COMPARISON OF RESAMPLING EFFICIENCY LEVELS OF JACKKNIFE AND DOUBLE JACKKNIFE IN PATH ANALYSIS

The assumption of normality is often not fulfilled. This causes the estimation of the resulting parameters to be less efficient. The problem with assuming that normality is not satisfied can be overcome by resampling. The use of resampling allows data to be applied free of distributional assumptions. In this study, a research simulation was carried out by applying Jackknife and Double Jackknife resampling in path analysis with the assumption that the normality of the residuals was not fulfilled and the number of resampling was set at 100 with the degree of closeness level of relationship between variables consisting of low, medium, and high closeness. Based on the simulation results, resampling with a power of 100 can overcome the problem of unfulfilled normality assumptions. In addition, the comparison of the relative efficiency level of the Jackknife and Double Jackknife resampling in the path analysis obtained that the Double Jackknife resampling has more efficiency than the Jackknife resampling.