@article{Andi_Rumengan_Suhardis_2021, title={Effect Of Work Discipline, Ethics, Communication, Job Satisfaction On Employee Loyalty (Case Study Of The Public Relations Protocol Bureau Of The Regional Secretariat Of The Riau Islands Province)}, volume={3}, url={https://aptikom-journal.id/conferenceseries/article/view/471}, DOI={10.34306/conferenceseries.v3i2.471}, abstractNote={<p><em><span style="font-weight: 400;">This study uses a causal model survey method using path analysis techniques, based on the reasons this study aims to confirm the theoretical model with empirical data, the study population is employees of the Riau Islands Province Secretariat Public Relations Protocol Bureau, which consists of employees with a sample of 174 employees, because the population limited from all members of the sample population so this study uses saturated samples taken through census techniques using proportional random sampling when collecting technical data using variable measurements using a questionnaire instrument in which each employee is given five questionnaire instruments for the source of the measurement variables studied In addition to research instruments that use questionnaire, the instrument was developed based on theoretical studies, then defined in conceptual definitions, operational definitions, and developed through lattice instruments and technical techniques. Knitted data analysis uses descriptive statistics and statistical analysis to test the significance of path coefficients, descriptive statistics to present data in the form of frequency distribution tables, histograms, and total statistics such as media, modes, averages, variants, and foreign exchange standards. Statistical tests are used to test the significance of path coefficients using Partial Least Square (PLS) which is a Multivariate Analysis in the second generation using structural equation modeling (SEM). PLS can be used for a small number of samples, and of course with a large number of samples will be better able to improve the accuracy of estimates. PLS does not require the assumption that data distribution must be normal or not. The construct form can use a reflective or formative model in which from the results of statistical analysis, the relationship between variables formulated in the formulation of a problem as many as 7 pieces obtained significant results.</span></em></p> <p><em><span style="font-weight: 400;"> </span></em></p>}, number={2}, journal={IAIC International Conference Series}, author={Andi, Aludin and Rumengan, Jemmy and Suhardis, Adnan}, year={2021}, month={Jan.}, pages={123–130} }