Table 6.

Summary of sensitivity analyses

Sensitivity analysis number	Purpose of sensitivity analysis	Method employed	Summary of results of sensitivity analysis
1	To adjust for multiple comparisons	When re-estimating the models in tables 4 and 5 and well as figures 1 and 2, we employed false discovery rate procedures.	In total, there were 427 p values in which 159 were significant in the original analysis. After the false discovery, p value correction, 127 of the 159 significant p values (80%) remained statistically significant (online supplemental tables 2 and 3 and online supplemental figure 2). Results were robust for comparisons between foreign-born and US-born NHW adults and for most results for comparisons with adults of Mexican and Puerto Rican heritage compared with NHW adults.
2	To investigate how results would be affected if we did not consider nativity/birthplace as a modifier of racial/ethnic differences in sleep	We combined both US-born and foreign-born participants; we then compared sleep characteristics among adults of Hispanic/Latino heritage versus NHW adults.	Combining foreign-born and US-born participants across both Hispanic/Latino heritage and NHW race/ethnicity would have missed important differences by nativity status (online supplemental table 3). For instance, the lower prevalence of non-recommended sleep duration observed among foreign-born Mexican versus US-born NHW adults (table 2) would either have been underestimated or not have been observed if participants were not stratified by birthplace.
3	To investigate how results would be affected if we considered sex/gender and age as potential modifiers39	We stratified the original models by sex/gender (men, women) and by age category (18–30 years, 31–49 years, ≥50 years), separately. In models that were also stratified by language acculturation, we combined low and medium acculturation to increase sample sizes and improve statistical stability.	After stratification by sex/gender (online supplemental table 3), point estimates were slightly stronger among men versus women for sleep quality across comparisons with foreign-born NHW adults and for very short as well as short sleep across comparisons with non-US born Mexican adults. Sex/gender did not modify the remaining associations among Mexican or Puerto Rican adults. The differences among both foreign-born NHW and Mexican adults compared with US-born NHW adults that were observed in the main analysis were greater among younger and middle versus older aged adults (online supplemental table 4). Across comparisons with non-US-born NHW adults, there was little variation by sex/gender for Mexican and Puerto Rican adults, but the differences were greater among younger versus older aged adults (online supplemental tables 5 and 6). In analyses stratified by language acculturation, lower prevalence of shorter sleep duration among foreign-born Mexican compared with NHW adults was stronger for men versus women and for younger versus older adults (online supplemental tables 7 and 8).
4	To investigate how results would be affected if we adjusted for time in the USA in the comparisons between foreign-born Hispanic/Latino heritage groups and their NHW counterparts9 19 37	Across comparisons of foreign-born Hispanic/Latino heritage groups with their foreign-born NHW counterparts, we additionally adjusted for time in the USA.	Results (online supplemental table 9) were consistent with the main analysis (table 3), which suggested that time spent in the USA was not a strong confounder across comparisons between foreign-born Hispanic/Latino heritage groups and their NHW counterparts.
5	To investigate how results would be affected if we used a different measure of acculturation in models9 19	We separated foreign-born NHW adults and Hispanic/Latino heritage groups by a different metric of acculturation, time lived in the USA (<15 years in the USA, ≥15 years in the USA),9 19 37 when compared with US-born NHW adults.	Results (online supplemental table 10) were consistent with those of the language acculturation-stratified analyses (figure 1).

NHW, non-Hispanic white.