To calculate a sample size estimate we used mean effect sizes (Cohen’s D = 0.55) reported in a study examining task-evoked pupil diameter while participants interpreted visual stimuli of varied difficulty [36 (link)]. With α = 0.05 and power = 0.95, a minimum sample size of 38 participants is advised. To increase our statistical power, we collected data from 89 pathologists as part of a larger study examining how resident physicians’ diagnostic expertise and eye movements change through their specialized training.
To reduce sampling bias, we intentionally recruited a geographically and experientially diverse sample of pathologists. This included recruiting participants from nine major university medical centers distributed across the United States (in eastern, western, northern, and southern regions of the country); the participants held highly varied experience levels, including 70 resident pathologists and 19 experienced (faculty) pathologists. While we cannot control which pathologists chose to participate in our study or guarantee that our results will generalize to other groups of pathologists or other specialized domains of medicine, we are confident that our recruitment and data collection procedures reduced selection bias. Sample characteristics are detailed inTable 2 .
All participants provided written informed consent, and all study procedures were approved by the appropriate Institutional Review Boards (IRB), with the University of California, Los Angeles acting as the IRB of record (Protocol #18–000327).
To reduce sampling bias, we intentionally recruited a geographically and experientially diverse sample of pathologists. This included recruiting participants from nine major university medical centers distributed across the United States (in eastern, western, northern, and southern regions of the country); the participants held highly varied experience levels, including 70 resident pathologists and 19 experienced (faculty) pathologists. While we cannot control which pathologists chose to participate in our study or guarantee that our results will generalize to other groups of pathologists or other specialized domains of medicine, we are confident that our recruitment and data collection procedures reduced selection bias. Sample characteristics are detailed in
All participants provided written informed consent, and all study procedures were approved by the appropriate Institutional Review Boards (IRB), with the University of California, Los Angeles acting as the IRB of record (Protocol #18–000327).
Free full text: Click here
