Feature requests: Delete cases and spreadsheet display

Discuss the jamovi platform, possible improvements, etc.

by Wake » Sun May 05, 2019 6:46 pm

Hi,

I'm starting to work on step-by-step guides for jamovi, both in order to create the material itself but also to see whether I can convince the psychology team to make a wholesale switch from SPSS to jamovi in September.

The first guide I'm working on is about opening and cleaning Qualtrics data. I'm using version 9.6.9 and I'm pleased to see that the variable descriptions and value labels are being correctly imported from the legacy SPSS file downloaded from Qualtrics. However, for ethical reasons, data from participants who do not complete the questionnaire should not be used, Simply adding a filter is not sufficient, since it can easily be turned off at a later date. Such data would not, for example, be suitable for sharing on an open data platform, or indeed with anyone outside the immediate research team.

I would therefore like to be able to remove deselected cases, either with a switch in the filter options or as a save option.

In order to get around this problem, I did try exporting a template file and the data to a csv file. However, when I deleted the appropriate cases in Excel and imported the edited csv file into a new jamovi file created from the template, the value labels had gone.

There are also a few other aspects which make SPSS far easier to use than jamovi at this point, so here is a list of interconnected things I would like to be able to do:

1) Delete cases based on a filter (as above).

2) Sort cases based on the value of a variable. This could help with 1 by grouping all of the cases to be deleted together which would allow them to be deleted manually.

3) Toggle between value labels and the underlying numbers. It is good to be able to see the value labels, but information can be parsed more quickly as digits. This also helps with checking value label errors (which can easily occur in Qualtrics if the questionnaire is edited or copied after it is activated.

4) Preserve value labels in the template file, so that new data can be imported.

5) Paste new variable names from a spreadsheet into multiple data variables. Copying variable names and descriptions from jamovi to Excel would be nice too, but that information could come from a separate csv download of the data from Qualtrics.

Best wishes,

Wakefield

Wakefield Morys-Carter
ATSiP Treasurer and Membership Secretary
Senior Psychology Demonstrator and Teaching Fellow
Oxford Brookes University
Wake
 
Posts: 29
Joined: Tue Jun 26, 2018 8:31 am

by jonathon » Mon May 06, 2019 1:39 am

these are great suggestsions wake, we'll see what we can do.

re: 4) try using the 'retain unused levels' option.

when a template is created, a copy of the data set is created and all the rows are deleted from it. the normal behaviour of jamovi is to remove unused levels, so all the level value/labels get nuked. the intention with 'retain unused levels' is that you can instruct it to keep these levels. so if you turn on 'retain unused levels' the value/labels should persist in the template.

does that solve the issue?

i should make a note of this in the user manual.

cheers

jonathon
User avatar
jonathon
 
Posts: 929
Joined: Fri Jan 27, 2017 10:04 am

by Wake » Mon May 06, 2019 5:38 am

Hi Jonathon,

I did wonder whether "Retain unused levels" would help with the template export. Is there a way to set this flag globally, since otherwise I would need to change in separately on every single question (or question type) since I am specifically talking about imported SPSS/Qualtrics data files, rather than ones created natively.

Cheers,

Wake
Wake
 
Posts: 29
Joined: Tue Jun 26, 2018 8:31 am

by jonathon » Mon May 06, 2019 6:31 am

if you open the variable editor (F3), and select all the columns ctrl+A (or cmd+A), that will let you change *all* the columns at once.

jonathon
User avatar
jonathon
 
Posts: 929
Joined: Fri Jan 27, 2017 10:04 am

by Wake » Mon May 06, 2019 7:13 am

Thank you Jonathon. That did work and is an acceptable workaround pending delete or sort cases. I didn't test it because I thought jamovi would complain if I tried to edit different types of variable simultaneously.

Cheers,

Wake
Wake
 
Posts: 29
Joined: Tue Jun 26, 2018 8:31 am

by jonathon » Mon May 06, 2019 7:36 am

yeah, i might tweak the behaviour here.

i might make it so it retains all the levels in the template, then once the import is complete it can trim the unused levels after that.

cheers

jonathon
User avatar
jonathon
 
Posts: 929
Joined: Fri Jan 27, 2017 10:04 am

by Wake » Mon May 06, 2019 9:09 am

Thank you. That would certainly be more intuitive.

Alternatively or additionally, retain unused levels could be set as the default.

Cheers,

Wake
Wake
 
Posts: 29
Joined: Tue Jun 26, 2018 8:31 am

by jonathon » Mon May 06, 2019 10:46 am

yeah, we don't "retain unused" by default, because there's lots of little traps associated with it.

i.e. a person could make a mistake entering data, perhaps entering 'Yex' rather than 'Yes'; this would create an additional level. then they might correct it to 'Yes'. with "retain unused" the level 'Yex' would persist after the person had corrected it, and then analyses wouldn't work: "A level contains no data".

of course there are situations where a level having no data you *want* it to throw an error, and that's why we provide the option, but off by default.

so i think tweaking the export of templates is probably the best solution here.

cheers

jonathon
User avatar
jonathon
 
Posts: 929
Joined: Fri Jan 27, 2017 10:04 am

by Wake » Mon May 06, 2019 11:09 am

Thank you.. I'm not used to having errors created by unused levels per se, which is why I didn't consider the repercussions of that option.

Cheers,

Wake
Wake
 
Posts: 29
Joined: Tue Jun 26, 2018 8:31 am


Return to General