如何在Stata中完成多维度数据重塑？解决reshape报错与格式转换问题

Hey there! You’re spot-on about why the initial reshape command threw an error—let’s walk through exactly how to transform your data into the desired format, step by step.

Why the Error Happened #

Your guess is correct: the i(country) identifier isn’t enough to uniquely label each row in your wide dataset. Each country has 3 separate rows for measures A, B, and C, so you need both country and measure as your unique identifiers for the first reshape step.

Step 1: Reshape Long (Year Columns to Rows) #

First, we’ll convert the year columns (yr1995, yr1996, etc.) into a single year variable. Use this command:

reshape long yr, i(country measure) j(year)

After running this, your data will look like this (abbreviated):

country measure year yr
USA A 1995 5
USA A 1996 4
USA A 1997 1
USA B 1995 1
UK A 1995 2
...

Step 2: Reshape Wide (Measures to Columns) #

Next, we’ll pivot the measure values (A/B/C) into separate columns, using country and year as the unique identifiers this time:

reshape wide yr, i(country year) j(measure) string

The string option tells Stata that measure is a text-based variable (since A/B/C are string labels). Finally, clean up the column names to remove the yr prefix:

rename yr* *

Combined One-Liner (Optional) #

If you want to run both steps in one go, chain the commands with pipes:

reshape long yr, i(country measure) j(year) | reshape wide yr, i(country year) j(measure) string | rename yr* *

Final Result #

Your dataset will now match exactly the format you wanted:

country year A B C
USA 1995 5 1 0
USA 1996 4 2 4
USA 1997 1 1 2
UK 1995 2 2 2
UK 1996 4 8 4
UK 1997 9 4 1

内容的提问来源于stack exchange，提问作者Jay