FAKTOR-FAKTOR DETERMINAN HUMAN DEVELOPMENT INDEX KABUPATEN/KOTA DI JAWA: PENDEKATAN GEOGRAPHICALLY WEIGHTED REGRESSION

Raden Witcaksono Setyo Putro, Prof. Dr. R. Rijanta, M.Sc

2012 | Tesis | S2 Magist.Prnc.Kota & Daerah

Human Development Index (HDI) merupakan indeks komposit yang mengukur rerata capaian tiga dimensi utama pembangunan manusia: usia yang panjang dan sehat, pengetahuan, serta standar kehidupan yang layak. Dalam beberapa dekade terakhir, capaian pembangunan manusia di Indonesia telah meningkat secara signifikan. Namun, peningkatan tersebut juga diikuti dengan adanya disparitas pembangunan manusia antarwilayah, termasuk antar kabupaten/kota di Jawa. Jawa merupakan pulau terpadat di Indonesia, dimana dengan wilayah 6% luas Indonesia, namun dihuni oleh 58% penduduk Indonesia dan memiliki kontribusi dalam 58% ekonomi Indonesia. Penelitian ini memiliki dua tujuan: Pertama, mendeskripsikan pola sebaran HDI di Jawa dan menguji dependensi spasial dengan menggunakan autokorelasi spasial. Kedua, menerapkan model Geographically Weighted Regression (GWR) untuk menguji hubungan antara HDI dan faktor-faktor yang mempengaruhinya. GWR merupakan pengembangan metode regresi yang mempertimbangkan unsur spasial. Data yang digunakan adalah data Potensi Desa Tahun 2008, dengan unit analisis kabupaten/kota. Data belanja APBD kabupaten/kota diperoleh dari Kementerian Keuangan RI. Untuk data spasial, digunakan Peta Rupa Bumi dari Bakosurtanal dan peta digital elevation model (DEM) dari United States Geological Survey (USGS). Analisis pola sebaran HDI menggambarkan terjadinya disparitas capaian pembangunan manusia antar kabupaten/kota di Jawa. Skor autokorelasi spasial yang bernilai positif juga menunjukkan adanya klaster sebaran HDI, yang sejalan dengan Hukum Geografi Pertama (Tobler, 1970). Dengan menggunakan pendekatan GWR, diperoleh bahwa rasio sarana pendidikan, persentase buruh tani, dan jarak ke sekolah terdekat berbanding terbalik dengan capaian HDI. Sebaliknya, rasio sarana kesehatan dan persentase keluarga pengguna listrik berbanding lurus dengan capaian HDI. Hasil estimasi parameter GWR menunjukkan adanya hubungan bervariasi antarwilayah, yang dapat digambarkan dalam bentuk peta (spasial). Dibandingkan dengan metode OLS, pendekatan GWR terbukti lebih baik dalam memodelkan faktor deterninan HDI, yang ditunjukkan dengan kriteria kebaikan model. Galat pada regresi GWR juga bebas dari autokorelasi spasial, yang merupakan salah satu syarat terpenuhinya asumsi regresi, dimana hal tersebut tidak dipenuhi oleh model regresi OLS. Lebih lanjut, penelitian ini juga menemukan bahwa rendahnya capaian HDI di Madura dan Tapal Kuda tidak serta merta disebabkan oleh buruknya kondisi sarana dan prasarana.

The Human Development Index (HDI) is a composite index that measures the average achievements of a region in three basic dimensions of human development: a long and healthy life, knowledge, and a decent standard of living. In last decades, progress of human development in Indonesia has been very impressive. However, there are also clear disparities of human development between regions, including between regencies and cities in Java. Java is the most populated island in the world, with only 6% of the total area of Indonesia, but inhabited by 58% of Indonesian population, and contribute 58% of Indonesia’s economy. Firstly, this paper examined human development patterns in Java. We performed a spatial autocorrelation measures to assess the level of spatial dependency. Secondly, we developed a model using Geographically Weighted Regression (GWR) to estimate the strength of relationship between HDI and factors associated. GWR is an extended development of regression method that incorporates spatial structure. The data used were the 2008 Village Census (Podes) from the BPS at regency/city-level, which were aggregated from villagelevel. The district’s budget data was compiled from Indonesian Ministry of Finance. For spatial data, we use administration map from National Survey and Mapping Coordination Agency (Bakosurtanal) and digital elevation map from USGS. Examining spatial structure of the HDI, the results show prominent disparities on HDI between regions of Java. Strong patterns of spatial association were found proving the presence of clusters on the distribution of HDI, corresponding to Tobler’s First Law of Geography. Using GWR, we found that educational infrastructure ratio, farm workers, and distance to the nearest school, were negatively associated with the HDI. Meanwhile, health infrastructure ratio, and percent of family with electricity were positively associated with the HDI. The results of the GWR model were compared to the global model. GWR can help better understand which predictors are associated at specific locations. The GWR maps produced were not only confirmed, but also demonstrated the spatial varying association. We found that the GWR performs better to model HDI determinants than the OLS model, shown with goodness-of-fit statistics. GWR’s errors were also free from spatial autocorrelation; one of regression assumptions which was not met by the OLS model. Additionally, we found that the factors shaping the HDI achievement were different between regions; and furthermore, the low HDI in the eastern tip of Java and Madura Island were not necessarily caused by the poor infrastructure.

Kata Kunci : Human Development Index (HDI), perkembangan wilayah, autokorelasi spasial, regresi lokal, Geographically Weighted Regression (GWR)