Analysis of Auditory Measures in Normal Hearing Young Male Adult Cigarette Smokers Using Multiple Variable Selection Methods with Predictive Validation Assessments

Table 2

Best-Subsets Linear Regression Coefficients of five optimal predictors (selected from 24 original predictors) of carbon monoxide level based on N = 16 cases.

Predictor	Estimate	StandardError	𝑡 value	Pr(>|𝑡|)	lmg statistic with ranking of importance in predicting outcome measured
Intercept	32.3172	1.8146	17.809	6.65e-09 ∗∗∗
RTABRPV80A	11.8526	4.5424	2.609	0.026071 ∗	0.1053425-D
LTABRPV80A	−61.7999	4.8624	−12.710	1.70e-07 ∗∗∗	0.2971096-B
RTABRPV40A	92.3103	5.6452	16.352	1.52e-08 ∗∗∗	0.3937055-A
RTDP	0.6637	0.1181	5.618	0.000222 ∗∗∗	0.0572460-E
LTDP	−1.4257	0.1549	−9.203	3.38e-06 ∗∗∗	0.1465964-C

Note 1: Total Multiple R-squared: .7449 (74.49 % variance accounted for in CO by the predictors). Significance of probability indicated by: 0 – “∗∗∗” 0.001 – “∗∗” 0.01 – “∗”.
Note 2: All 74.5% of variance in the outcome variable (CO) is accounted for by 5 independent (non overlapping) sources of variance (optimal predictor set). Consequently, the total sum for lmg across predictors adds to 1.0.