COMMENTARY Toward a Valid View of Human Factors Research: Response to Vicente (1997)

Abstract

INTRODUCTION Vicente (1997) presents several interesting ideas regarding human factors research and the relationship between basic and applied research. Many of the issues regarding basic and applied work have been discussed recently textbooks (e.g., Payne & Conrad, 1997), chapters (e.g., Payne, Conrad, & Hager, 1997), and articles (e.g., Koriat & Goldsmith, 1994), and we feel that both these sources and Vicente's commentary provide important reminders of the importance of both of research. Although we agree with some of Vicente's views, we differ on at least two important points. First, we question the utility and accuracy of Vicente's four-type categorization scheme for human factors research. Second, we disagree with his characterization of the work of Payne, Lang, and Blackwell (1995) and the alleged between their work and that of Hansen (1995). The first point concerns Vicente's prescriptions regarding human factors research. Vicente characterizes human factors research as including four types of research. We find this characterization problematic two ways. First, the criteria used to classify research into these four are vague, insofar as no operational definitions are given. The definitions Vicente provides are all relative; for instance, 1 experiment highly controlled laboratory experiment, and experiment less controlled but more complex experiment (p. 324). Such definitions make it difficult to categorize any single study objectively. For example, Vicente cites the Gould et al. (1987) study of reading from paper versus CRT as 1 research. However, if one were to compare the work of Gould et al. with tightly controlled experiment examining eye movements reading, the work of Gould et al. might be considered research. Second, we take issue with the picture Vicente paints concerning the relative strengths and weaknesses of each type of research. For example, he argues that research more likely to generalize than 1 research because the former more representative of operational settings. Vicente states that this assertion a fact (p. 326). We agree that the more closely an experimental setting emulates specific real-world setting, the more likely the results are to apply to that specific setting. However, the extent to which research findings can or should be generalized across settings depends on the extent to which critical factors controlling behavior are common across the original setting and the setting to which one generalizes. If basic laboratory study identifies factors that influence performance, then these factors will allow one to make predictions about the real world. Vicente misses the point that knowing the extent to which Type 2 experiment representative requires knowing which factors determine representativeness and the setting for which one wishes to generalize. It not statistical fact. For applied researchers it absolutely essential that the results of studies generalize beyond the original research setting. If studies lack generalizability, then with each new operational setting one forced to conduct research specific to that setting. Such an approach expensive and inefficient. Our second major point concerns the characterization of the research by Hansen (1995) and Payne et al. (1995) that Vicente uses to motivate his arguments. Vicente quotes statements from these two articles that, when taken at face value and out of context, appear to be at odds. Vicente asserts that there the assertions made these two and that in fact one of the two papers is incorrect (p. 324). As authors of one of the papers question, we think it important to set the record straight. In our opinion, the glaring contradiction between Hansen and Payne et al. simply does not exist. …