When I choose which statistical analysis to do, I am expressing my data with a bias, either to tell a story (or, in more negative terms, to push an agenda). The story is built into the language and assumptions associated with each statistical analysis. prediction through multiple regression? anova? odds ratios? They all give similar results on the same data but with very different implications and responses from the people who are receiving the results. I express emotion just fine through words. I also express emotion through numbers.