significantly contribute to low student achievement. Further, student achievement will not be significantly improved by trying to artificially increase a child's SES."

    But in order for me to benefit from these books, two concurrent things must take place: first, the books must be present in the home, and second, my cultural environment must favour reading.


    You are leaving out one critical condition: that you successfully learned not only how to read, but also to become an expert reader. For most people this requires positive instruction and much practice. I assure you that you would not have benefitted from these books without this.

    Now suppose there are no books in the home, but the environment contains plenty of books - there's a school library, say. Then, given a cultural environment that favours reading, I get the same benefit.

    Only if you learned how to read in the first place.

    Now from this it looks like it doesn't matter whether or not the parents have books in the home. And this is the outcome of the twin studies. But now suppose there are no books in the home and no books in the environment. Even if the culture supports reading, I cannot benefit. But I *would* have benefited had the resources been available. Unfortunately, this counterfactual is never measured.

    Lots of counterfactuals weren’t measured or tested. The question remains is this a relevant counterfactual. I’d say no. The subjects in these studies did have both the culture and the books available and still performed as if they didn’t. I would like to see the putative school that doesn’t have any books available for the students to read.

    Poverty creates ripple effects that bounce back and forth through a child's life. The child may have to deal with prenatal or infant malnourishment, substandard living conditions and health conditions, poor access to resources both in home and at school, community attitudes that enforce a norm by discouraging achievement, chronic health and social issues caused by pollution, crime, and other factors, systemic discrimination based on race, appearance, accent, and other factors.

    And yet when we conducted reasonably controlled experiments that eliminate most, if not all of these variables (your speculation notwithstanding), student outcomes failed to improve according to your hypothesis. I understand your reluctance to discard the hypothesis and your desire to find potential confounders, but your lack of supporting evidence and the consistency of the findings that support the genetic theory is increasingly marginalizing your position. I’m sure you’ll keep up the good fight, but for what purpose?

    The evidence doesn't bear that out. The evidence suggests that much more effort is made to find 'magic bullet' solutions … everything EXCEPT to acknowledge the role of poverty in an unequal society in educational outcomes.

    I’d say that much more resources are presently devoted to reducing poverty than the magic bullets you list. I also think that acknowledgement needs to be earned.

    let me suggest a more productive, research-based, and enlightened strategy: feed them.

    As I stated above, America’s poor are far more likely to be supernutriated and obese than undernourished, much less malnourished. Here are some 2002 stats relating to non-cash benefits related to food: food stamps (about $2,200 per family), the school lunch and breakfast programs (as much as $600 per child), and the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) (about $400 per person).

    The rest of the solution involves the broader initiatives we see in the countries that score well in the PISA evaluations

    About the only thing PISA results tells us is that you are better off academically if you are northeast Asian or white.

    I've read te PISA report, it is mostly opinion based on dubious correlations extrapolated to extremes. It is an embarrassing for PISA and you to pawn this off as legitimate science.

    ReplyDelete
  2. The legitimate science bit got me intrigued. It took 'half-an-hour' just to read the rest.

    Whether SL is 'real'could be a scientific study in itself. Most importantly we await your post Stephen that reveals your alter ego's in SL and the games that your playing out that we are not yet told about.

    Surely you cant only have one Facebook login ?

    ReplyDelete
  3. ps. it may appear off topic but the question/s are about the very identities we propound and hold ourselves out to "talk" through.

    Academic achievement has done as much to support poverty as it has to lift people out of it.

    An educated mass knows themselves as society. They have the research to prove that they exist and that they are making a substantial difference.

    Likewise the OLPC will ensure that we have enough landmarks in 10,000 years to come to prove that humans really did enjoy "green" plastic.

    It's all ciphers and signs. Poverty and expectancy of bandwidth are not a popular juxtaposition.

    ReplyDelete
  4. My Second Life identity is Labatt Pawpaw, but I don't log on very often.

    I have only one Facebook account.

    There's nothing mysterious happening behind the scenes - what's happening in plain view is more than enough to occupy me.

    ReplyDelete

I welcome your comments - I'm really sorry about the moderation, but Google's filters are basically ineffective.