Analysis & Critique of Who benefits from learning with 3D models? The case of spatial ability (Huk, 2006)
The purpose of this study was to determine if interactive three-dimensional (3D) models of plant and animal cells had an effect on students’ learning of cell biology in a hypermedia learning environment, and whether the effect was different between students with high vs low spatial ability. In the context of this paper, spatial ability was considered to be the students’ ability to visualize and rotate 3D images in their mind. The researcher identified a gap in research examining the educational value of 3D models, stating that most prior research had not found either advantages or detriments when using 3D vs 2D images. There was also little or no previous research connecting spatial ability with the educational value of 3D models.
The most significant prior studies linked to the research are: Keehner et al. (2004), which revealed that there is an effect on comprehension of 3D computer modeling that depends on spatial ability, Mayer (2001) which presented the ability-as-enhancer hypothesis (where higher spatial ability increases comprehension of 3D models), and Hays (1996), which proposed the alternate ability-as-compensator hypothesis (where people with lower spatial ability benefit more from the 3D models as these models compensate for the student’s lower ability to visualise 3D structures). A compelling idea in this study was that the addition of interactive 3D models may increase the cognitive load of learners, especially in a hypermedia environment, and that a learner’s spatial ability may affect whether they are cognitively overloaded by the extra information.
Spatial ability of participants was measured by their score on a 21-question tube figures test. A median split was used to categorize students as having either high or low spatial ability in the graphical representation of results; it was not stated if this was the method used in calculations or if a raw score out of 21 was used. Knowledge acquisition was measured by student scores on a pencil and paper post-test with a total of 7 questions. The first 3 questions were designed to test auditory recall and the second 4 questions to test visual recall. Cognitive load was measured indirectly by students’ self-reporting of agreement or disagreement on a 5-point scale with the statement “The presentation of the animal and plant cell is easy to comprehend” (Huk, 2006, p. 398).
The research was quantitative as the researcher used test scores and statistical analyses to measure the results (both test results and survey results). It was experimental because there was an intervention (the addition of 3-D models to the learning material) and the participants were randomly assigned to control or experimental groups. There was an element of problem-based research because the researcher was directly exploring the problem of whether 3-D models were beneficial to learners of cell biology. The research was also partially theory-based because the researcher framed the experiment in such a way that it could support/refute two conflicting hypotheses that had been proposed in previous studies: either the ability-as-compensator hypothesis or the ability-as-enhancer hypothesis. The support of one of these hypotheses could possibly be used to generalize about a larger population of learners or help to refine the theories themselves.
One independent variable was the presence or absence of interactive 3-D cell models in the software that students used prior to their knowledge acquisition test. The experimental group was given identical software to that of the control group, except with the addition of 3-D cell models. The second independent variable was the spatial ability of the participants. A median split was used to categorize students as having either high or low spatial ability (at least for the purposes of graphical representation). It was unclear if the researcher used the raw score out of 21 in their statistical analysis.
One dependent variable was students’ knowledge acquisition, measured by the number of correct answers on a post-test. Knowledge acquisition was split into the sub-categories of auditory and visual recall. The other dependent variable was students’ impression of the module, a rating of whether the students found the information easy or difficult to understand. This was interpreted as the self-reporting of cognitive load. The attribute variable was students’ prior knowledge of the subject material. This was analysed using the scores obtained on a pre-test one week ahead of the actual test.
The research design was an experimental, randomly assigned, 2 x 2 design. Control procedures that were used included randomization of subjects into intervention or non-intervention groups and the statistical control using students’ prior domain knowledge (and in some cases, the amount of time spent on the content module) as covariates. The author noted that the research took place in the students’ everyday classroom surroundings to increase external validity of the experiment.
The sample of research participants consisted of 106 high school or college-level biology students from more than one school (with the total number of schools not specified) in Germany. The author reported that there were 54 students randomly assigned to the control group and 54 to the experimental group. About 67% of participants were female and the mean participant age was 18.49 years (SD = 2.16 years).
As an alternative hypothesis, the author posited that gender differences may influence spatial ability and an imbalance of male to female participants could have introduced bias to the results. However, the random assignment of participants ensured that the ratio of male to female participants was nearly equal between the two groups.
The data were analyzed using linear regression models with prior knowledge as a covariate, and in the case of auditory recall, time spent on the module was used as a covariate as well. The inclusion of time spent as a covariate made no statistical difference for visual recall, so was not included there.
A major finding of the study was that students with high levels of spatial ability showed higher mean post test scores (both auditory and visual) when the 3D model was present in their software, while the opposite was true for students with low spatial ability. The results suggest that only students with high levels of spatial ability benefitted from the inclusion of interactive 3D models.
The most important point made in the discussion section was that students with lower spatial ability may be faced with cognitive overload when they are faced with integrating the information from a 2D drawing with that of a 3D computer model.
There were a few methodological issues in this study. The most obvious was that the number of participants in the research study did not add up. It was reported that there were 54 students in each of the groups (intervention and non-intervention), but the total number of participants in the study was reported as 106. It is possible that there was an error made in the writing of the paper and each group only had 53 students, or perhaps 2 participants were lost throughout the course of the study (although this was not reported). It may also have been possible that there were 2 non-binary participants, but it is unclear why the researcher would not include their data, especially if it were one participant per group, if this were the case.
Additionally, there were no details describing how student answers on the pre- and post tests were graded. Whether the answers were rated by one or more people and whether those raters had a high level of inter-rater reliability could have an impact on the validity of the results. If multiple raters had differing opinions on the students’ answers and did not reconcile these differing opinions through averaging or some other means, the data would not be reliable, and one would have to question the results of the statistical analyses.
In the presentation of the data, the researcher used a median split to show the difference in knowledge acquisition between students with high spatial ability and low spatial ability. There was no clarification in the methods section whether the actual spatial ability score of the students or median split was used in data analysis. If it were a median split, this would reduce the validity of the data.
On the other hand, the same computers were used at each of the study locations and the same instructor gave the directions to the students. This control for confounding effects of technology and different instructors was a notable strength in the research design. As well, random selection of test subjects controlled for differences in prior knowledge, which when calculated, was not statistically different between groups. Although there were a higher number of females in the study, the proportion of female to male participants was not different between groups either.
Overall, I found this study useful both for my personal work as a biology laboratory instructor and as a deeper investigation into spatial ability and cognitive load of learners. In order for the study to have repeatability, and in order to effectively gauge the reliability of the study, more detail would be required in the methods. However, the researcher did appear to pay attention to detail and consider alternative hypotheses and control for confounding. Therefore, I would recommend that this study be considered when making decisions regarding the introduction of computer models to students learning cell theory. In cases where students have limited time to interact with software, the addition of more learning tools may impact their ability to recall information. This study indicates that more research is required to explore the connection between spatial ability, cognitive load, and 3D computer models in other areas of study.
References
Hays, T.A. (1996). Spatial abilities and the effects of computer animation on short-term and long-term comprehension. Journal of Educational Computing Research (14), 139–155.
Huk, T. (2006). Who benefits from learning with 3D models? The case of spatial ability. Journal of Computer Assisted Learning, (22)(6), 392-404. https://doi.org/10.1111/j.1365-2729.2006.00180.x
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. In Proceedings of the 26th Annual Conference of the Cognitive Science Society (eds K. Forbus, D. Gentner, & T. Regier). Erlbaum, Mahwah, NJ, 1576 pp.