Creating a categorical variable based on observations

[Ajay Joshi]

I want to create a new categorical variable that will have 5 categories based on groupings of observations. For example, fees = 5 will include observations 11769/12911 & 18862/19769 and a similar approach for the other categories. I'm not sure on how to create all 5 categories?

[Carlo Lazzaro]

Ajay:
without further details, I would go step-by-step following a safe but inefficient way:

g wanted=5 in 11769/12911
replace wanted=5 in 18862/19769

...and the -replace- festival will go on and on until you reach the lower bound of your ranking.

[Nick Cox]

Carlo Lazzaro is bang on in terms of the question asked. I would add that

1. Specifying such a new variable depends sensitively on the current sort order.

2. It's essential to document why these observations are used and indeed almost always better practice to have a rule for generate and replace that is more explicit.