Wang, H., Chun-Yen, C., & Tsai-Yen, L. (2007). "The comparative efficacy of 2D- versus 3D-based media design for influencing spatial visualization skills"

From MathWiki

Wang, H., Chun-Yen, C., & Tsai-Yen, L. (2007). The comparative efficacy of 2D- versus 3D-based media design for influencing spatial visualization skills. Computers in Human Behavior, 23, 1943-1957.

Available Online:

Table of contents

Study Outline & Method

Wang et al. (2007) explore the differential effects of 2D and 3D virtual representations on the spatial skills of 23 undergraduate students at the National Taiwan Normal University. The students were tested for spatial ability using a test element from the Purdue Spatial Visualization Tests, the Purdue visualization of rotations test (PVRT), a common tool for assessing spatial ability in the literature.

The students were then randomly assigned to one of two groups: the 2D group (n=10) or the 3D group (n=13). The experimental treatments were 2D and 3D versions of a program called CooTutor (Coordinate Tutor) – “a web-based system incorporating multiple media representations aimed at facilitating learners’ spatial reasoning” (Wang et al., 2007, p.1946). Both of the treatment conditions contained the same information (i.e. narrations and examples), which “were intended to instruct skills for handling spatial visualization tasks and let students construct their own senses of spatial transformation” (Wang et al., 2007, p.1949). However, the 3D version of the program involved the ability to manipulate 3D objects to enhance the textual description, while the 2D version contained only text and static images.

The interactivity in this study was slightly different than Keehner et al. (2004). (See Keehner M, Montello D.R., Hegarty M. & Cohen C. (2004). "Effects of interactivity and spatial ability on the comprehension of spatial relations in a 3D computer visualization" for more information.) In both conditions, participants had the ability to navigate through the tutorial at their own pace (within a 35 minute period), via hypertext navigation links that the participants were assumed to be very familiar with. Therefore, there was interaction in both conditions in this study.

Students were allowed to leave (to write the post-test) as quickly as they wanted, as long as they viewed all of the included pages in the tutorial. This was an effort by the authors to mimic web-based learning, which they identify as learning at one’s own pace. After the tutorial treatment, both groups were re-tested following the treatment condition, once again using the PVRT.


In the post-test, the 3D group scored higher than the 2D group (15.124 vs. 13.238 out of 20). However, the 3D group also had a higher mean score on the pre-test as well. ANCOVA analysis on the post-test scores revealed no significant difference between the two means (2D vs. 3D). However, the authors point out that the effect size of the difference was “very close to a medium effect size” (criteria for effect size as determined by Cohen, 1988, in Wang et al., 2007, p.1952).

From a statistical point of view, the authors point out that the small sample size of the study is its major limitation. They express the fact that statistical significance may be more of a reflection of sample size than actual difference due to the treatment conditions. Thus, they take the near-emergence of a medium effect size to “signify the possibility of finding statistical significance of this comparison with a future replication of study with a larger sample size” (Wang et al., 2007, p.1953).

Participants in the 2D group spent significantly less time than the 3D group working with the tutorial (M=4.13 vs. M=8.33 minutes). Statistical analysis (t-test) showed that this difference was significant, which suggests that the participants interacted with the treatments in unique ways.


Interaction with the software

Why did the 3D group spend more time on the tutorial than the 2D group? The authors provide the following possibilities:

  • Students may have preferred the 3D environment, and were willing to spend time learning through it.
  • Students were “perplexed” by the 3D environment, and thus had to spend more time learning how to work with it. Here, it is possible that the students may be improving their interface manipulation skills but not necessarily their spatial visual skills.
  • Student attitudes towards the different mediums may be central, but the study did not delve into this possibility.

Does this type of instruction help develop spatial visualization skills?

The statistical insignificance between the treatment groups does not allow the authors to make a conclusive decision regarding the possible benefits of using 3D technology over 2D representations to improve spatial skills. “However,” the authors reason, “a medium effect size of the difference between groups on adjusted post-test, different using behaviors, and the possibly disparate attitude underlay students were found… [And] future replications of study with a large sample size may be able to derive a statistically significant result for the same hypothesis” (Wang et al., 2007, p.1954).

Other Resources

  • PowerPoint presentation by the authors (2004) – contains information related to this current study: IEEE Presentation (


Back to Mordy's Links

Back to Spatial Reasoning Home