Skip to main content

Table 3 Maximum likelihood estimation of sample selection and ordered responses

From: Consumer response to media information: the case of grapefruit-medicine interaction

  Outcome equation (Y21)) Selection equation (Y12))
  Estimated parameters Standard error Estimated parameters Standard error
Thresholds 1 0.059 (1.348) −2.550b (0.242)
Thresholds 2 3.229b (1.059) -  
Age 0.020a (0.011) 0.013b (0.004)
Gender 0.103 (0.125) −0.029 (0.056)
Income −0.046 (0.059) 0.012 (0.026)
Education 0.174 (0.124) 0.117b (0.056)
Yes warning 1.505b (0.566) 0.798b (0.314)
No warning 0.273 (0.396) 0.208 (0.186)
Occasionally consumed 1.646b (0.539) 0.780b (0.214)
Frequently consumed 2.150b (0.669) 1.754b (0.24)
Trust -   0.012 (0.066)
Influence -   0.173b (0.067)
Search -   0.257b (0.063)
Positive tone 0.845b (0.177) -  
Negative tone −0.277a (0.167) -  
Positive frequency 0.050 (0.038) -  
Negative frequency −0.038 (0.038) -  
Old news −0.398b (0.165) -  
New news −0.505b (0.181) -  
Age*Yes warning −0.037b (0.011) −0.006 (0.006)
Age*No warning −0.007 (0.009) −0.003 (0.004)
Age*Occasionally consumed −0.035b (0.010) −0.011b (0.004)
Age*Frequently consumed −0.030b (0.011) −0.022b (0.005)
ρ 0.278 (0.457)   
LR test (H o  : ρ = 0), χ 2 (1)   270.99b
Observations   3497
Log Likelihood   −1700.55
Wald test, χ 2 (18)   169.51
  Estimated probabilities Mean Min Max
p(y1 = 1) 0.157 0.016 0.654
p(y1 = 1, y2 = 0) 0.011 0.000 0.247
p(y1 = 1, y2 = 1) 0.127 0.134 0.455
p(y1 = 1, y2 = 2) 0.019 0.000 0.496
  1. b and a indicates that estimated parameters are significantly different from zero at 5 % and 10 % levels, respectively
  2. 1) Y2 = 0 if respondents consumed less than before, Y2 = 1 if respondents did not change their consumption and Y2 = 2 if respondents consume more than before
  3. 2) Y1 = 1 if respondents have been exposed to grapefruit/grapefruit juice news over the past month