Absence of conflict of interest.
Citation
Steiner, L. J., & Burgess-Limerick, R. (2013). Shape-coding and length-coding as a measure to reduce the probability of selection errors during the control of industrial equipment. IIE Transactions on Occupational Ergonomics and Human Factors, 1(4), 224-234. https://doi.org/10.1080/21577323.2013.844209
Highlights
- The study’s objective was to examine the impact of shape-coding and length-coding of roof-bolting machine levers on selection errors made when controlling the machines.
- The authors conducted three separate laboratory experiments to test the role of shape-coding and length-coding of roof-bolting machine levers on the amount of selection errors made by participants. In experiments 1 and 2, participants were randomized into different lever conditions (shape-coded, length-coded, or non-coded). In experiment 3, the participants used shape-coded levers only and the lever order was changed half-way through the experiment.
- The study found that the proportion of errors significantly improved over time for all lever conditions, indicating a learning effect. The study also found that participants in the length-coded group made significantly fewer selection errors than those in the non-coded group when the order of levers changed.
- The quality of causal evidence presented in this report is high for experiments 1 and 2 because they were based on well-implemented randomized controlled trials. This means we are confident that the estimated effects are attributable to shape-coding and length-coding, and not to other factors. The quality of causal evidence provided in this study is low for experiment 3 because there are not sufficient pre and post measures of the outcome. This means that we are not confident that the estimated effects are attributable to shape-coding; other factors are likely to have contributed.
Features of the Study
The study tested roof-bolting machine safety measures (redesigned levers for reducing injuries by roof bolt operators who accidentally used the wrong levers) on the number of selection errors. The authors recruited 64 participants from the National Institute for Occupational Safety and Health’s (NIOSH) Office of Mine Safety and Health Research and the National Personal Protective Technology Laboratory in Pittsburgh, Pennsylvania.
The study was conducted in a laboratory and used a simulation of a single-boom roof-bolting machine (Fletcher Roof Ranger II) with five levers that are used to control the boom. Four levers (elevation, slew, sump, and stabilizer jack) each move the boom in a different direction, and the last lever (color lever) causes a light to turn on at the end of the boom. The levers were shaped in one of three ways: shape-coded, length-coded, or identical handle (non-coded). The authors conducted 3 experiments to test the various levers:
- In experiment 1, the authors randomly assigned 48 participants to one of three lever conditions (16 participants in each): (1) non-coded, (2) shape-coded, or (3) length-coded. Each participant completed 6 blocks of 40 trials (240 trials total). The participants viewed a demonstration of the function of the levers prior to the experiment. In each trial, participants viewed a short video clip and had to choose a lever and move it in one of the two directions to replicate what they saw in the video. After each attempt, the bolter arm was moved back to the starting position. Each block lasted approximately 6 minutes, with a one-minute break between blocks. The location and control arrangement of the levers were constant throughout the experiment.
- Experiment 2 was conducted three months after experiment 1 with the same group of participants (42 participants) and using the same procedures. The participants remained in the randomly assigned conditions from experiment 1: 14 participants in the non-coded condition, 13 in the shape-coded condition, and 15 in the length-coded condition. Each participant completed 3 blocks of 40 trials with the identical lever arrangement to experiment 1. Next, each participant completed 4 blocks of 40 trials with the order of the levers changed.
- Experiment 3 included 16 new participants who completed the trials using shape-coded levers only. Using the same procedures as experiment 1, the participants completed a total of 240 trials. Each participant completed 3 blocks of 40 trials, then the relationship between lever shape and function was changed for the last 3 blocks of 40 trials.
The outcome measured was the number of selection errors made by the participants. The authors conducted statistical analyses to compare the outcomes of the participants assigned to each lever condition (shape-coded, length-coded, or non-coded), the performance of each group over the timed trials, and the interaction between the lever condition and repeated performance.
Findings
Health and safety
- In experiments 1 and 2, the study did not find a significant effect of lever shape on the proportion of selection errors. However, the study found that selection errors were significantly higher in the first three trial blocks compared to the later blocks for all three conditions, indicating a learning effect.
- In experiment 2, the study found a significant increase in the error rate between blocks 3 and 4, when the order of the levers changed. Further, participants in the length-coded group made significantly fewer selection errors than those in the non-coded group when the order of levers changed.
- In experiment 3, the study found that selection errors were significantly higher in block 1 versus blocks 2, 3, 5, and 6 and significantly higher in block 4 versus blocks 3, 5, and 6.
Considerations for Interpreting the Findings
The authors did not explicitly provide information to calculate attrition, but as the experiments were conducted in a laboratory setting, we can imply that there was no attrition. However, the authors note that the study results may not generalize to a real-world setting since the experiments were conducted in a controlled environment. Although experiment 3 was conducted in a laboratory setting, it was not a randomized controlled trial. The authors compared the proportion of selection errors over the course of six blocks and after the lever function changed using an analysis of variance (ANOVA). The study did not have the required number of pre and post repeated measures to meet the standards established by CLEAR. For ANOVA designs, the highest rating a study can receive is a low.
Causal Evidence Rating
The quality of causal evidence presented in this report is high for experiments 1 and 2 because they were based on well-implemented randomized controlled trials. This means we are confident that the estimated effects are attributable to shape-coding and length-coding, and not to other factors. The quality of causal evidence provided in this study is low for experiment 3 because there are not sufficient pre and post measures of the outcome. This means that we are not confident that the estimated effects are attributable to shape-coding; other factors are likely to have contributed.