Problems with roster

Hello @sergiy I am downloading the information from the server, however I have two rosters where I only get “0”, i.e., I can get the real answers … please see the screenshot. How can I solve this?.. Thanks


If by “real answers” you mean the words “Maiz” and “Cacao” then the *__id variables will indicate what object the data row is about. It is basically the @rowcode value that you use in the syntax when writing conditions. Refer to the trigger question to recover the entered names if needed. (I can’t figure out what type of q4 really is: multiselect or list).

Best, Sergiy

Thanks. yes, real is maiz and cacao (sorry about that). If I download data from another roster, I got the id linked to the products, in this case, it is not. I am not sure the reason why. Please see on the screen question´s details.

What is the program that is on your 2nd screenshot?

I used R, I had to create a script to merge different versions,

The issue is not related to rosters whatsoever then. You should be having the same problem looking at any categorical question. The “real answers” you are seeking are the value labels corresponding to the codes you introduced in the Designer. Since you are reading only the “*.tab” data files the labels are not used. Tab-delimited data files contain only data, not metadata.

Import a Stata or SPSS file instead. If R has a capability to work with value labels, then it should recognize their presence in the original file and read them in.

An R-expert can possibly explain this better.

1 Like

Hello @sergiy , thanks. I am not sure if I understood, but even If I open the .tab file in other editor, “Visual Studio”, I got the same, please see in the screenshot the id = 55-41-04-99.

Hi @Jaime,

As @sergiy pointed out, i seems that your main issue is that you are reading a .tab file, which does not contain any label information. No matter what software you use to open the file, it will only contain the variable names (columns) and value codes (0,1,2 etc.).

As Sergiy also mentioned, you can instead try using another file format, such as SPSS. When you export Survey Solutions data, select SPSS as the output format. Then, if you are using R, use the haven package to read the contents of the file.

You will use the read_sav function from the haven package (instead of read.delim) to read the SPSS file. You should now be able to see both values and labels. (In this case, it is the variable label that will be useful to you. When you use the View() function to see your data, the columns should show you the variable labels underneath the variable names)

1 Like

Correct @giansib! And thank you for bringing in the particular package name in R that can achieve this.

Hello @giansib and @sergiy thanks, I will try. However, I don´t understand why I opened .tab files in R, and only I have “issues” with this roster. If you suggest to download SPSS format, then, the process I have done until now is wrong??.. Thanks

Perhaps if you can provide some examples/screenshots of the .tab files that do “work” for you I can see if I can provide some more insight @Jaime

@Jaime, if your problem is really that you want the names instead of the codes, using SPSS files with the haven library as suggested by @giansib is a way to do it.
Alternatively you have the maps of option codes - option labels in the .do files and can read those like the .tab files.

But in your original post you say:

I only get “0”, i.e., I can get the real answers

If you are worried about the "0"s, these are the options in your multi-select question mainProdCatAgri which were not selected by the interviewer. Where he selected options you see their codes (like 4, 25, 14, etc.).

@klaus @giansib thanks, sorry I did the wrong question. Now it is clear. My issue here, is because I can see the ids in mainProdCatAgri roster when I use .tab format, I only have “0”. My next step is try with the labels, to see if I got the labels of the answers. Thanks

Hello @giansib and @sergiy I used read_sav, cat<-read_sav("xxx/xxx/rosterCatAgri.sav").

The result was the same, please see screenshot. Did I do something wrong?

@Jaime, I think you found a bug.

It seems that there is an issue with the data export indeed! When you export your data, instead of seeing all 0s in your mainProdCatAgri__ columns, you should be seeing the option code of the selected categories.

The very last section of the Multi-Select question documentation explains what is the expected output for ranked multi-select questions.

I have just confirmed that this is not properly happening. I created my own questionnaire with a multi-select, combo-box, ranked, reusable category question:

And I entered some data:

But the exported file has all 0s for all variable columns:

@sergiy, I will attempt to post a support ticket with reproducible example by tomorrow, as it appears to be a real bug. Let me know if you have any other insights at the moment.

thanks @giansib I hope You should fine the way to solve it, on the contrary I will have to write down manually the answers, to do my report.

See:

which has been fixed already.

@Jaime, please update and re-export.

1 Like

Thanks @sergiy!

I had not noticed

@sergiy thanks, I need to work on the transcripts, do you have more examples of how can I download them?.

Hi, I don´t understand how to download the transcripts. Any way to get a real support???