Table 3.

Results from the models selected for hypothesis testing in which β₀ was estimated from the data.

Gene	n	df	β0	P(β0 = 0)	β1	P(β1 = 0)	m	LCI	UCI	R2	Model
UMOD	50	46	1.189	0.065	−0.967	1.309 × 10−14 *	0.031	−0.073	0.135	0.909	PWL
CALM2	50	46	−0.058	0.853	−0.537	5.382 × 10−5 *	−0.048 +	−0.071	−0.025	0.742	PWL
KRT14	50	46	−0.956	0.043 ◆	0.907	1.158 × 10−12 *	0.063 +	0.012	0.113	0.908	PWL
MMP1	50	46	−0.111	0.800	0.052	0.165	−0.196 +	−0.391	−0.001	0.153	PWL
EMP1	48	44	0.376	0.429	0.221	3.186 × 10−5 *	−0.093	−0.202	0.016	0.441	PWL
TR-β	50	48	−0.237	0.213	−0.044	7.650 × 10−5 *	N/A	N/A	N/A	0.281	L
TR-α	50	48	−0.060	0.803	−0.071	1.830 × 10−6 *	N/A	N/A	N/A	0.381	L
GLUL	50	48	0.005	0.987	−0.037	0.034 *	N/A	N/A	N/A	0.090	L

df is the sample size (n) minus the number of parameters estimated for a given model (for piecewise models β₀, β₁, Ψ, and β₂ = four parameters and for linear models β₀ and β₁ = two parameters). P(β₀ = 0) is the probability that β₀ = 0 and P(β₁ = 0) is the probability that β₁ = 0. LCI is the lower 95% confidence interval for m and UCI is the upper 95% confidence interval for m. Lower and upper 95% confidence intervals for m that do not overlap zero are deemed statistically different from zero (denoted by +). PWL = piecewise linear and L = linear.

^◆indicates β₀ terms that statistically differ from zero at the 0.05 level.

^*indicates β₁ terms that statistically differ from zero at the 0.05 level. R² is the proportion of variation in the data explained by the model. N/A = values that are not obtained via linear models. Results for KRT6A are not shown because attempts to fit a piecewise model that estimated β₀ failed. Units for β₀ are log₂(R). Units for β₁, and m are log₂(R)/day of T₄ treatment.