ABSTRACT
However, as we noted, single-parameter models are only infrequently of substantive or theoretical interest. In many ways, the example from the last chapter, where we wanted to test the hypothesis that people were willing to pay the expected value of $50, is unusual in the behavioral sciences. More frequently such a priori values do not exist, and instead we may be asking whether the mean in one group of respondents (e.g., those who were trained in the meaning of expected values) differs from the mean in another group of respondents (e.g., those who received no such training). Or, returning to the data on internet access, while it is certainly possible that we would be interested in testing whether some a priori percentage (e.g., 44%) is a good estimate of mean internet access, it is much more likely that we would be interested in examining the determinants or correlates of internet access rates. In other words, our interest is more likely to center on attempts to explain the internet access data than on tests of alternative values for the mean access rate.
