Abstract:
The objective of the study is to explore the MAUP in spatial perspective.
Openshaw (1984) noted that the results of the analysis of aggregated spatial data will
depend on the scale and zoning being used. These two factors potentially affect analysis
of spatially aggregated data and are referred to as the modifiable areal unit problem or
MAUP. The MAUP can distort an interpretation of result of the analysis. For
example, Minot and Baulch (2005) indicated that the most poor people do not
live in the poorest districts but in the two lowland deltas, where poverty incidence
is intermediate.
Census and survey often provide a main sources of small area statistics. But
analyzing of census data often ignores spatial characteristics and usually assumes
that observations are distributed independently and identically (IID). This may
lead to incorrect results. One observation may be influenced by others at nearby
locations. As a result, census data may show inter-dependence between
observations, or the IID assumptions are not be appropriate.
The study exercises on characteristics of poverty, employment, unemployment,
lapor participant rate, literacy rate, and some others socio-economics indicators.
The data sources are come from Australian Bureau of Statistics, Indonesian
Statistical Office, and Vietnam Statistical Office. The poverty data were obtained
from The Poverty Mapping Project at CIESIN (The Center for International Earth
Science Information Network). This institution maintain poverty data of several
country and provide as well some report on the study of poverty and other related
material (Center for International Earth Science Information Network, 2006).
Poverty is one of the welfare characteristics, which is a multidimensional
phenomena within a society. The basic idea of measuring poverty is to assess the
poverty status of each individual within the region and also constructing an index
of poverty using the available information on the poor (Sen, 1976). The Foster
Greer and Thorbecke index is used in this study. Other measurements are used
such as Socio Economic Index for Area - SEIF A and also an index derived by
geographical weighted regression. The SEIF A may be used as an alternatif
measured of the socio-economic index. The geographical weighted regression
method also give a better localize estimates of the index across the geographical
regions.
The semivariogram models are developed to portray spatial distribution at each
level. Some interesting results come out from this study, one of them is the
nugget and range of the semivariogram model may give a direction of
understanding spatial distribution at the unobserved scale, for example subdistrict
level. The semivariogram model also provides a better information than other
measure of spatial autocorrelation, i.e. Moran coefficient. The semivariogram
model's parameters gives information about variance (si11), inter-dependency in
term of geographical distance (range) and also the nugget effect, which discloses
an interesting information about a latent inter-relationship of the characteristics
within a particular area.