An assessment of the validity and discrimination of the intensive time-series design by monitoring learning differences between students with different cognitive tendencies

Authors


Abstract

Intensive time-series designs for classroom investigations have been under development since 1975. Studies have been conducted to determine their feasibility (Mayer & Lewis, 1979), their potential for monitoring knowledge acquisition (Mayer & Kozlow, 1980), and the potential threat to validity of the frequency of testing inherent in the design (Mayer & Rojas, 1982). This study, an extension of those previous studies, is an attempt to determine the degree of discrimination the design allows in collecting data on achievement. It also serves as a replication of the Mayer and Kozlow study, an attempt to determine design validity for collecting achievement data. The investigator used her eighth-grade earth science students, from a suburban Columbus (Ohio) junior high school. A multiple-group single intervention time-series design (Glass, Willson, & Gottman, 1975) was adapted to the collection of daily data on achievement in the topic of the intervention, a unit on plate tectonics. Single multiple-choice items were randomly assigned to each of three groups of students, identified on the basis of their ranking on a written test of cognitive level (Lawson, 1978). The top third, or those with formal cognitive tendencies, were compared on the basis of knowledge achievement and understanding achievement with the lowest third of the students, or those with concrete cognitive tendencies, to determine if the data collected in the design would discriminate between the two groups. Several studies (Goodstein & Howe, 1978; Lawson & Renner, 1975) indicated that students with formal cognitive tendencies should learn a formal concept such as plate tectonics with greater understanding than should students with concrete cognitive tendencies. Analyses used were a comparison of regression lines in each of the three study stages: baseline, intervention, and follow-up; t-tests of means of days summed across each stage; and a time-series analysis program. Statistically significant differences were found between the two groups both in slopes of regression lines (0.0001) and in t-tests (0.0005) on both knowledge and understanding levels of learning. These differences confirm the discrimination of the intensive time-series design in showing that it can distinguish differences in learning between students with formal cognitive tendencies and those with concrete cognitive tendencies. The time-series analysis model with a trend in the intervention was better than a model with no trend for both groups of students, in that it accounted for a greater amount of variance in the data from both knowledge and understanding levels of learning. This finding adds additional confidence in the validity of the design for obtaining achievement data. When the analysis model with trend was used on data from the group with formal cognitive tendencies, it accounted for a greater degree of variance than the same model applied to the data from the group with concrete cognitive tendencies. This more conservative analysis, therefor, gave results consistent with those from the more usual linear regression techniques and t-tests, further adding to the confidence in the discrimination of the design.

Ancillary