Table 5.

Model 1: random effects generalized least squares regression examining the role of time (week effects), sociodemographics (age and gender), diary characteristics (word count and entries), and baseline well-being (moderate and high well-being, compared with low well-being) in determining the emotional tone of diary entries.^a

Covariates					Coefficient			P value			95% CI
Time
	Week 1			Referenceb			N/Ac			N/A
	Week 2			4.57			.19			−2.24 to 11.38
	Week 3			4.72			.20			−2.57 to 12.02
	Week 4			3.78			.30			−3.33 to 10.89
	Week 5			23.90d			<.001			16.89 to 30.90
	Week 6			26.62d			<.001			19.35 to 33.88
	Week 7			4.73			.19			−2.38 to 11.84
	Week 8			−3.41			.35			−10.62 to 3.79
	Week 9			4.56			.23			−2.91 to 12.03
Sociodemographics
	Gender
		Male	Reference			N/A			N/A
		Female	−2.44			.34			−7.48 to 2.59
	Age			1.98e			.008			0.51 to 3.46
Diary characteristics
	Word count			−0.01			.10			−0.02 to 0.00
	Total diary entries			0.43			.66			−1.47 to 2.33
Baseline well-being
	Low well-being			Reference			—			—
	Moderate well-being			5.03f			.046			0.08 to 9.98
	High well-being			7.48e			.009			1.84 to 13.12

^aWald chi-square, Χ²₁₅=137.0 (N=855); P<.001. This model provided an improved model fit, compared with model 0, which only included time (week effects) presented in Multimedia Appendix 3.

^bReference category for factor variables.

^cN/A: not applicable.

^dP<.001.

^eP<.01.

^fP<.05.