In the previous chapter the model describing how judges reach sentencing decisions in cases comprising multiple disparate and separate counts was used to derive hypotheses so that it was possible to test this model by means of a series of sentencing problems. This chapter describes how the data were collected for this purpose. Three sets of data were used, each set generated by a different technique, although the sentencing problems provided a common framework within which these three techniques were applied. The three techniques were: (1) sentences imposed by a sample of judges for the cases in the sentencing problems, to investigate which aspects of the model were consistent with the judges' sentencing behavior and which were not; (2) verbal protocols based on immediate retrospective reports and providing direct, detailed and accurate information on the content and sequence of the judges' thoughts as they worked through the sentencing problems; and (3) reflective retrospective reports in which the judges were asked to comment on whether the various elements of the model were accepted by them as valid descriptions of the method to be adopted when sentencing the multiple offender. For these three techniques of data collection, the judges in the sample were tested individually.

Each of the three techniques is now described, justified, and its place in the testing of the model explained, in turn. Following this, the procedural details relating to the administration of these techniques are described.