Skip to main content
. 2019 Aug 12;8(8):1203. doi: 10.3390/jcm8081203

Table 4.

Dynamic Model of Regression Analyses.

Probit Logit Mixed Model Likelihood
2002 −0.645 *** −1.139 *** −0.77 ** −0.64 **
(−3.73) (−3.66) (−4.27) (−4.68)
2003 −0.086 −0.151 −0.09 −0.087
(−0.60) (−0.58) (−0.54) (−0.65)
2004 0.026 0.044 0.13 0.13
(−0.22) (−0.2) (0.95) (0.95)
2006 0.226 ** 0.397 ** 0.319 ** 0.377 **
(2.07) (2.01) (2.44) (2.44)
2007 −0.038 −0.071 −0.097 −0.009
(−0.31) (−0.33) (−0.07) (−0.07)
2008 0.055 0.083 0.087 0.08
(0.42) (0.36) (0.61) (0.60)
Chi 192.846 180.525 55,741.766
p value 0.00 0.00 0.00
AIC 5800.953 5804.472 5997.446
BIC 6040.738 6044.257 6237.198

Note: Omitted year for baseline comparison is 2005, when the NHI payment policy was changed. Numbers in parentheses are z-values. Likelihoods are estimated from odds ratio based on the results from the mixed model. Likelihoods estimated by the logit model generate similar results; thus, they are not reported due to space considerations. Full regression results are not reported since all three models generate comparable results as in Table 3. **, *** represent the 0.05, and 0.01 statistical significance level, respectively.