Different learning environments provide different learning experiences and ought to serve different achievement goals. It was hypothesized that constructivist learning environments lead to the attainment of achievements that are consistent with the experiences that such settings provide and that more traditional settings lead to the attainments of other kinds of achievement in accordance with the experiences they provide. A meta-analytic study was carried out on 32 methodologically-appropriate experiments in which these two settings were compared. Results supported one of the hypotheses showing that overall constructivist learning environments are more effective than traditional ones (ES=.46) and that their superiority increases when tested against constructivist-appropriate measures (ES=.90). However, contrary to expectations, traditional settings did not differ from constructivist ones when traditionally-appropriate measures were used. A number of possible interpretations are offered among them the possibility that traditional settings have come to incorporate some constructivist elements. This possibility is supported by other findings of this study such as smaller effect sizes for more recent studies and for longer lasting periods of instruction.

For the last three decades numerous kinds of technology-intensive learning environments have been developed and their effectiveness repeatedly compared with that of more traditional settings (Hong, 1973; Kulik, Kulik, & Bangert-Drowns, 1985; Liao, 1998; Kulik, & Kulik, 1991; Bayraktar, 2001/2002). Most of these comparisons employed standard criteria by means of which learning achievements of both kinds of learning environment were studied. The evaluation criteria and the achievement measures were usually common to both kinds of learning environment, thus overlooking the possibility that due to their inherent differences, these qualitatively different kinds of learning environment are likely to serve qualitatively different learning goals and outcomes. The use of the same achievement measures is therefore a search for the lowest common denominator which may possibly mask important differences in the kinds of learning outcomes attained. 

            The purpose of the present study was to test the above assertion through a meta-analysis of comparative studies. Toward that end, the learning attainments of constructivist technology-intensive learning environments (CTILEs) were compared with those of more traditional, didactic ones. The differential effects of the learning environments were computed on the basis of a meta-analysis of studies that carried out such comparisons and employed differential outcome measures that corresponded to the rationales, activities and experiences of the learning environments under study.

    The following hypotheses were tested in the current meta-analysis:

1. Math students in CTILEs show higher learning achievements when tested against constructivist-appropriate criteria, as compared to students in more traditional learning environments.


2. Math students in traditional learning environments show higher learning achievements when tested against traditionally-appropriate criteria, as compared to students in more constructivist learning environments.  

Method

         The statistical analysis was conducted according to the meta-analytic approach (Glass, McGaw, & Smith, 1981; Hedges, & Olkin, 1985) of comparative studies existing in the field of mathematics education. This system of analysis enables the reevaluation of the findings in existing research through an inclusive quantitative approach. The “subjects” are 32 studies that meet the criteria set in this study. In each study the experimental group was engaged in learning math in a CTILEs setting whereas the control group studied math in a traditional way. Each study classified by its characteristics (constructivist-appropriate or traditionally-appropriate measures, grade level, duration of treatment, year of publication). In cases of combined constructivist and traditional outcomes in the same study, the main outcome measured was selected. Inter-rater reliability of the classification of mathematical goals was assessed on the basis of the judgments of ten independent expert judges. Inter-rater agreement reached 87%;

Results

  Of the 32 studies included in the present meta-analysis, 27 (85%) of the ESs were positive, favoring CTILEs, three studies (9%) were negative and favored traditional learning environments, and two (6%) showed no difference between CTILEs and traditional learning environments. The range of the ESs was from -1.15 (higher achievements of the traditional learning environment) to 2.00 (higher achievements of CTILE). An effect is considered to be small when ES = 0.20 standard deviations, of medium magnitude when ES = 0.50 standard deviations, and large when ES = 0.80 or more standard deviations (Cohen, 1977). Mean effect size for all 32 studies was 0.46, suggesting that overall, the learning attainments of CTILEs are of medium magnitude when compared with traditional learning environments. Large weighted average effect size (d+ =.90, Qb=142.61, p<.01), was found in that group of experiments that evaluated learning according to constructivist criteria in CTILEs, as compared with those who studied in a traditional learning environment. That is, in accordance with the first hypothesis, learning attainments are profoundly better in constructivist learning environments when constructivist-appropriate measures are used. However, the second hypothesis could not be supported: Even when traditional measures are used, achievements are not any higher in traditional learning environments than in less traditional ones (d+ =.11, Qb=142.61, p<.01).  It thus appears that regardless of the measures used, traditional learning environments yield either poorer or similar outcomes when compared with CTILE. However, it needs to be noted that the variation of effect sizes between studies using traditional achievement measures is larger than that within the group of constructivist-using studies (SD=.66 and .48, respectively). This is an unexpected observation One would have expected didactic instruction to be more uniform than instruction in CTILE. But this, apparently, was not the case.

Discussion

The main purpose of the present meta-analysis was to compare the relative effectiveness of these two kinds of learning environment in differentially achieving learning objectives that are appropriate for each. As predicted by the first hypothesis, math students in CTILEs, compared to students in more traditional learning environments, show higher learning achievements (ES=.90) when tested against constructivist-appropriate criteria. On the other hand, findings did not support the second hypothesis: Math students in traditional learning environments showed learning achievements similar to those studying in CTILEs when tested against traditionally-appropriate criteria (ES=.11). Although our findings lend some support to the novelty interpretation, a more reasonable interpretation may well be that with the increased integration of computers into education (National Center for Education Statistics, 2000; Reiser, 2002) and with the increased acceptance of constructivist-like instructional goals (Gagnon, & Collay,  2001; Karagiorgi, & Symeou, 2005), traditional instruction gradually moves away from the strict didactic approach. It become less uniform as it comes to incorporate constructivist-like elements such as team work and team-based problem solving, which are then reflected in achievement measures.