Comparing online versus in-person auditory experiments on talker adaptation

Abstract

Online experiments are increasingly used as an alternative to in-person lab-based experiments. While online experimental platforms offer convenience, conducting auditory experiments online also presents several challenges, including variations in latency and jitter in auditory stimulus presentation across participants’ devices. Therefore, it is important to carefully examine whether response time data collected online are comparable to the reliability and precision of data collected in well-controlled laboratory settings. Here, we directly compared whether well-documented context- and talker-adaptation effects on response times are replicable between the laboratory and an online platform. Participants (N = 58) performed an established word identification task measuring response time in a 2 × 2 manipulation of context (isolated words versus continuous speech) and talker variability (single versus mixed talkers). The results from the laboratory session, but not the online platform, reliably replicated all main and interaction effects found in prior laboratory-based studies. While mean response times were comparable between environments, the variance of online data was significantly greater than in the lab, even after discarding outliers. These results suggest that careful considerations are needed when implementing online experiments that depend on precise auditory timing, particularly in considering how greater measurement error will impact power, sample sizes, and statistical inferences.