R Code and reporting output for randomized causal inference study to answer question on whether there is bias in following directions from gendered voices.
Implicit bias surrounding gender is ever-present in society and, in recent times, more attention has been brought to the issue. The increased attention has given rise to more effort to identify gender biases and attempt to mitigate and remove them. However, even with increased attention to the issue, gender biases persist in our everyday life - from the technologies we use such as voice assistants on smartphones and the audio commands in public service announcements. A CNN article (October 2011) discusses how the choice of the gendered voice of the Siri audio assistant in Apple I-Phone smartphone, which is female in the US market, has to do with biological studies that suggest that people generally find women's voices more pleasing than men's. The same article cites historical references such as the use of female voices in navigation devices during World War II, when women's voices were employed in airplane cockpits because they stood out among the male pilots. Moreover, “telephone operators have traditionally been female, making people accustomed to getting assistance from a disembodied woman's voice.” [@griggs_2011] Recent media reports also cite cases of women in high positions deepening their voices to sound more baritone in order to convey authority (for example, the case of Theranos CEO Elizabeth Holmes) [@hesse_2019]. With this motivation, we pursued identifying the existence of gender bias in giving directions through an experimental study.
The experimental study attempts to isolate the effects that a speaker’s gender has on the willingness of a targeted listener to follow through on the direction of the speaker. The purpose is to provide insight on whether the verbal directions of men are more effective than that of women in steering behavior though some simple game playing. The experiment should serve to examine the existence of the effect and the strength of the effect, if it exists.
Is there a gender bias in the response to audio (voice) directions?
Randomized control trials to assess response to genderized audio commands.