For the csv outputs, I noticed that each column does not come exactly as the order of the questions; for example, Q30 might appear in a earlier column than Q25 despite we programed Q25 to appear on an earlier interview stage. I was wondering why the variables did not go as the order of interview surface, and if there is anyway to fix it prior to data exportation?
CSV is actually not a standardised data format, meaning that there are no ârulesâ about how it should be structured. Our exporter code simply encodes the attributes as it finds them. This may or may not correspond to the order they are collected. More than likely it will not, since the language we have written the software in does not guarantee object property order, and besides variables are âkeyedâ in the node object by an internal unique identifier rather than the variable name you assign (in order to avoid collisions). What this means ultimately is that there is no concept of variable âorderâ in Network Canvas.
With that said, the column order really shouldnât matter. If you parse the CSV into a dataframe using any data analysis package, you will almost certainly be able to access a given variable by the variable name itself - the ordering should be inconsequential.
Is there a reason you feel you need this behaviour?
Thanks for the explanation. I will look more into using the data analysis package to reorder the variables. The reason I wanted the variables to be listed in order is for a better presentation on CSV for my colleagues, so itâd be easier for them to follow alone the network canvas interviewer protocol.
Have you considered naming your variables using a scheme that has the order they are collected included (for example, 1_name, 2_age and so on), and then just sorting the columns in excel?