. Author manuscript; available in PMC: 2019 Jan 1.

Published in final edited form as: J Affect Disord. 2017 Aug 15;225:180–187. doi: 10.1016/j.jad.2017.08.015

Table 3.

Clinical Differences Among the Five Latent Classes

	Moderate Negative Coping Class	Disengaged Class	Reactive Coping Class	Positive Coping Class	Severe Negative Coping Class	df	F/ χ2	Partial η²
T1 NSSI	1.18 (1.55)^ac	0.45 (1.11)^b	0.98 (1.34)^ab	0.76 (1.23)^b	1.83 (1.86)^c	4	4.43**	0.04
T2 NSSI	0.47 (0.51)^a	0.11 (0.31)^b	0.23 (0.43)^ab	0.21 (0.42)^ab	0.67 (0.52)^ab	4	24.02***
T1 Dep	10.75 (6.36)^a	4.07 (4.40)^b	10.97 (6.66)^a	4.64 (4.46)^b	21.50 (10.49)^c	4	68.48***	0.35
T2 Dep	6.33 (5.46) ^a	3.15 (4.36) ^a	8.11 (7.02) ^a	3.67 (4.45) ^a	12.57 (7.85) ^a	4	0.96	0.02
T2 WB	8.73 (2.24)^a	10.21 (2.05)^b	10.06 (2.41)^b	10.54 (2.31)^b	9.30 (3.16)^ab	4	5.69***	0.09

Note. T1 NSSI = Time 1 Last-Year NSSI Frequency; T2 = Time 2 NSSI Presence; T1 Dep = Time 1 Beck Depression

Inventory; T2 = Time 2 Beck Depression Inventory; T2 WB = T2 Internal State Scale - Well-Being Subscale. ANCOVAs and post-hoc analyses were used to compare classes on our outcome measures: depressive symptoms, NSSI frequency measured at baseline, and well-being measured at follow-up. Multinomial logistic regression was used to determine whether there was a significant main effect of classes on engagement in NSSI over the follow-up period; Binary logistic regressions were used to conduct post-hoc analyses.

^a,b,c

Results for post-hoc analyses; classes with the same subscript did not significantly differ.