The non-significance of the effect size in long-term duration (>6 months) is counterintuitive, but consistent with those of Kulik and Kulik (1991), who found that computer-based instruction had a greater effect when the duration was shorter. Kulik and Kulik (1991) and Cheung and Slavin (2013) proposed three reasons for why short-term treatments have better effects: high novelty value, stronger interventional supports, and different measurement tools for the dependent variables. These explanations are also applicable to the present findings. In most studies with intervention durations less than 6 months, the use of mobile devices and the applied teaching methods were both novel, so the students were more easily engaged in the
activity. Cross-analysis of intervention duration with other moderator variables provides data that supports these arguments. For example, most research that took place over a 6-month period used general-purpose software (66.7%; Table C2 of Appendix C), which did not necessarily match the needs of the learning scenarios in specific learning topics. Furthermore,
around half of the studies (44.4%; Table C2 of Appendix C) with durations of >6 months placed the computers directly in the classroom and did not specify the teaching methods to be used to achieve specific educational goals. Conversely, 57.1% of the
studies lasting for >1 month and 6 months used learning-oriented software for specific teaching and learning goals, and 94.3% specified a specific teaching strategy instead of simply using computers for some unspecified purpose in the classroom (Table C2 of Appendix C).