Interrelationship among the Gender Indicators across Districts Before proceeding further in terms of unfounded descriptive conjectures, which rule the major market of gender literature in general, let us try to understand district-wise linkages of different indicators used in this study in order to have a more clear understanding of the phenomena of gender disparity. The relevant correlation matrices for rural and urban areas of the districts for 2001 and also for the pooled data set are presented in Table 7(a), 7(b), 8(a) and 8(b).As evident from Table 7(a) and 7(b), only six correlation coefficients for rural areas and five for urban areas appear to be statistically significant, which might cause some problem for OLS estimation. But a deeper scrutiny removes these possibilities. For example, female and male literacy rates should be naturally high, and can not be treated as substitutes. Sex ratios for infant and all ages also have common base, and are not jointly used in the following estimation. Female literacy and purchasing power are moderately high in rural areas, but they are also not substitutes. Third, higher purchasing power is related to lower poverty, but they do not contain any direct gender bias. The same is true for poverty and regional dummy. The only case, which is to be mentioned explicitly, relates to purchasing power and regional dummy for both rural and urban areas. Note that regional development dummy is derived directly from above average values of purchasing power. So they are not selected simultaneously in any single OLS. For urban areas, female main workers and female WPR are not chosen jointly in any set. But inverse relation of female literacy with Hindu female population share and positive relation with Christian female population are themselves new findings of the present study. As will be clear from these correlations, it is the urban areas, which pose real threat against gender equality in India. This will be further substantiated in the OLS analysis. As far as dependent variables are concerned, namely sex ratio and female work participation rate, there does not appear to have any unwarranted correlation problems. In the pooled data set presented in Table 8(a) and 8(b), all the nice properties are maintained except total population for female and male (TPFR, TPMR, TPFU and TPMU).
