33.6 The Role Of Learner Characteristics

Educators know that individual learner characteristics play a huge role in how fast and how well overall learning occurs. Generally speaking, since its inception CBI has continually been held up as a promising vehicle able to somehow tailor instruction to meet the individual needs of each learner (Suppes, 1966; U.S. Congress, Office of Technology Assessment, 1988). Just how these instructional adaptations are best concocted, however, is a matter of debate.

Some models for adaptive CBI are largely program con-trolled and attempt to present to each student appropriately matched instructional events according to some relevant individual difference variable. Examples of such approaches include regression models that assign optimized instructional conditions based on either stable trait variables (McCombs & McDaniel, 198 1; see also 22.3) or on-task state variables (Rivers, 1972; Ross & Morrison, 1988; see also 22.5) and schemes to branch instruction according to some optimiz-ing mathematical model (R. C. Atkinson, 1972; Holland, 1977; Smallwood, 1962; Tennyson, Christensen & Park, 1984; see also 22.4). Few of these approaches have made it into commercially produced CBI, however.

On the other hand, for reasons of feasibility, attractive-ness, and understandability, most of the CBI found in school software libraries has at least some learner-con-trolled features that their manufacturers tout as helping to accommodate the learning needs of each individual student. The idea, as Merrill (1973, 1975) and Federico (1980) have propounded, is that students will make their own decisions throughout a lesson so as to best match their own learning styles, personality, or other relevant traits.

As we have seen, however, learner control does not seem to be a superior overall instructional strategy. A closer examination does seem to indicate differential student effectiveness of instructional choices, although perhaps not in the way the software producers had intended. That is, some students are able to use learner control to their advan-tage; others, however, use it actually to their detriment.

33.6.1 Paradigms

It is useful at this point to review the major paradigms employed in research that focuses on the relationship between learner characteristics and learner-controlled CBI. There are several methods available to researchers wishing to study such interactions. One common approach adopts an "aptitude-by-treatment interaction" or ATI perspective (see 22.5; Carrier & Jonassen, 1988; Cronbach & Snow, 1977). That is, students' stable cognitive and personality "trait" variables are viewed as possibly interacting with predetermined instructional features to produce differentially effective learning, particularly within a learner-controlled context (Snow, 1980). One example (Judd et al., 1974a) found that the personality variable of "achievement via independence" predicts certain behaviors under learner control (see 22.3.4.6). Snow (1979) also takes an ATI approach and presents some data using various statistical profiling techniques that appear to be fairly successful at sorting college students enrolled in a BASIC programming course into good and poor options selectors according to their scores on a variety of aptitude measures.

The usual aim of ATI studies is to find instructional treatments that would somehow benefit students possessing different learner characteristics or profiles. However, in spite of Federico's (1980) suggestion that learner control might allow students to select effectively instances based on their own cognitive requirements, there is ample evidence that learner control serves to magnify student differences rather than eliminate them. Wilcox (1979), for instance, presents a review of non-computer-based ATI studies and concludes that learner control tends to exacerbate problems arising from individual differences instead of minimizing them. Snow (1980), too, argues that a learning environment that allows learners to control instruction might possibly produce stronger relationships between individual differences and learning to the degree that these individual differences are free to operate than would "fixed" instruction. In a variation on ATI approaches, which Tobias (1976) calls "achievement-by-treatment interactions" (see 22.3.7), dif-ferential results have also been found for effectiveness of options selection for students with differing amounts of prior knowledge (Ross & Rakow, 198 1; Tobias, 1987a).

Merrill (1975) discusses several frequently invalidated assumptions regarding "aptitudes" and "treatments" in a typical ATI model. He points out that quite often the most germane learner characteristics are actually unstable, vary-ing from moment to moment during instruction. Likewise, treatment effects may similarly not always hold under the variety of conditions present in typical educational settings. Lastly, he argues that instead of instruction being adapted to the individual, we should allow students to adapt the instruction for themselves. This forms one of the bases for the inclusion and importance of learner control in his theory of instruction.

While standard ATI approaches seek to understand the differential effectiveness of learner control with individual differences measured prior to instructional intervention ("trait" variables), other approaches choose to explore learner variables measured during the instructional task, so-called "within-task" variables (Federico, 1980) or situational "state" variables. These presumably reflect momentary variations in certain learner characteristics that also could interact with the specific instructional situation. Tennyson and Park (1984) discuss the need to investigate the phe-nomena of moment-by-moment interactions of instruction and individual differences, in particular within learner-controlled environments.

Studies by Seidel, Wagner, Rosenblatt, Hillelsohn, and Stelzer (1975) examining students' ongoing expectancies of success, by Fisher, Blackwell, Garcia, and Greene (1975) on momentary changes in attributions for success and failure, and by Goetzfried and Hannafin (1985), Johansen and Tennyson (1983), Tennyson (1981), and Tennyson and Buttrey (1980) investigating on-task mastery self-assessment by students, illustrate the utility of variables that occur during the course of learner-controlled CBI.

However, still another possibility not discussed by either Carrier and Jonassen (1988) or Merrill (1975) is to not necessarily adapt instruction to fit the student, but rather to attempt to change the student to optimally use the instruc-tion. That is, if we can identify modifiable characteristics of the students which typically produce dysfunctional interac-tions with instructional treatments, we might attempt to alter those characteristics so that the student and instruction are better matched. Suggested approaches using this para-digm are presented later in this chapter.

Thus, both person and instruction variables can be considered either stable or unstable, perhaps reciprocally changing throughout the course of instruction. This paradigm also allows for the occurrence of aptitude-by-treatment "corrections" (Gehlbach, 1979), that is, selecting treatments to eliminate the effects of individual differences rather than to accommodate diem.

This expanded "adaptive instruction" paradigm presents a revised set of larger questions to the researcher: When might instruction respond to variations (both stable and unstable) in individual learners, and how might learners react and respond to changes (macro and micro) in the instruction? Within this framework, there seems to be suffi-cient theoretical, empirical, and practical justification for investigating the mutual relationship between learner differ-ences and instruction under some degree of learner control.