Correlation and causation, revisited

I'm dumbfounded at the stupidity of a recent statistical analysis recently reported in a press release from the Association for Psychological Science.

The way some people talk, you’d think that a flat tax system—in which everyone pays at the same rate regardless of income—would make citizens feel better than more progressive taxation, where wealthier people are taxed at higher rates. Indeed, the U.S. has been diminishing progressivity of its tax structure for decades.

But a new study comparing 54 nations found that flattening the tax risks flattening social wellbeing as well. “The more progressive the tax policy is, the happier the citizens are,” says University of Virginia psychologist Shigehiro Oishi, summarizing the findings ...

. . .

The researchers analyzed the relationship between tax progressivity and personal well-being in 54 nations surveyed by the Gallup Organization in 2007—a total of 59,634 respondents. Well-being was expressed in people’s assessments of their overall life quality, from “worst” to “best possible life,” on a scale of 1 to 10; and in whether they enjoyed positive daily experiences (such as smiling, being treated with respect, and eating good food) or suffered negative ones, including sadness, worry, and shame. Finally, the analysis looked at the participants’ satisfaction with their nation’s public goods, from schools to clean air.

The degree of progressivity was measured by the difference between the highest and lowest tax rates, corrected for such confounding factors as family size, social security taxes paid, and tax benefits received by individuals.

The results: On average, residents of the nations with the most progressive taxation evaluated their own lives as closer to “the best possible.” They also reported having more satisfying experiences and fewer discomfiting ones than respondents living in nations with less progressive taxes. That happiness, Oishi says, was “explained by a greater degree of satisfaction with the public goods, such as housing, education, and public transportation.”

. . .

The study, like others Oishi has done looking at connections between economics and personal life, has important social implications. “If the goal of societies is to make citizens happy, tax policy matters,” he says. “Certain policies, like tax progressivity, seem to be more conducive to the happiness of the people.”

There's more at the link.

Three things jump right out at any informed reader:

1. The factors rated by respondents were subjectively measured, rather than objectively, thereby making this study less than 'scientific' and rather more 'intuitive';
2. The greater (subjective) happiness expressed by respondents in 'nations with the most progressive taxation' is ascribed to that taxation by the researcher, without any objective evidence that the two factors are related in any way, shape or form;
3. The researcher proceeds to draw conclusions based on his findings: but since the findings are not objective and are not scientific, they clearly can't be used to justify 'scientific' recommendations for action!

This is a classic example of the 'correlation versus causation' problem in data analysis. The pre-existing political, social and economic beliefs and biases of the researcher(s) clearly influenced this 'scientific' study from the outset, rendering it meaningless from a logical, rational, evidential point of view. If this is an example of the 'science' delivered and endorsed by the Association for Psychological Science, I can only suggest they begin marketing their services to (and to the customers of) purveyors of homeopathy, astrology and palmistry. The credibility of all four 'sciences' will thus be operating at an equal level - along with the credulity of their patients customers!





Peter