News and Media
In 2005, the Iranian Studies Group at MIT undertook a comprehensive survey of Iranian-American community characteristics to better understand what it means to be Iranian-American. The survey included over 40 questions and 3880 individuals filled it online. The initial phase of the reporting of the results focused on the basic characteristics and statistics regarding this survey. In this paper we report some of the early results from partial regression analysis on some of the interesting themes in the survey.
While online surveys typically tap into a selective sample of the population and therefore their aggregate statistics (e.g. average age, education, etc) do not necessarily represent the characteristics of the community as a whole, conducting regression analysis on survey results often yields more reliable insights. For example, 60 percent of the respondents to an online survey may have a bachelors degree and 30 percent to earn over 100000 $/year, even though the underlying population on which the survey is conducted has only 30 percent bachelor holders and 20% of that population earns over 100000 $/year. However, if we conduct a regression to find the determinants of income in this population it is very likely that we get the same determinants of the income (e.g. education, age, and so on), because many of the same variables influence the income regardless of which social group one belongs to.