An integration of research findings from investigations of pictorial stimulus complexity

Abstract

The purpose of this study was to generate future research hypotheses from research integration results. This integration was limited to the body of research studies utilizing static illustrations in instructional materials. Differential instructional effectiveness of five iconic coding elements (representation realism, chroma, context, embellishment, and shading) was examined in relation to time of testing, presentation pacing, and grade level. A three-tiered screening procedure was used to pare 701 research reports to the sample's 121 data sets. Meta-analysis techniques were used to calculate a common metric of achievement effect for these studies. The data yielded 2,607 effect-size measures, interpreted as standard deviation differences between treatment groups. Multiple linear regression and frequency analyses identified six potential interactions (grade level by time of testing, grade level by pacing, grade level by chroma, representation realism by chroma, pacing by embellishment, and representation realism by pacing). Jackknifing techniques produced a grand mean and separate means for main effects and for individual cells of the potential interactions. Jackknifing the effect-size means equalized the contribution to the overall mean from studies with unequal numbers of effect-size values. Jackknifed mean differences were described in terms of small, medium, and large levels of practical significance. Effect-size values were interpreted both in terms of the size of the mean differences between the predictor variables and verbal control groups, and in terms of the size of the mean differences between levels of the coded variables. Hypotheses for both types of effect-size differences were proposed at each practical significance level. A total of 39 hypotheses were derived. Positive picture effects were noted for 65 of the 68 jackknifed means. Small practical significance levels were reached on 54 occasions, and large significance levels were reached 13 times. By combining the various effect-size values, the most consistently effective illustrations were nonembellished colored chirographs, presented with external pacing, in grade levels 7-12. The least effective illustrations were any black and white pictures (either chirographs or photographs) presented with internal pacing at the college level, and evaluated under delayed testing conditions.