Aptitudes and Instructional Methods

Background

When asked about the effectiveness of different treatments, educational researchers often respond that “it depends” on the type of student exposed to the treatment, implying that the treatment interacted with some student characteristic. Two types of interactions are important in ATI research: ordinal and disordinal, as shown in Figure 1. In ordinal interactions (top two lines in Figure 1), one treatment yields superior outcomes at all levels of the student characteristic, though the difference between the [Page 39]outcomes is greater at one part of the distribution than elsewhere. In disordinal interactions (the bottom two lines in Figure 1), one treatment is superior at one point of the student distribution while the other treatment is superior for students falling at another point. The slope difference in ordinal interactions indicates that ultimately they are also likely to be disordinal, that is, the lines will cross at a further point of the student characteristic distribution than observed in the present sample.

Figure 1 Ordinal and Disordinal Interactions

Research Design

ATI studies typically provide a segment of instruction by two or more instructional methods that are expected to be optimal for students with different characteristics. Ideally, research findings or some strong theoretical basis should exist that leads to expectations of differential effectiveness of the instruction for students with different characteristics. Assignment to instructional method may be entirely random or random within categories of the student characteristic. For example, students may be randomly assigned to a set of instructional methods and their anxiety then determined by some measure or experimental procedure. Or, in quasi-experimental designs, high- and low-anxiety students may be determined first and then—within the high- and low-anxiety groups—assignment to instructional methods should be random.

ATI research was traditionally analyzed with analysis of variance (ANOVA). The simplest ATI design conforms to a 2 × 2 ANOVA, with two treatment groups and two groups (high and low) on the student characteristic. In such studies, main effects were not necessarily expected for either the treatment or the student characteristic, but the interaction between them is the result of greatest interest.

...