Category Archives: Uncategorized

“Business donations to judges’ campaigns often equal friendly rulings”

Leave a reply

McClatchy article on a new study, via Rick Hasen. I like this part of the news article:

As with legislative campaign contributions, though, the new analysis raises a vexing chicken-and-egg question about whether donations change voting behavior or simply reflect common interests between donor and recipient. Many judicial decisions, moreover, defy any connection to campaign contributions.

The study is authored by a law professor with an economics PhD, and was published by an advocacy organization.

Interesting exchange on measuring preferences for inequality

Leave a reply

These two pieces came across the Google-alert wire:

American’s desire for less wealth inequality does not depend on how you ask them — Michael I. Norton and Dan Ariely

The available evidence suggests the percent measure should not be used to study inequality: Reply to Norton and Ariely — Kimmo Eriksson and Brent Simpson

Both pieces from Judgment and Decision Making, Vol. 8, No. 3, May 2013.

The backstory: first Norton and Ariely published this; then Erikkson and Simpson published a critique.

I like how Eriksson and Simpson sum up their counter-counter-counter:

We sum up by noting that the original Norton and Ariely (2011) paper has received a very high level of mostly uncritical attention, probably because of the importance of the topic, the surprising findings, and the fact that both authors are well-known for doing excellent research. In this case, however, the available evidence suggests that their original conclusions are artifacts of an invalid measure. Our original paper was motivated by our concerns that policy recommendations might be founded on unsupported conclusions and that the paper might set a precedent for future research to employ similar invalid measures of perceived and desired inequality.

My read of the exchange is that the balance of evidence favors the critique. It doesn’t help that Ariely and Norton don’t address the major point of their interlocutors, namely that their measure of preferences for inequality is too computationally difficult for subjects to understand. Ariely and Norton, in turn, could have attacked Erikkson and Simpson for relying solely on a convenience sample — the original authors used Survey Sampling International, which is not a population-based sample but is nationally diverse.

 

Kahneman on why economics is the only social science that policy makers listen to

Leave a reply

You note in the foreword to the recently released The Behavioral Foundations of Public Policy that economists have a “monopoly” on policy making, that. “Like it or not, it is a fact of life that economics is the only social science that is generally recognized as relevant and useful by policy makers.” Why is that?

 

Policy makers, like most people, normally feel that they already know all the psychology and all the sociology they are likely to need for their decisions. I don’t think they are right, but that’s the way it is. On the other hand, people who have not studied economics are fully aware of their ignorance. The use of mathematics adds a touch of magic to economics. Indeed it makes perfect sense for economists to be the interpreters of policy-relevant research, because they understand and are trained to use big data. This, and the fact that policies always involve tradeoffs and almost always involve money, explains the dominant role of economics in policy.

Full article here (via here).

New free online course from Stanford: Statistics in Medicine

Leave a reply

Yesterday I received an e-mail inviting me to the following online course, offered through Stanford University’s online learning platform:

This course aims to provide a firm grounding in the foundations of probability and statistics. Specific topics include:

1. Describing data (types of data, data visualization, descriptive statistics)
2. Statistical inference (probability, probability distributions, sampling theory, hypothesis testing, confidence intervals, pitfalls of p-values)
3. Specific statistical tests (ttest, ANOVA, linear correlation, non-parametric tests, relative risks, Chi-square test, exact tests, linear regression, logistic regression, survival analysis; how to choose the right statistical test)

The course focuses on real examples from the medical literature and popular press. Each week starts with “teasers,” such as: Should I be worried about lead in lipstick? Should I play the lottery when the jackpot reaches half-a-billion dollars? Does eating red meat increase my risk of being in a traffic accident? We will work our way back from the news coverage to the original study and then to the underlying data. In the process, students will learn how to read, interpret, and critically evaluate the statistics in medical studies.

The course also prepares students to be able to analyze their own data, guiding them on how to choose the correct statistical test and how to avoid common statistical pitfalls. Optional modules cover advanced math topics and basic data analysis in R.

I think it sounds like a great way to teach introductory statistics, mainly because of the “teasers” idea.

Supreme Court makes democracy (and social science research) harder

Leave a reply

Last week, the U.S. Supreme Court unanimously ruled that the state of Virginia can refuse to entertain Freedom of Information Act requests from persons residing outside of Virginia. A New York Times article is here. I heard about the decision just by chance from turning on NPR during On the Media this weekend. But I knew the case was coming, and had paid attention to the arguments when they were heard in February.

The reason is that, as part of my dissertation research, I’m studying local elections in Virginia, and have been requesting that county election offices provide me with election results dating back to the early 1990s. A few counties outright refused to help me, citing the state law that says they don’t have to. The majority just did not respond. The state election office even decided to get in touch with me, apparently after some localities contacted them about my request, reminding me that their offices are under no obligation to help me. (Naturally, I tried to tailor my requests to be as polite as possible, and made clear that I was willing to reimburse local offices for any fees associated with my request.)

Of course I knew about this clause of the state law going into my request — probably I learned about it from the Virginia Coalition for Open Government web site, which was involved with the case and was on the side of the plaintiffs who argued the residency restriction was unconstitutional. All is not lost for me, anyway — I have since been in contact with a researcher in Virginia and we’re planning on completing the requests soon.

I have two comments about the On the Media report about the decision. First, I found it really strange that Justice Alito, writing for the Court, seems to believe that “Requiring noncitizens to conduct a few minutes’ of Internet research in lieu of using a relatively cumbersome state FOIA process cannot be said to impose any significant burden.” Assuming this is not horribly taken out of context, it’s just wrong — if government information was readily available on the Internet, then no one would be doing FOIA requests in the first place. Indeed, in my experience, the results of FOIA requests are often paper records — if you’re lucky, the person processing the request will scan them for you. So they often aren’t even digitized, let alone on a web site.

My second comment relates to a point raised by Mark Caramanica of Reporters Committee for Freedom of the Press. He points out that these type of residency requirements threaten evaluations of government policy — how well are different states doing on certain benchmarks:

For example, USA Today did a story where they compared how the No Child Left Behind initiative was impacting teacher behavior in a variety of jurisdictions and pulled records dealing whether teachers were disciplined for potentially coaching students and, you know, trying to cut corners on the standardized tests, and so forth, to boost up their scores. That was a case where, you know, you need access to Virginia records, obviously, to put Virginia in the mix.

ProPublica did a story where they investigated dialysis centers all across the country, their effectiveness and safety records, and so forth. To get that national perspective, you need Virginia. And, you know, it leaves a hole in the entire story and the complete picture, but it also shortchanges citizens of Virginia because they don’t get the benefit of knowing where their state stacks up on certain issues.

(Of course, this type of cross-state comparison is more commonly done by social scientists, including political scientists who study states but also, maybe even more often, by economists. While a large number of press organizations, including the New York Times, filed an amicus brief on the side of the plaintiffs in this case, did any social science organizations? Apparently not. SCOTUSblog has all the amica briefs for the case, and I don’t see any.)

Maybe it’s not surprising that the most insular branch, not especially friendly to sharing information in its own domain, decided this way. On the other hand, a famous Justice once spoke eloquently of the potential of states as generators of good public policy. With this decision, the Court has made this potential just a little bit harder to realize.

I want to be careful not to exaggerate. For one, already, workarounds are apparently in the works. For another, VA is apparently one of 8 states with such a clause. But I’ve found some states don’t specify residency — as a researcher, I worry that the next time I make a request to a state with an ambiguous law, they might decide to refuse me on residency grounds.

Evaluating government programs

Leave a reply

Via the Freakonomics blog, this Washington Post article discusses the problem of evaluating government programs:

The case of Even Start stands out because it is so rare. At a time when the federal budget is increasingly squeezed — and lawmakers are wrestling with tough choices on what to cut or to keep — the government does very, very little to find out which programs produce the best results for the money spent on them.

There are several reasons for that, most wrapped up in politics. There’s no natural ideological constituency for program evaluations. Lawmakers who champion social programs often fear that attempts to measure them will be only thinly disguised excuses to kill the programs. Fiscal hawks don’t often love the idea of spending more money on evaluations.

The frame of the article is that recently some economists are calling for a more institutionalized process for evaluating government programs. One proposal is to start by including evaluations in decisions for federal transfers to state and local governments:

[Harvard economist and former Obama budget official] Jeffrey Liebman’s most specific proposal would change the flow of federal money to state and local governments. He would require that 1 percent of such grants be set aside for programs whose effectiveness has been proven through randomized or other rigorous research methods. Over time, the requirement would rise to 5 percent.

Given there is “no natural ideological constituency” for this sort of thing, how do the economists expect policy makers to get on board? Apparently there are some assumptions about the wisdom of voters and the electoral incentive.

But if Washington ever hopes to provide the services voters say they want, at the tax rates voters say they’re willing to pay, economists say the government will need to ramp up its efforts to figure out which programs work and which ones don’t, and shift resources accordingly.

[...]

That’s especially bad at a time when Washington is debating tax increases and spending cuts to reduce the federal deficit, said Jeffrey Liebman, a Harvard economist and also a former Obama budget official. “It’s imperative to be able to show that the things [voters’] tax dollars are being spent on work, and that we’re trying to improve performance and do it in a data-driven way,” Liebman said. “That’s just good stewardship.”

“Does International Child Sponsorship Work? A Six-Country Study of Impacts on Adult Life Outcomes”

Leave a reply

New paper at the Journal of Political Economy:

Child sponsorship is a leading form of direct aid from wealthy country households to children in developing countries. Over 9 million children are supported through international sponsorship organizations. Using data from six countries, we estimate impacts on several outcomes from sponsorship through Compassion International, a leading child sponsorship organization. To identify program effects, we utilize an age-eligibility rule implemented when programs began in new villages. We find large, statistically significant impacts on years of schooling; primary, secondary, and tertiary school completion; and the probability and quality of employment. Early evidence suggests that these impacts are due, in part, to increases in children’s aspirations.

Appear to be a few ungated copies floating around the web. Science Daily claims that this is the first study to show that such programs actually work:

Despite the billions of dollars that flow to child sponsorship each year and the millions of American families who sponsor overseas children, this is the first published study to investigate whether such programs actually benefit the children they intend to help. Evidence from the study points to the positive effects of child sponsorship on the adult life outcomes of these children.

Do small businesses (really) create most jobs?

Leave a reply

Something I’ve wondered about for a while, given that it’s a staple of political rhetoric. Now a forthcoming paper in the Review of Economics and Statistics looks at this issue systematically:

The view that small businesses create the most jobs remains appealing to policymakers and small business advocates. Using data from the Census Bureau Business Dynamics Statistics and Longitudinal Business Database, we explore the many issues at the core of this ongoing debate. We find that the relationship between firm size and employment growth is sensitive to these issues. However, our main finding is that once we control for firm age, there is no systematic relationship between firm size and growth. Our findings highlight the important role of business start-ups and young businesses in U.S. job creation.

The authors story is apparently a straightforward spurious relationship: “Importantly, because new firms tend to be small, the finding of a systematic inverse relationship between firm size and net growth rates in prior analyses is entirely attributable to most new firms being classified in small size classes” (first page of the unnumbered PDF at the link).

Does that invalidate the “descriptive” claim that small businesses create more jobs? For a similar issue(?), see here.

The study is apparently gated, but NBER did a press release about the study a few years ago, suggesting an ungated copy may be out there.

But apparently there’s been lots of popular commentary about whether this claim is true as well.

More on policy advice in the presence of political biases

Leave a reply

Via a Google Scholar Alert for “political equality”, I just came across this review article by UCLA Political Scientist Ronald Rogowski which is quite consistent of my recent post on policy advice. Here is the abstract:

Political science produces highly policy-relevant research, but politicians ignore it in favor of their own (or their supporters’) biases. I give examples from such fields as anti-immigrant politics, political business cycles and the politics of redistribution. The sole area in which politicians do attend closely to scholarly research is where it assists their own efforts at electoral success (e.g. effect and duration of political advertising). But politicians equally ignore the expertise of climatologists, physicists, biologists, economists and even spies, where that expertise contradicts their own preferred policies. All of this points more to a problem of democratic politics than of political (or any other) science.

Unfortunately the full article is gated and I can’t locate an ungated copy. I just can’t resist also posting the first paragraph:

Contemporary political science suffers from too much policy relevance, not too little. Politicians simply do not like the policies that scholarly research supports, prefer policies (often put forward by charlatans) that better suit their interests, and seek to suppress or ignore evidence-based research that contradicts their own, or their ‘base’ voters’, ideologies. When these same politicians assert piously that political science offers no policy- relevant research, what they really mean is that it offers no research that supports their own biases. Politicians accept research from political science, as I shall argue below, only when it assists their own efforts at re-election.

I think this point — “When these same politicians assert piously that political science offers no policy- relevant research, what they really mean is that it offers no research that supports their own biases” — may also extend to many political scientists who complain about a lack of policy relevance. There, though, I think there is also a failure of communication — like not understanding one another’s methods or motivations, so failing to see the relevance — as well as a difference in political agendas.