ABSTRACT
The second approach is generally preferable. Not only does it allow more powerful statistical methods to be used, it retains the information provided by the finer grained measurement of interval-level variables. The first part of this chapter focuses on techniques that do not require that we collapse the values of interval-level variables.
