Men producing /s/ with higher first spectral moment (M1) are more likely to be perceived as gay, yet it is unclear if M1 differs in production. Inconsistent results might be caused by inherent change in M1 over time. Therefore, we explored M1 change over time and tested if the length and location of the analysis window affects results on gay-straight differences. 37 gay and 29 straight male speakers of Australian English produced two /s/ tokens in continuous speech. M1 was extracted in each quarter of /s/ to explore change over time and in three windows: the entire fricative, the mid-50 ms, and the third quarter. Gay and straight men produced lower M1 in the first and last quarters relative to the midpoint. Despite the M1 change over time, we found no effect of analysis window on gay-straight differences, as gay men consistently showed higher M1. The lack of effect of analysis window on M1 is attributed to the overlap between the analysis windows caused by the duration of /s/.