Q Research Software

From a user: When I say it would also be amazing if Q automatically coded based on mentions within the verbatim another way of saying it would be to be able to code based on "contains" the word or phrase in the verbatim. I think this is a particular issue when using a code frame to code overlapping (multi) codes. E.g. at the moment if I code "Innovation" as innovative the next month it will automatically code "Innovation as "Innovative" but if someone says "Innovation and quality" I have to code it as "Innovative" and "Quality". Then the next month someone might say "Innovative & Quality" and because it is slightly different it means I have to code "Innovative" and "Quality" again. When people are asked to give words they associate with a brand and there is only one text box, the combinations are literally endless!

Allow users to sort code of codeframe alphabetically when using manual/semi-automatic coding interface

User describes this well. Essentially she wants to input a 2-column spreadsheet with text in one column and code in a second column to get Qcodes file. I have historic data in the form of column A (text input) and column B (assigned code) which I want to turn into Qcodes files and keep the codeframe numbering consistent. I imagine this scenario is pretty common when clients are switching from manual coding or different systems. In one of the other automatic coding processes in Q it allows me to enter labels and variations when importing categories, but I think there is an upper limit on the number of new codes and variations it will accept. I'm dealing with codeframes of 1000+ codes so the limit was too small for me. I have been using python to transform this historic data and the pre-existing codeframes and reverse-engineer them into Qcodes files. It would be good to have a way to do this within the software.

When you categorize a text variable as a Pick Any, it will create a variable for each category. The variable names are based on the original "variable name + category label". It would be better for other automations (and tidier) to instead create variable names for these new variables from "original variable name + category value".

From user: I am using the manual coding functionality and the challenge I have is that if I use the find field it doesn’t return back an exact match. E.g. if I ask it to find ‘reliable’ it also gives me ‘unreliable’ which means I then have to code reliable then take unreliable out and code as unreliable. Another example would be ‘class’ and ‘classic’.

It’s a lot easier to code using legacy coding. Even a beginner will know what to do but using R coding is confusing. I have used Q for a lot of years and still found it confusing to code using R. I guess ease of use should be given a priority

The user wishes to be able to use the semi-automatic algorithm for languages like Japanese. The only way to currently do this is to translate the text variable first.

The request to allow auto-coded outputs to work with general coding QCodes files. For example, being able to export a QCodes file from automatic categorization would make it easier to continue manual coding.

User sought to use a pick any variable in Automatic Categorization - Unstructured text > Existing categorization. It is currently set up to only take ONE variable, however, I could see the feature being expanded to take a pick-any and return an overlapping codeframe. The workaround for this user was to run the Unstructured Text feature for each binary variable they were coding into, then set the output variables together in a pick any. Except, in this case, the user had over 100 categories, so it would have taken a long time. !image-20230214-200357.png|width=432,height=530!

This question comes up fairly often. People see the n statistic in the categorization tool: and think that that is supposed to be n count in the table created by the results. It's actually the number of _responses_ rather than _respondents_ (which matters when you're coding a text-multi question). Can we better label the metrics inside the () so they are more intuitive? i.e. verbatims=2 responses=324 or something or add some hover help text or something?