Abstract
Data from two previous investigations (Sadoski & Goetz, 1985; Sadoski, Goetz, Olivarez, Lee, & Roberts, 1990) were reanalyzed to investigate: (1) validity of imagery ratings and reports as measures of reader response to a story, (2) similarity of final regression models for recall and imagery reports. Imagery ratings of story paragraphs were found to predict both imagery reports and recall of a second, independent group of readers, lending convergent construct validity to these measures. Hierarchical regression analyses of within‐subjects variance (cf. Anderson, Mason, & Shirey, 1984; Cohen & Cohen, 1983; Herman, Anderson, Pearson, & Nagey, 1987) ensured that rating effects could not be attributed to surface‐level text factors. Comparison of final regression models for imagery and recall revealed differences that appeared to support Paivio's (1971, 1986) dual code hypothesis.