RUDY HARTONO, 090214706 M (2004) KOMPARASI HASIL UJI KEHANDALAN (RELIABILITAS) INSTRUMEN PENELITIAN DENGAN BEBERAPA TEHNIK BELAH DUA. Thesis thesis, UNIVERSITAS AIRLANGGA.
|
Text (ABSTRAK)
gdlhub-gdl-s2-2006-hartonorud-734-tkm_20-05.pdf Download (669kB) | Preview |
|
|
Text (Halaman Depan)
35811_Part1.pdf Download (533kB) | Preview |
|
|
Text (Fulltext)
35811_Part2.pdf Download (1MB) | Preview |
|
|
Text (Lampiran)
35811_Part3.pdf Download (2MB) | Preview |
Abstract
This research was intended to compare the results of reliability_tests on research instrument using split-half technique, i.e., the confirmatory factor analysis a part of multivariate analysis, and conventional method, which is part of univariate analysis. This was an applied research. The application of confirmatory factor analysis was aimed to test the reliability of research instrument and to compare with conventional method, Spearman Brown coefficient, Guttman coefficient and Pearson correlation coefficient along with a simulation of 36 research instruments. and 30 cases using split-half technique. The findings of the research showed that confirmatory factor analysis, which belonged to multivariate analysis, was more complicated and although there were more positive matrix elements, the negative ones were not eliminated and could produce negative reliability. Similar findings were also found in the case of lambda value, which is negatively dominant like that found in even odd split-half technique. Whereas, in conventional method, which was a univariate analysis, the analysis technique was easier, as it only added up all values so that the time variables having negative correlation would be neutralized by the process of adding up where the positively dominant value would produce positive reliability coefficient. The findings also showed that there were t and r values which were not suitable for confirmatory factor analysis, since the model was not convergent because it could not respond to the matrix that would be analized so that the parameter estimate remained an intermediate solution. There was a small case in confirmatory factor analysis where the result of reliability was not obtained due to not convergent model and very small sample. This did not happen with the conventional method. By adding up in pairs, the variables and items became one, so that problems would not be found in dealing with small number of cases. In conclution, there is a difference between the results of reliability test in some analyses. There are significant differences in the conventional method than the confirmatory factor analysis where there is a negative correlation value. In the conventional method all reliability test values are positive. In the confirmatory factor analysis the more items, the bigger cases are needed. Therefore; in same analyses there is no complete result, whereas in conventional method larger number of cases are not required.
| Item Type: | Thesis (Thesis) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Additional Information: | KKC KK TKM 20/05 Har k | |||||||||
| Uncontrolled Keywords: | Reliability, Split-half, Confirmatory Factor Analysis, Spearman Brown, Guttman, Pearson. | |||||||||
| Subjects: | H Social Sciences > HA Statistics > HA1-4737 Statistics R Medicine > R Medicine (General) > R735-854 Medical education. Medical schools. Research R Medicine > RA Public aspects of medicine > RA1-1270 Public aspects of medicine > RA421-790.95 Public health. Hygiene. Preventive medicine |
|||||||||
| Divisions: | 10. Fakultas Kesehatan Masyarakat > Magister Ilmu Kesehatan Masyarakat | |||||||||
| Creators: |
|
|||||||||
| Contributors: |
|
|||||||||
| Depositing User: | Nn Anisa Septiyo Ningtias | |||||||||
| Date Deposited: | 2016 | |||||||||
| Last Modified: | 02 Jul 2017 22:41 | |||||||||
| URI: | http://repository.unair.ac.id/id/eprint/35811 | |||||||||
| Sosial Share: | ||||||||||
Actions (login required)
 |
View Item |