Canonical solution for demographics predicting attitudes toward welfare for canonical functions 1–5.

The attitude variate and demographic variate of each function are “latent variables”. The canonical correlation Rc is the correlation between the two variates, and the percentage of variance is presented as R2c in the middle row of the table. The structure coefficients (rs, 3rd column within each function) represent the Pearson correlation between an item and it associated canonical variate as determined by factorization of the correlation matrix (e.g., the correlation of Disagree cuts damage lives with the attitudinal variate in the first function is -0.700). Structure coefficients greater than |.35| are underlined, and those greater than |.70| are double underlined. Each individual’s variate scores are calculated by summing the product of the structure coefficient and observed value for each item in the variate (i.e., for attitudes, the items from “Disagree cuts damage lives” to “Disagree single parents deserve”; for demographics, the items from “Bachelor degree plus” to “exposure to benefits”). The correlation of each item with the alternative variate (e.g., “Disagree cuts damage lives” with the demographic variate) is represented in the column of canonical cross loadings (cros., 2nd column within each function). Also presented are the standardized canonical function coefficients (coef., 1st column within each function). These are the standardized β coefficients from simultaneously regressing each item in the variate on the variate itself, and as such can be thought of as adjusted structure coefficients. This adjustment process accounts for their typically smaller size relative to the structure coefficients and their sometimes divergent directions. As the reader moves through the table, the function changes. Each function has its own attitude and demographic variate, and across functions the variates are orthogonal. It is thus useful to know how much of the original items variance is represented by the reported canonical functions, that is, each variables communality (h2 reported as a %, final column of table). The communality is calculated by taking the sum of squared structure coefficients (rs) across the 5 reported functions. Communality coefficients greater than |35.00| are underlined, and those greater than |70.00| are double underlined. The bottom section of the table presents the correlation of the two extracted attitude components from the principle component analysis with the demographic variate of the canonical function.Canonical solution for demographics predicting attitudes toward welfare for canonical functions 1–5.