Tintin Sukartini (2005) Komparasi Antara Analisis Conjoint Data Non Metrik dengan Skala Peringkat dan Skala Likert untuk Menentukan Ketepatan Prediksi Dimensi Kepuasan Pasien Dalam Pelayanan Keperawatan di Ruang Tropik RSU Dr. Soetomo Surabaya. Thesis thesis, UNIVERSITAS AIRLANGGA.

Text (FULLTEXT) KKA KK TKM_2005_TINTIN SUKARTINI.pdf Download (17MB)

Abstract

Conjoint analysis is a multivariate technique to estimate an individual's perception on an object that comprises one or several parts. The final result of conjoint analysis is the utility function of levels in each attributeland individual level part worth. Predictive accuracy is an important factor in estirriting the utility function of each attribute. The purpose of this study was to analyze the difference of predictive accuracy of conjoint analysis on non-metric data from stimuli estimated in rank scale and Likert scale on the dimension of patients' satisfaction in nursing care. Each respondent in male and female Tropical Wards, Dr Soetomo Hospital, Surabaya, estimated the dimension of patients' satisfactory in nursing care for data in rank as well as Likert scales. Data estimated in rank scale for respondent 1 revealed Kendall's Tau correlation coefficient value of 1 and significance of p = 0.0003, and aggregate results revealed Kendall's Tau correlation coefficient value Hof 0.929 and significance value of p —0.0006. Data estimated in Likert scale kir respondent 1 revealed Kendall's Tau correlation coefficient of 0.853 with significance of p 0.0076, and aggregate results revealed Kendall's Tau coefficient' correlation of 0.964 and significance of p = 0.0005. High Kendall's Tau correlation coefficient indicated a string correlation between rank estimates part worth and actual estimates part worth or high predictive accuracy from the respondents in determining the dimension of satisfaction in nursing care. Results revealed that 92% of predictive accuracy in rank scale had lower significance value compared to that of ykert scale. In aggregate results, predictive accuracy of rank scale (p = 0.0006) and Likert scale (p = 0.0005) was relatively not different, showing that both scales had high predictive accuracy in determining the dimension of patients' satisfaction in nursing care.

Actions (login required)

View Item