Screening For Dichotic Acoustic Context And Headphones In Online Crowdsourced Hearing Studies

Abstract

Experimental evidence suggests that crowdsourced online experiments, where suitable, may produce better data than in-lab studies. The absence of a reliable screening method for headphones and dichotic auditory context (perfect separation of the stereo channels) is one of the main reasons why online crowdsourcing is rarely possible for auditory studies. As the evidence demonstrates, the responses to the questions “Are you wearing headphones” and “Are the headphones stereo channels separated” are providing unreliable and even conflicting results. Here we show that the interference beating phenomenon can be used as a screening method for dichotic context with satisfactory accuracy. We collected data through an in-lab experiment to test the method’s performance against the reference (the truth), avoiding the uncontrolled biases of the online experiments, achieving Cohen’s Kappa of 0.79 (95% CI, [0.52, 1.06], p<0.001), yielding “Substantial agreement” when calculated over the whole sample, and Cohen’s Kappa of 1 (95% CI, [1, 1], p=0.001), yielding “Almost perfect agreement” when calculated only over the true dichotic cases. Also, we collected data by using the only other method found in the literature that attempts screening for headphones usage, to compare both methods over the same participants and auditory contexts. The usage of the new method is tested in a crowdsourced setting, involving over 2000 online participants. The in-lab and online results suggest that the method introduced in this study is suitable, and therefore, an enabler of auditory online crowdsourced studies.