* Proportion is significantly different from Primarily Research (for Carnegie Classification) or from Doctorate (for Highest Degree Offered) or from Public (for Form of Control) at the 5% level.
** Proportion is significantly different from all other disciplines combined at the 5% level. We used regression analysis for these tests with a binary (0-1) variable for the level of interest. If the coefficient for the binary variable differed significantly from 0, then the interpretation from regression is that the discipline differs from all other levels combined.
Statistical significance depends on a number of factors, not solely the absolute difference between two values. While differences that are not marked as significant may seem to be the same size as, or even larger than, those marked as significant, they are not statistically significant. The most likely factors contributing to the lack of significance when the absolute difference seems “large enough” are a smaller sample size or a larger variation within that discipline.
Source: Susan White, Raymond Chu, and Roman Czujko, The 2012–13 Survey of Humanities Departments at Four-Year Institutions (College Park, MD: Statistical Research Center, American Institute of Physics, 2014). Study conducted for the American Academy of Arts Sciences’ Humanities Indicators Project.