Working across platforms with SPSS, Excel - best practices

[Will Hall]

Hi everyone, one of my colleagues works with Stata and the other has SPSS.

Will SPSS users be able to work with my data readily?

I have tried to export my Stata dataset to .xls or .xlsx excel format using the Export... command but get corrupted data.

I don't have spss to try to open the .dta file.

Any help on best practices here so our team can work with the data on different platforms?

Thanks!

- Will

[Nick Cox]

I work with many people who also use Excel spreadsheets and have often used export excel on their behalf. It's worked well for me and them. To comment on "corrupted" we would need an example of Stata data, your exact Stata code and details on what you mean by corrupted. The command even works well with date variables, a notoriously awkward area.

[Will Hall]

Thanks a lot for your reply. This is reassuring.
I exported my .dta dataset as an .xlsx file as I found suggestion online, this time it seems to work -

So you are suggesting that's the practice, export as an excel file and then my colleagues with SPSS import the excel file is that it?

- Will

[Clyde Schechter]

In addition, you might consider using Sergiy Radyakin's -savespss- program, available from SSC. This will enable you to save your data sets as SPSS files. The advantage to your colleague is that this approach preserves things like variable and value labeling, that are lost when exporting to Excel and require programming to re-create. I have, at this point, only recently begun using this program, and so far it has given me good results, except with some datetime variables. Going in the other direction, if you are running version 16, Stata can directly read SPSS files using the -import spss- command.

Another possibility to consider is StatTransfer, by Circle Systems. I used this software happily for many, many years. It worked very well, was easy to integrate into Stata workflow thanks to Roger Newson's -stcmd-, -inputst-, and -outputst- commands (which come bundled with StatTransfer and are also available on SSC). When I ran it on a standalone computer under an individual license it was great value for the money. Unfortunately, I have been involuntarily switched over to working with these particular colleagues over a network, and the cost for a network license is beyond my budget (and nobody else on the network has any need for it, so they won't chip in). But if you are not installing it on a network, the standalone computer price is still quite reasonable.

[Joseph Coveney]

I would consider Stat/Transfer to be central in any consideration of best practices for working between colleagues some of whom have only one statistical analysis software package and some only another. You can try to use Excel workbooks as the go-between, but keep in mind that such an approach makes sharing metadata, such as variable labels, less convenient.

Another approach is to set up and maintain a relational database on a central server (or cloud, whatever) and keep the data and metadata in it. There are concrete advantages to that, but it's much more administratively involved.