Economists Donohue and Levitt (2001) claim that legalizing abortions explains the drop in crime rates, reasoning that legalization allowed women to avoid childbirth until they were in a position to care for their children. The underlying assumption is that unwanted babies growing up with mothers who are not ready to care for them are likely to become violent criminals. This assumption, however, is just an assumption; the authors do not provide evidence to support it. The political spin here is that if there is a relationship between abortions and crime reduction, those who support a woman’s right to choose might argue that restricting access to abortions would have an unintended consequence of increasing future crime rates. These social scientists provide evidence based on research to back up their theory, using multiple regression in a display of statistical wizardry. The violent crime rate is the dependent variable. Looking at state data over time, they constructed a model that includes the canadian pharmacy email address effective abortion rate, along with prisoners per capita, police per capita, percent unemployed, income per capita, poverty rate, welfare generosity, the right to carry concealed weapons law, and gallons of beer consumption per capita as the independent variables. This regression model has more variables than the other examples presented in this chapter but it gets interpreted the same way. The key information is an R-squared of .938 between the effective abortion rate and the violent franchise viagra crime rate, and an R-squared of .942 for the model with all the variables. An R-squared of 0.9 seems to explain almost all the variation in crime rates. However, it is very rare to see such a high R-squared , so I am immediately suspicious. Certainly some of the independent variables in the model are likely to be highly correlated, such as income and poverty rate. That correlation will throw off regression models, which work best when the independent variables are not correlated with each other. It is also possible to get a high R-squared by eliminating extreme scores in the data, which by definition will reduce the variance. For some reason that is not clear to me, there is a ridiculously high correlation between the effective abortion rate and the violent crime rate. In an explanation that is difficult for me to fully understand, the authors describe the effective abortion rate as a viagra for dogs calculated figure that includes an estimate of the number of arrestees. Hmmm—there might be viagra kullanım şekli a problem here in including arrest data because arrest rates are likely to be highly correlated with crime rates. In order to determine the credibility of this study, I would have to dig a whole lot deeper. However, despite my skepticism, it does not mean that the researchers have not done a good job. They might indeed have found a relationship between legalization of abortions and crime rates. I need more than my skeptism to dismiss max dose of cialis its results. I need solid evidence to refute this study. I also need to be mindful that some of my skepticism might be because the study challenges some personal beliefs. This awareness is a warning that I might not be able to make an objective assessment. What else might explain the crime rate reduction? Could it be the reduction in lead poisoning? That is what social scientist Rick Nevin thinks (Vedantam 2007). Lead is toxic to the brain and lead poisoning is associated with aggression. Prior research revealed that geographic areas with high levels of lead had substantially higher murder rates than areas with lower levels, after controlling for socioeconomic and environmental factors. Nevin tracked data looking at crime rates and lead poisoning in nine countries. Because other countries introduced policies to reduce lead exposure at different times, Nevin was able to do a comparative time series analysis using a regression technique. “Sixty-five to ninety percent or more of the substantial variation in violent crime in all these countries was explained by lead,” he states. He is talking about R-squared here. He noted that lead poisoning is not the only factor but it is the strongest one according to his research. Again, there is much statistical wizardry here with a very high R-squared, which is a signal for closer inspection if a policy-maker is being pressed to make a decision. The difficulty in explaining the drop in violent crime is not unique. The policy arena is cluttered with competing theories about why something happened, some with strong analysis and some not. Some explanations may be designed to serve different political agendas, but truthfully, it is just plain hard to determine causality in complex social issues even with very advanced statistical techniques and the intention to be objective. From Research Methods for Public Administrators, 2nd edition.