Using Hierarchical Linear Modeling for Proformative Evaluation: A Case Example

Abstract

Proformative evaluation—first introduced in Scriven’s (2006) The great enigma: An evaluation design puzzle—“is motivated, like formative evaluation, by the intention to improve something that is still developing, but unlike formative, the improvement is only possible by taking action, hence proactive instead of reactive, hence both, hence proformative” (M. Scriven, personal communication, March 9, 2006). An exploratory study was conducted to proformatively assess the effects of a pilot intervention intended to reduce negative attitudes toward the poor using a non-probability, convenience sample of 202 (149 women, 53 men) predominately White college students from a midsized Midwestern university. The study was conducted so that the program, prior to full-scale implementation, could be refined and more effectively designed. Hierarchical linear modeling was used to fit a two-level baseline (unconditional means) and growth model. The baseline model regressed the within individuals (nested) measures of attitudes toward the poor on time. The growth model regressed attitudes toward the poor on level two individual characteristics of gender, receipt of public assistance, race, and just-world beliefs.