Sleep attitudes as a predictor of sleep outcomes: a secondary data analysis

Figures & data

Table 1. Bivariate Correlations Between Predictors and Outcome Variables.

	%(N)	M(SD)	1	2	3	4	5	6	7
1. Age		33.31(9.86)	–
2. Gender^a (% female)	41(70)		.23**	–
3. Race^b (self-identified minority)	32(54)		.19*	−.02	–
4. SES		4.76(1.76)	.07	−.12	.07	–
5. Sleep Attitudes		5.14(.80)	.24**	.21**	.18*	.02	–
6. Sleep Hygiene^1		78.21(24.41)	−.10	−.07	−.22**	−.04	−.58**	–
7. Sleep Duration^2		7.35(1.47)	−.07	−.11	.14	.11	.18*	−.20**	–
8. Sleep Quality^3		5.55(3.78)	−.01	.11	−.16*	−.05	−.35**	.60**	−.49**

Note: ¹Higher scores indicate poorer sleep hygiene; ²Higher scores indicate longer sleep duration; ³Higher scores indicate poorer sleep quality. A score of 5 or greater for sleep quality indicates a ‘poor’ sleeper.

*p < .05; **p < .01. ^aMales = 0, Females = 1; ^bSelf-identified racial/ethnic minority = 0, Self-identified white = 1. Correlations between dichotomous (race, gender) and continuous variables are point-biserial.

Figure 1. Proposed moderated mediation model of the indirect effect of sleep attitudes on sleep outcomes via sleep hygiene, with demographic variables moderating the direct effect of sleep attitudes on sleep hygiene and the indirect effect of sleep attitudes on sleep outcomes (Hayes, 2018, Model 8).

Table 2. Test of Moderated Mediation Effect for Sleep Duration.

Moderator variable	Conditional indirect effects	Effect	SE	BootLLCI	BootULCI
Age
	25 years (−1 SD)	−.15	.07	−.31	−.02
	31 years (Median)	−.13	.06	−.27	−.02
	42 (+1 SD)	−.10	.05	−.22	−.01
	Age Index	.01	.01	−.01	.01
Gender
	Males	−.15	.08	−.31	−.01
	Females	−.08	.05	−.19	.01
	Gender Index	.07	.06	−.01	.20
Race
	Racial/ethnic minorities	−.14	.07	−.29	−.03
	Whites	−.14	.07	−.29	−.03
	Race Index	.01	.05	−.10	.09
SES
	Lower SES (−1 SD)	−.10	.05	−.21	−.01
	Median SES	−.13	.06	−.27	−.02
	Higher SES (+1 SD)	−.18	.08	−.35	−.02
	SES Index	−.02*	.01	−.05	−.001

Note: N = 172. *p < .05. Effect = unstandardized regression coefficient; Standard errors and bootstrapped confidence interval limits (BootLLCI, BootULCI) have been biased-corrected with a 10,000-resample bootstrap method (see Hayes, 2018).

Table 3. Test of Moderated Mediation Effect for Sleep Quality.

Moderator variable	Conditional indirect effects	Effect	SE	BootLLCI	BootULCI
Age
	25 years (−1 SD)	−1.36	.28	−1.95	−.86
	31 years (Median)	−1.24	.23	−1.71	−.83
	42 (+1 SD)	−1.02	.26	−1.56	−.53
	Age Index	.02	.02	−.02	.06
Gender
	Males	−1.49	.26	−2.02	−1.01
	Females	−.78	.26	−1.32	−.27
	Gender Index	.71*	.31	.14	1.36
Race
	Racial/ethnic minorities	−1.31	.30	−1.91	−.71
	Whites	−1.19	.26	−1.74	−.73
	Race Index	.11	.34	−.59	.75
SES
	Lower SES (−1 SD)	−.93	.26	−1.47	−.46
	Median SES	−1.26	.21	−1.71	−.87
	Higher SES (+1 SD)	−1.59	.27	−2.11	−1.06
	SES Index	−.17*	.08	−.31	−.01

Note: N = 172. *p < .05. Effect = unstandardized regression coefficient; Standard errors and bootstrapped confidence interval limits (BootLLCI, BootULCI) have been biased-corrected with a 10,000-resample bootstrap method (see Hayes, 2018).

Figure 2. N = 172, *p < .05, **p < .01. Path models highlighting the relationship between sleep attitudes and sleep outcomes (duration and quality) via sleep hygiene. Coefficients for the c and c’ paths represent the total effect and direct effect of IV on DV, respectively. The direct effect is provided in parentheses.

Hayes, A. (2018). Introduction to mediation, moderation, and conditional process analysis (3rd ed.). New York: Guilford Press.

 Google Scholar

Data availability statement

The authors plan to provide full publicly available access to the present study’s data should the paper be accepted for publication.