Peak startle latency (ms (SEM)), on PA trials (0 ms prepulse intervals), and 30, 60 and 120 ms prepulse+pulse trials, across the 7 COGS sites. Inset shows distinct patterns of peak reflex latency modulation in women vs. men. Despite the lack of significant site × trial type interaction for peak latency, inspection of the data revealed some inter-site variability in the patterns of latency facilitation across the different prepulse intervals. For example, peak latency for 120 ms prepulse trials was significantly slower than for 30 and 60 ms trials for some sites (A,B,C,G) but not others (D,F), and fastest latencies were detected with 30 ms prepulse trials at some sites, and with 60 ms prepulse trials at others. Neither hearing threshold nor electrode impedance correlated significantly with reflex latency on PA or any prepulse trials (all r's<0.13). No significant relationships were detected between reflex latency and either WRAT scores or ethnicity. Ambient noise levels also did not correlate significantly with mean peak latency for each site on PA (rs= -0.03, p=0.93) or on any prepulse trials (rs's= -0.24 - (-0.01); p's=0.54 - 0.97). ANOVA across all sites revealed a significant interaction of gender × trial type (F=3.33, df 3,447, p<0.02). As seen in the inset, this interaction is reflected in patterns of reflex latency across sites: sites testing predominantly men (e.g. site F) exhibited latency patterns characteristic of males in the present study (minimal difference in latency across prepulse intervals), and those testing predominantly women (e.g. site A) exhibited latency patterns characteristic of females in the present study (less facilitation with 120 vs. 60 ms prepulse intervals).