Table 3.

Differences in sleep measures by race and by education and the percentage explained by sound levels

	Racial Groups		Education Groups
	Model 1	Model 2	Model 1	Model 2
Total Sleep Time (minutes)	−41.8 (−78.3, −5.3)	−40.8 (−78.7, −2.9)	−5.3 (−44.1, 33.6)	n/a
% attenuation	–	2.4%	–	–
N2 (minutes)	7.9 (−26.5, 42.3))	n/a	−29.0 (−62.9, 5.0)	n/a
% attenuation	–	–	–	–
N3 (minutes)	−23.4 (−40.0, −6.9)	−22.0 (−39.6, −4.4)	18.9 (1.7, 36.2)	16.9 (−1.2, 35.1)
% attenuation	–	6.0%	–	10.6%
REM (minutes)	−35.1 (−55.1, −15.1)	−30.0 (−49.9, −10.2)	8.4 (−14.1, 30.9)	n/a
% attenuation	–	14.5%	–	–
N3 (%)	−4.0 (−8.3, 0.3)	n/a	5.0 (0.7, 9.2)	4.7 (0.1, 9.2)
% attenuation	–	–	–	6.0%
REM (%)	−5.7 (−9.5, −1.9)	−4.5 (−8.2, −0.9)	2.4 (−1.7, 6.5)	n/a
% attenuation	–	21.1%	–	–

All values are reported as estimate (95% CI) for either the race coefficient (Black v. White) or education coefficient (more than a college degree vs less education) in models for each sleep outcome

Only models where the sleep measure significantly differed by race or by education are included in Model 2

Model 1 includes racial or education group, age, and sex. Model 2 adds average sound level across the night (except for N3/N3%, which used average sound level for first 3 h) to Model 1

Percent attenuation is calculated as the percent reduction in the race coefficient estimate relative to the estimate in Model 1