I work with a large survey data that is pooled cross-section, i.e, people interviewed are randomly chosen in each time period. This is not panel data. But, I aggregate the individual cells according to age and gender groups and then run the regressions. Now that I work with age-gender groups, is this considered panel data?
$\begingroup$
$\endgroup$
3
-
$\begingroup$ indicate an example of data compilation based on age /gender separately . That should help to know your problem. $\endgroup$– Subhash C. DavarAug 9, 2019 at 8:23
-
$\begingroup$ it is basicalliy as the following: I have a labor force survey for several years. The same individuals may or may not participated in the survey in different years. I do not conduct the analysis on individual level, but on age-gender group level. An age gender group, for example, is males 30 to 34 years old. $\endgroup$– user23955Aug 9, 2019 at 8:50
-
$\begingroup$ do you have similar groupings for females ? And finally pooling these groups for regression ? what is your dependent variable ? O.K. - what is the goal of your study ? $\endgroup$– Subhash C. DavarAug 9, 2019 at 8:59
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Yes, you can consider this as panel data, but the key is to understanding why, as this affects how you explain and interpret your panel regression. In treating each cohort as the same unit over time, you are assuming that each cohort has a constant, unobserved fixed effect. That is, people in a particular age category and gender tend to have an unobserved effect on the dependent variable that is different to other cohorts.
Please note:
Age increases over time, so, for example, the group aged 30–34 in year 1995 should be treated as the same group as the 35–39 year-olds in 2000. I suggest your time variable increases in steps of five years, as you indicated that your age categories are five-year bins.
