sort command does not work perfectly

[Mahdie Raj]

I have a column in my dataset containing age groups such as 50-51, 52-53, ..., 100-101. When I sort this column, the 100-101 value shows up at the first line followed by 50-51, 52-53, etc. Obviously, it should show up after 98-99. I appreciate your thoughts.
Best,
Nader

[Joseph Coveney]

Your column is a string variable. You'll need to encode it to get the categories (or prefix the strings under 100 with a zero) to sort as you want. At the command line in Stata, type

help encode

and

help label

for more information about how to create the set of value labels and encode the string variable using the set.

[Clyde Schechter]

In Stata, what you see is not necessarily what you get (nor what you have). Items like 50-51 are not numbers. Although your brain perceives them as such. To Stata "50-51" is a string. The variable that contains these is a string variable. So when you sort it, it sorts in dictionary order, just as all strings do. In dictionary order 100-101 comes before 50-51.

If you want the data to sort more appropriately for the meaning of these values, you need to create a value-labeled numeric variable where the actual values of the variable are numbers like 1, 2, 3,... but these numbers are associated with labels in the appropriate order, "50-51", "52-53",...with "100-101" far down the list. The -encode- command is often a convenient way to create such variables. Do read -help encode-. But -encode- will, by default, put things in alphabetical order, so in order to get this to work, you actually need to first write (or have Stata build) a label where 50-51 is associated with 1, and so on. You can then specify that label as an option to -encode-.

Added: crossed with #2 which says the same things more succinctly.