By employing the Kruskal-Wallis analysis of variance by ranks to test whether the effectiveness of the interventions on the dependent variables was influenced by
moderators, no significant difference was found for the moderators of setting,
interventionist, category of participants, age and gender of participants, implying that the effectiveness of the interventions on the dependent variables can be generalized to different settings, interventionists, categories of participants, and ages and genders of participants. In a test as to whether it was the case that the longer the length of the treatment the higher the effectiveness, a Pearson correlation coefficient was calculated and no significant correlation between the length of treatment phase and the effect size was found, r(1089) = -.01, p = .78.
The only two moderators which showed significance were the order of pair of phases and the kind of design. The second pair of the reversal designs showed a significantly larger mean effect size than the first one. The Mann-Whitney U test showed that Z = -2.36, p = .02. The results in the research employing reversal design demonstrated a higher mean effect size than those using multiple-baseline designs.
The Mann-Whitney U test showed that Z = -2.98, p = .01.
Discussion
The effectiveness in terms of the mean effect size of the 12 reinforcing strategies investigated in the present study ranges from .83 to .96, i.e., from a moderate to large effect size as compared with the criterion suggested by Scruggs et al. (1986).
The finding that the effectiveness of “activities” and “token” were significantly higher than that of “praise”, “edible”, and “object” may possibly be explained in that the American participants in the relevant studies were not deprived personally in ordinary daily life in respect to food and object reinforcers, and thus it may not generalize the finding to individuals of less wealthy countries. Praise as a secondary reinforcer acquires its power for reinforcement after being paired by association with a primary reinforcer. Agents who deliver positive reinforcers frequently, though not always, conjoin them with praise. Such a combination functions as a kind of
conditioned positive reinforcer. A nearly large mean effect size of .89 of praise found in the present study supports the findings in Reinke, Lewis-Palmer and Merrell’s (2008) study, which indicated that the performance feedback on the rate of teachers’
praise helped the teachers to increase behavior-specific praise and that increased praise by the teachers contingent on appropriate behaviors led to the reduction of disruptive behaviors of their students, and that, on the contrary, reprimanding inappropriate behavior led to an increase in disruptive behavior of their students.
Hence, the best strategy of classroom management for a teacher should be to focus on praising an appropriate behavior rather than on reprimanding inappropriate behavior.
The effect of this strategy was demonstrated experimentally as early as 1968 by Thomas, Becker, and Armstrong.
The finding that there was no significant correlation between the length of a treatment phase and its effectiveness is the same as in the results of the study conducted by Vegas, Jenson, and Kircher (2007). What is important to the
magnitude of the effectiveness is not the length of the treatment phase but the power of the reinforcement provided by a reinforcer. The results of the present study, that the mean rank of the effect size of the second pair of baseline-treatment phases was higher than that of the first pair can serve to reduce concern on the orthogonal slope
change mentioned by Scruggs, Mastropieri, & Casto (1987). They anticipated that the orthogonal slope change would threaten the effect size of the second baseline
treatment pair. But the result of the present study demonstrates that such is not the case. In Ma’s (2006) study, the mean effect size of the second baseline-treatment pair was higher than that of the first one, although the difference was small, and in the present study the same result was not only replicated but furthermore, the difference reached a significant level of .02.
The result that not all positive reinforcers have the same effectiveness also justifies the importance of the element of RN in the formula (1), that is, the reinforcer must satisfy the need (or reduce the deprivation) of the individual. This conclusion was also partially supported by the result of the present study in that an intervention showed a higher effectiveness if the reinforcer was determined through the use of a preference test rather than simply by asking for the suggestion of significant others such as the parent. The finding that reinforcer “activities” had the highest mean effect size (.95) confirmed indirectly the Premark principle, which states that a high
frequency activity can be used to reinforce a low frequency activity. Logically
inferred, the Premark principle can be alternatively expressed as the use of an activity towards which the student feels a strong intrinsic motivation as an extrinsic reinforcer to motivate him or her to learn an important academic or social behavior for which his or her intrinsic motivation is weak. This finding has practical implications. As
suggested by Kern, Babara, and Fogt (2002), academic activities can be associated with opportunities to make choices, such as choice of activity, choice of teaching of learning materials, and choice of task sequence, and the findings can be used to modify class-wide curricula. Their research demonstrated that curricular modification resulted in increased levels of engagement and decreased levels of destructive
behavior and that it can be compatible with school policy.
The present study does not address the controversy between intrinsic and extrinsic motivation. However, all the studies located for this present research were conducted to improve behaviors in completing a task or performing an action according to an acceptable standard and were based on the assumption that the participant had weak intrinsic motivation. Thus the results of the present study imply that extrinsic
reinforcement may motivate participants with a weak intrinsic motivation to improve quantitatively and/or qualitatively in their academic or social behaviors. Deci (1975) defined intrinsic motivation as a motivation to satisfy the individual’s needs for feelings of competence and self-determination. Table 2 shows that “activity” was the most effective reinforcer. Among the activities listed in the independent variable coded “13”, a prominent part was “allowing choice of activities”, which by
operational definition is in accord with the element of “self-determination” in intrinsic motivation. Although the average effect size of the reinforcer on the quality of an academic behavior was only moderate (.85), depicting that such quality is among the most difficult to be changed because it is also influenced by the prior achievements of the individual student, however, this result also shows that when the quality of an academic behavior was reinforced, a feeling of competence would be elicited in the student, which would facilitate the building of intrinsic motivation. Cameron and Pierce (1996) stated that the negative effect of a reward on intrinsic motivation was caused by non-contingent reward and could be prevented by rewarding people for completing work, solving problems successfully, or attaining a specified level of performance. The dependent variables in the current study “quantity of academic behaviors” and “quality of academic behaviors” fall within the category of
“completing work” and “solving problems successfully, or attaining a specified level of performance”, respectively and therefore would not be harmful to intrinsic
motivation. Hence, by logical inference, intrinsic and extrinsic motivation could be
mutual facilitators, that is, a student demonstrating excellence in intrinsically
motivated academic behavior would be likely to accept extrinsic primary or secondary reinforcers delivered by significant others, such as parents, teachers and/or peers in the natural setting, and conversely, an extrinsically reinforced academic behavior would be likely to turn into a intrinsically motivated one because a frequently reinforced behavior would be likely to become a habit, which resembles an intrinsically motivated behavior. Future research should be done on the mutual fostering of both kinds of motivation.
The feasibility of the PEM approach suggests that PEM scores can be used to describe and judge the effectiveness of a treatment based on the figures provided in an article employing a single-case experimental design in accordance with the criterion set by Scruggs et al. (1986).
The analysis of whether a different schedule, duration, intensity, or amount of reinforcement would produce a different effect remains for further study as the number of studies included in the present investigation is too small to allow for such analysis.
References
Baker, M. J. (2000). Incorporating the thematic ritualistic behaviors of children with autism into games: Increasing social play interactions with siblings. Journal of
Positive Behavior Interventions, 2(2), 66-84.
Cameron, J. & Pierce, W. D. (1994). Reinforcement, reward, and intrinsic motivation: A meta-analysis. Review of Educational Research, 64, 369-423.
Cameron, J. & Pierce, W. D. (1996). The debate about rewards and intrinsic motivation: Protests and accusations… . Review of Educational Research, 66, 39-51.
Chen, C. -W., & Ma, H. -H. (2007). Effect of intervention on disruptive behaviors: A quantitative synthesis of single-subject researches using the PEM approach. The Behavior Analyst Today,
8(4), 380-397.
Deci, E.L.(1975). Notes on the theory and meta-theory of intrinsic motivation.