2. Between explanatory variables there explanatory variables which can be seen in should be no multicollinearity: to the Appendix 1. extent that one independent is a linear function of another independent, the Method problem of multicollinearity will occur The methods used in this research were: in , as it does in OLS 1. preparation. regression. As the correlation among This step consist of selecting regencies each other increase, the standard errors with backward region status namely fairly of the logit (effect) coefficients will backward, backward, very backward and become inflated. Multicollinearity does the most backward regions. not change the estimates of the 2. Early data description. coefficients, only their reliability. High 3. The assumption of a logistic regression standard errors flag possible examination. multicollinearity (www.chass.ncsu.edu). 4. Data analysis. Analyze selected data with ordinal logistic Biplot Analysis regression. This analysis is conducted for Biplot similarity provides plots of the n each sub criteria of determining backward observations, but simultaneously they give region status. plots of positions of the p variables in two 5. Determine the prior factors that influence dimensions. Furthermore, superimposing the backward region status. two types of plots provides additional 6. Significant variables were further analyzed information about relationships between through biplot and then explain the variables and observations not available in relationship of these variables based on either individual (Jolliffe, 2002). globally and part of regions (west and The plots are based on the singular value east). decomposition (SVD). This state that the (n x The Software used in this research are p) matrices X on observations on p variables Microsoft Excel 2007, Minitab 14, SPSS 13 measured about their sample can be and SAS 9.1. written RESULTS AND DISCUSSION ׳X = ULA where U, A are (nxr),(pxr) matrices respectively, each with orthonormal columns, Early Description L is an (rxr) diagonal matrix with elements According to the data released by KNPDT, 1/ 2 1/ 2 1/ 2 t1 ³ t1 ³ ...³ tr , and r is the rank of there are 434 regencies in Indonesia. KNPDT X. has determined five categories of region index To include the information on the and status based on six major criteria, such as variables in this plot, we consider the pair of (1) economic, (2) human resources, (3) eigenvectors. These eigenvectors are the infrastructures, (4) regional finance, (5) coefficient vectors for the first two sample accessibility, and (6) characteristic of region. principal components. Consequently, each Each criteria has indicators which are relevant row of matrix positions a variable in the to measure the criteria score. Then the GoI graph, and the magnitudes of the coefficients calculated region score with giving weight for (the coordinates of the variable) show the each criteria based on their experiences and weightings that the variable has in each then multiply it with standardized data. principal component. The positions of the variables in the plot are indicated by a vector.

MATERIAL AND METHODS

Source of Data The data used in this study were collected from the KNPDT. These data were derived from data Potensi Desa (Podes) 2005 and Survei Sosial Ekonomi nasional (Susenas) 2006 conducted by Central Bureau of Figure 1. The number and percentage of (CBS). The data consists of five regency with each status categories as response variable and 33 Regencies with advance status were not used in this analysis because this research