Table 2.

Arithmetic ability test performance predictions (odds ratios for making a mistake, and response times) based on sleep condition and time-of-day in the final mixed-effects model.

	Mistake			RT
Fixed-effect predictors	Odds ratios	CI	p	Estimates	CI	p
(Intercept)	0.04	0.03–0.05	<0.001	5,610.75	5,392.14–5,829.35	<0.001
Presentation order	1.11	1.10–1.13	<0.001	−32.38	−39.15 to −25.61	<0.001
TSD condition at baseline (22:30)	0.81	0.58–1.13	0.21	122.00	−129.59–373.60	0.34
Session 1 (08:00)	1.02	0.81–1.28	0.90	90.74	−48.81–230.28	0.20
Session 2 (12:30)	0.86	0.68–1.08	0.18	−188.52	−323.67 to −53.36	<0.01
Session 3 (16:30)	0.74	0.59–0.93	<0.01	−300.16	−434.82 to −165.49	<0.001
TSD	1.32	1.04–1.68	0.02	392.82	150.26–635.38	<0.01
TSD:Session 1 [TSD effect compared to session 3]				−212.34	−398.36 to −26.32	0.03
TSD:Session 2 [TSD effect compared to session 3]				62.52	−118.46–243.51	0.50
Observations	11,085
Marginal R2	0.15			0.03

Note: RT, response time in milliseconds; CI, 95% confidence interval; TSD, total sleep deprivation; Marginal R² represents the variance explained by the fixed-effects predictors (Nakagawa et al., 2017). Response time effects are in milliseconds (ms). Presentation order represents the order a given trial was within the test. All rows representing categorical variables report the effect compared to a reference level. The reference level for session was the baseline session (22:30 h) therefore all other effect estimates for different levels of session compare to performance at 22:30 h. The reference level for TSD was the normal sleep group. For models including an interaction between TSD and session, the effect estimate for TSD refers to the effect of TSD on performance at Session 3. Interactions between TSD and session therefore refer to differences in the effect of TSD between the stated Sessions (i.e., 1 or 2) and Session 3.