Alternative to rowjoinbyname

[Henry Strawforrd]

Is rowjoinbyname a more recent command? In the desktop Stata version 16 the code below works, in the virtual machine I have Stata 15.0 and it gives me the error "matrix operation not found".

How can I accomplish the same without using it?

MWE below:

sysuse auto.dta

reg price weight mpg
mat output = r(table)
mat list output

matrix estimates = output["b",.]
matrix standarderrors = output["se",.]
matrix lowerbound = estimates-1.96*standarderrors
matrix upperbound = estimates+1.96*standarderrors

matrix rowjoinbyname results = estimates lowerbound upperbound

[Fei Wang]

Equivalently, you may use

matrix results = estimates \ lowerbound \ upperbound

[daniel klein]

Yes, the rowjoinbyname subcommand was introduced in Stata 16. I know that because

help whatsnew15to16

contains the entry

1.3.15 What's new in programming
[...]
4. New commands matrix rowjoinbyname and matrix coljoinbyname join matrix rows (columns) matched on column (row) names.

In your specific example, you can just

matrix results = estimates\ lowerbound\ upperbound

and obtain the same result. However, the confidence intervals (not relying on the crude approximation of 1.96) are already present in r(table), so I fail to see why you would even want that.

[Henry Strawforrd]

thank you!