LATEN VARIABLES PATH NONPARAMETRIC ANALYSIS: ESTIMATING FUNCTIONS AND TESTING HYPOTHESES

This study is to identify patterns of relationships between manifest variables and latent variables. Parametric path analysis is one of the proper analytical techniques used to form unknown curves. The phenomenon states it is difficult to get a curve shape or curve shape is unknown. Nonparametric path analysis is used to estimate the function with the truncated spline approach which contains parameters of polynomial degrees and knots. The case of post-mining revegetation technique is used as a case study because the pattern of relationships between variables is not linear. PCA is carried out to measure the indicators of each variable, so that the contribution of component variables and variables used are reflected by the strongest indicators. The results showed that the best model of path analysis with linear polynomial degrees and knot point 3 with GCV values of 1738,303.