originally posted by Rahsaan:
originally posted by Jonathan Loesberg:
Suppose instead of listing 90 after a score, David listed XC. And suppose he meant by that that he gave the wine "Xavier Cougat," which for him meant "very damned fine." And suppose he gave another wine LXXXV, which for him meant Lawrence Xavier Xavier Xavier Victor, which meant "I like this better than Roman numerals can express." After you had corrected your original impression that those Roman numeral looking things actually were Roman numerals, what would modeling the variables mean that would add to the meaning of the phrases?
One technique (if you're feeling Googley) is multinomial logistic regression, which analyzes the probability of results falling in specific categories (e.g. Xavier Cougat vs. LXXXV and so on).
The standard econ example of this would be statistical analysis of whether commuters use cars, buses, trains, bicycles, etc. In political science we use this to analyze vote choice across multiple candidates. You can't classify those choices on an ordinal scale but you can determine the factors that lead people to choose the different forms of transportation or make their vote choice.
And yes, you can test for whether or not the Xavier Cougat categories are used consistently.