Descriptive statistics were calculated as means (eg, average amount per award) or as summations (eg, total amount per award). Data and results from the analyses are given for the 20 COMs that completed the 2004 survey plus an additional 6 that did not exist in 2004, thus giving a sample size for 2009 of 26. Descriptive statistics for these schools are presented as mean (standard deviation) with variable ranges and summations, if applicable. The distributions of skewed 2009 variables were normalized by means of log or square root transformations. Bivariate relationships among continuous variables were examined using Pearson product moment correlation. Point biserial correlation was used for school type (a dichotomous variable) and the continuous variables of interest. To guard against an experimentwise error rate associated with the bivariate correlation procedures, a Bonferroni correction for multiple tests was made. Therefore, the level of statistical significance was set at .007 (0.05 ÷ 7 tests) for bivariate correlations. Two multiple linear regression models were constructed to explore predictors of the total amounts (model 1) and number (model 2) of awards active in 2009. The independent variables were school type (0=private, 1=public), the number of faculty, and research expenditures in dollars; these variables were found to be related to the total amounts and number of awards in the bivariate analysis. The level of statistical significance was set a priori at .05 for the multiple regression analyses, and all analyses were conducted using SPSS statistical software version 20.0 (SPSS for Windows, SPSS Inc, Chicago, Illinois).