The impact of childhood temperament on the development of borderline personality disorder symptoms over the course of adolescence

Table 2 Fit statistics for growth models

	χ²	df	TLI	CFI	RMSEA	RMSEA (90% CI)	AIC	BIC	Δχ²	df, Δχ²
Free curve slope intercept model	78.78***	12	0.98	0.98	0.05	0.04-0.06	41255.27	41341.26
Factor model with means	223.96***	14	0.94	0.95	0.08	0.07-0.09	41396.44	41470.97	145.18***	2
Factor model with means-shift	221.85***	13	0.94	0.95	0.08	0.07-0.09	41396.33	41476.59	143.07***	1
MANOVA	237.05***	19	0.96	0.94	0.07	0.06-0.08	41399.54	41445.40	158.27***	7
MANOVA w/o compound symmetry	258.14***	14	0.93	0.94	0.09	0.08-0.10	41430.63	41505.16	179.36***	2
Linear slope intercept model	150.17***	16	0.97	0.97	0.06	0.05-0.07	41318.66	41381.72	71.39***	4
Quadratic slope intercept model^Ep	70.83***	11	0.98	0.98	0.05	0.04-0.06	41249.32	41341.04	--	--

Note. The Free Curve Slope Intercept Model was the base model against which the other forms of growth were compared using χ² difference tests (Δχ²). The quadratic slope intercept model was not nested within the FCSI model so the χ² difference test was not conducted. ^EpEstimation problems occurred when estimating the quadratic slope intercept model: the intercept latent variable had a negative error variance. ***p < .001.

ISSN: 2051-6673