Skip to main content

Table 2 Analysis results

From: Prioritising health service innovation investments using public preferences: a discrete choice experiment

  MNL Scale-adjusted latent class
Parameters    Class-1 (54%) Class-2 (34%) Class-3 (12%)
  coef. s.e. coef s.e. coef s.e. coef s.e.
ASC −1.270* 0.076 −1.172* 0.133 −2.146* 0.208 0.549** 0.247
Cost −0.031* 0.002 −0.036* 0.003 −0.036* 0.003 −0.036* 0.003
Target people:         
People with disability 0.590* 0.065 0.832* 0.146 0.086 0.084 0.479** 0.225
People with drug addiction −1.570* 0.083 −2.232* 0.231 −0.442* 0.135 −1.775** 0.715
People with cancer 1.560* 0.119 2.017* 0.277 0.645* 0.171 1.706* 0.371
People with mental health problems 0.310* 1.728 0.677* 0.154 −0.070 0.134 −0.622** 0.271
People with obesity −1.150* 0.078 −1.583* 0.161 −0.450* 0.107 0.094 0.276
People with asthma 0.260* 0.085 0.289** 0.120 0.231*** 0.138 0.118 0.256
Target age:         
Young (less than 18) 0.090** 0.040 0.175* 0.054 0.036 0.046 −0.422** 0.212
Adults (18–65) 0.180* 0.041 0.274* 0.068 0.071 0.051 −0.175 0.166
Elderly (more than 65) −0.270* 4.300 −0.449* 0.074 −0.107** 0.052 0.597* 0.163
Implementation time:
Between 6–12 months −0.027 0.038 0.027 0.046 −0.075*** 0.042 0.164 0.132
More than 12 months −0.029 0.038 −0.036 0.046 0.015 0.040 −0.254 0.163
Between 0–5 months 0.056 1.604 0.009 0.045 0.060 0.044 0.090 0.123
Whether it works:         
It works but not scientifically proven −0.250* 0.041 −0.201* 0.060 −0.194* 0.052 −0.417** 0.164
It works and scientific studies confirm this 0.210 5.612 0.175* 0.046 0.113** 0.045 0.295** 0.125
Experts say it works elsewhere in the NHS 0.040 0.039 0.026 0.051 0.081*** 0.044 0.122 0.131
Potential health benefits:
Moderate health (50%) −0.360 −8.731 −0.283* 0.060 −0.261* 0.056 0.015 0.130
Good health (75%) 0.110* 0.037 0.061 0.041 0.072** 0.037 −0.011 0.122
Best health (100%) 0.250* 0.038 0.222* 0.061 0.189* 0.050 −0.004 0.121
Scale   λ1 = 1, λ2 = 2.65 (s.e. = 0.375)
Probabilities         
Class size    0.541* 0.040 0.338 0.037* 0.121* 0.026
Scale λ1    0.652* (s.e. = 0.086)
Log-likelihood −2704.018 −2383.155
Pseudo R-squared 0.20 0.39
Number of parameters 15 47
Number of observation 3000 3000
  1. Variables effects-coded. For clarity, we also present the baseline levels, which are arbitrarily chosen. *significant at p < 0.01; **significant at p < 0.05, ***significant at p < 0.10.