Geographical variation in mobile phone ownership and SMS literacy among women (age 15–49) in India: A cross-sectional analysis based on National Family Health Survey-4

Abstract

In India, studies on mobile phone ownership and digital literacy have focused on the gender dimension, perspectives of women and men towards mobile phone use, and socio-economic determinants of ownership. There are no studies with a large sample size that present the variation in mobile phone ownership and ability to read short message services (SMS) message among women across different geographic levels. We use multilevel models to estimate the contribution of geographic levels in explaining the variation in mobile phone ownership and SMS literacy. The data from the fourth round of National Family Health Survey (NFHS) which collected information on health and nutrition indicators was used for our analysis. 122,351 women were interviewed about mobile phone ownership and SMS literacy. Information on mobile phone ownership and ability to read text messages were used to create the dependent variables. The independent variables included women's education, age, religion, social class, wealth quintile and place of residence. Two and four level variants of a multilevel model, with individuals (level-1) nested within primary sampling unit (level 2), districts (level 3) and States (level 4), were used to estimate the probability of mobile phone ownership and ability to read text message. The results from multilevel model and variance partition analysis indicate that contribution of State and primary sampling unit towards explaining variation in mobile phone ownership and ability to read text message is greater than the districts. The variance estimates are sensitive to the number of levels included in the multilevel model, so relying on estimates obtained for a particular level could lead to a bias. Our findings suggest that understanding the magnitude of inequalities at different geographic levels should be warranted more attention before tackling the socio-economic factors. State estimates should be supplemented with information available for lower geographic levels. It is important to identify the geographical clusters with high and low coverage of mobile ownership and ability to read text messages before implementing mhealth interventions.