psychometric conversion scale - standard score to z-score

[Libby Smith]

I have a number of variables which are standard scores or scaled scores (language and congnitive test standardised scores) that I would like to convert to z-scores. I can do this manually by simply looking them up in a psychometric conversion table, however there are a lot of scores so I would like to do it in Stata. The command -egen(varname2) std(varname) - will not give the right answer, because it will create a new variable based on my data. I just want to transfer the standard scores onto the z-score scale, not actually create z-scores based on my dataset. Can I do this in Stata?

[Joro Kolev]

I do not know what you mean by "standard scores or scaled scores " . I take your question as "I have a bunch of variables, and I want to standardise them by subtracting the mean and dividing by the standard deviation, but I do not like that egen cannot replace".

You can check out the user written command -ereplace- "extends the popular egen and egenmore (if installed) modules to permit replacing.

[William Lisowski]

For those of us unfamiliar with psychometric techniques, your question really isn't clear without more detail, or at a minimum it is too difficult to guess at a good answer from what you have shared. I've searched "psychrometric conversion table" online, but what I found assumed I knew what I was doing and what my data looked like, neither of which is the case for me.

Please help us help you. Show example score data. Show us the conversion table you wish to apply to the scores - a link to one of the many online PDFs will do. Show us some examples of what you expect the result to be for several of your example scores. The Statalist FAQ provides advice on effectively posing your questions, posting data, and sharing Stata output.

[Bruce Weaver]

Hello Libby Smith, and welcome to Statalist. I would advise you to not overwrite existing variables with new z-score versions. The following excerpt is from Raynald Levesque's book on programming and data management for SPSS, but the advice applies to users of any stats package, IMO.

Do Not Overwrite Original Variables

It is often necessary to recode or modify original variables, and it is good practice to assign the modified values to new variables and keep the original variables unchanged. For one thing, this allows comparison of the initial and modified values to verify that the intended modifications were carried out correctly. The original values can subsequently be discarded if required.

Source: https://gsg.uottawa.ca/data/teaching...mt_20_book.pdf, p. 7

If you follow this advice, a series of -generate- commands should give you what you want. If the variable names are var1, var2, etc., you might use zvar1, zvar2, etc. as the names for the new variables. This can likely be done neatly in a -foreach- loop.

HTH.