Variable | Coefficient^{‡‡} | SE | 95% CI | |

Travel time^{†} | 0.171^{*} | 0.013 | 0.145 | 0.198 |

Cost^{‡} | −0.291^{*} | 0.018 | −0.326 | −0.255 |

Provider^{§} | ||||

Doctor | 2.104^{*} | 0.058 | 1.989 | 2.218 |

Nurse | 1.654^{*} | 0.049 | 1.556 | 1.752 |

Good provider attitude/performance | 1.445^{*} | 0.049 | 1.349 | 1.542 |

Available drugs and medical equipment | 3.877^{*} | 0.074 | 3.732 | 4.022 |

Available transportation | 0.449^{*} | 0.027 | 0.396 | 0.502 |

n (observations) | 8045 | |||

Root likelihood^{¶} | 0.777 | |||

Per cent certainty^{**} | 77.1% | |||

p(χ^{2})^{††} | <0.001 |

↵* p<0.01.

↵† In hours.

↵‡ In 10 Birr.

↵§ In reference to health extension worker.

↵¶ With three alternatives per choice, a chance model would have a root likelihood of 1/3. A perfect model fit would have a root likelihood of 1.

↵** Indicates how much better the solution is than chance, as compared to a perfect solution. 0% corresponds to a model fit at the chance level, and 100% corresponds to a perfect model fit. This corresponds to a log likelihood of −2029.87.

↵†† The probability that the fully specified model is not different from a null model (in which all attribute coefficients are zero).

↵‡‡ Mean of individual coefficients.