Teaching science lab safety: are virtual simulations effective?
Document
Description
The purpose of this study was to investigate the impact of immersion on knowledge, cognitive load, and presence in a simulation designed to deliver a lesson on science lab safety training. 108 participants were randomly assigned to one of three conditions: high immersion (played an interactive simulation about lab safety in a VR headset), medium immersion (played the same interactive simulation on the computer), or low immersion (watched a video and read about lab safety procedures). Participants completed a pretest, a science lab safety training, a posttest (same as the pretest), a questionnaire with subjective presence questions, and a questionnaire with subjective cognitive load questions. Participants were again asked to complete a follow-up test (same as the pretest and posttest) a week later.
The results revealed three significant findings:
(a) Participants in the high and medium immersion conditions had significantly higher knowledge scores at posttest and follow-up than their peers in the low immersion condition,
(b) Participants in the high and medium immersion conditions reported higher presence scores than participants in the low immersion conditions.
(c) Correlation coefficients suggested that the higher the immersion and presence, the higher the knowledge scores are at posttest and follow-up.
In addition, multiple hierarchical linear regression models were conducted out of which one was significant.
The results revealed three significant findings:
(a) Participants in the high and medium immersion conditions had significantly higher knowledge scores at posttest and follow-up than their peers in the low immersion condition,
(b) Participants in the high and medium immersion conditions reported higher presence scores than participants in the low immersion conditions.
(c) Correlation coefficients suggested that the higher the immersion and presence, the higher the knowledge scores are at posttest and follow-up.
In addition, multiple hierarchical linear regression models were conducted out of which one was significant.