(Stata13) Panel Data Descriptive Analysis (Histograms) : CrunchEconometrix

In this video, using Stata13, I show the various ways of using the histogram to reveal various heterogeneities within and among the groups with the summary statistics of the outcome variable (Gini coefficient) and the different categorical variables such as countries, regions, income groups and inequality categories.

Besides, why do you want to perform panel data analysis? Some of the reasons could be to explore the behaviour of a variable across a sample of groups (e.g. firms, schools, countries, agro produce, individuals, cars, production style, political parties etc.); to explore the behaviour of the groups in the sample with respect to a variable; the groups have some semblance of commonness; the groups have varying heterogeneities (i.e. differences) with respect to a variable; perhaps due to inadequate data over a long period for your sample or maybe you just don’t know why! (lol).

I give 5 tips required for building an engaging panel data structure. They are (1) generate identifiers (watch my video on “Reshape Wide to Longitudinal Data” and “Tips to Building a Panel Data”); (2) reshape the data (watch my video on “Reshape Wide to Longitudinal Data”); (3) group classification (to explore the heterogeneities in the data); (4) categorise the outcome variable (to explore the heterogeneities in the data) and (5) create dummy variables (regional and time dummies).

Also note that a histogram (1) is not a bar chart, (2) shows the underlying distribution of a variable, (3) gives an accurate representation of the distribution of a numerical data, (4) gives the estimate of the probability distribution of a variable, (5) graphically displays data using bins (bars) of different heights and (6) shows the frequency distribution of a variable. Remember, graphics convey lots of information than words. So, load your data and let’s get started.

[Watch video tutorial]


Leave a Reply

Your email address will not be published. Required fields are marked *