Presented by Professor Amaresh Dubey and Professor Wendy Olsen.

In the morning session, Professor Dubey introduces the Indian Human Development Survey. As a large longitudinal field survey data at two time points, 2004/05 and 2011/12, it represents all regions of a huge country. He shows how to use design weights and calculate the design coefficient DEFF. He also introduces several STATA commands, notably svyset, reshape, generate, replace, label define, merge, keep, drop and rename. Participants are given the do files to run, making their experience easy.

Participants will be exposed to merging two rounds into a panel. Professor Dubey then presents the regression modeling of an outcome in Round 2 based on structural variables measured in Round 1. The response variable is likely to be a change-factor (transitions in/out of poverty versus staying in poverty). It is likely that logit regression will be used.

The afternoon session is shared with Professor Wendy Olsen, who will introduce measures of human capabilities using microdata. The afternoon will cover regression on an index of well-being or capabilities using two competing methods---a classical scale and confirmatory factor analysis -- in Stata. One can use SEM or GSEM commands to combine a regression with a CFA, as taught elsewhere. The more advanced commands used in the afternoon include recode, replace, regress, sem, estat eqgof, and predict. Again, participants are given the do files to run.

This workshop is of intermediate level. Some past experience with SPSS or STATA is recommended. Participants are not able to take away the whole IHDS dataset with them. Indeed participants use a masked (anonymized) dataset version in the workshop.

Anyone can register online for access to the actual IHDS data for their research. See data access URL http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/22626?searchSource=find-analyze-home&sortBy=&q=dubey

Indian Human Development Survey: Panel data analysis using STATA