You can't jump back to a skipped question or a different question branch

@salty - you would be better suited to move q5 to be after q15… you are going backwards in the current logic and that doesn’t work. as you have discovered. 

des

@john.desborough Sorry if this is a really stupid question, but how do I do that?

7->15 is handing an edge case. My main flow is 1, 2,3, 4,5, s2, 16. 

My understanding is that moving 4 after 15 would mean that everyone would have the do the edge-case screening even if they don’t meet the criteria (assessed in Q3)
 

Hi @salty I hope you’re doing well! Did Des’s solution work for you? Let us know!

Nope, couldn’t execute on it. I can’t even see a loop backwards. I see 

1 → 2 → 3 → 4 

or 

1 → 2 → 3 → handle edge case → 4

In both branches, things seem to move only forwards

Hm, as Des mentioned, it’s not possible to go backwards, but what issues did you run into when trying out his solution? Happy to take a look at the logic jumps you have setup!

to confirm, by going backwards,do you mean connecting a higher number to a lower number? Do you mean backwards, as in further to the left on the sheet? Neither of the those would make sense to me. I can only understand it making sense if you can’t go back in the series of questions such that someone could be asked the same question twice, and I can’t see that here. 

What am I missing?

Here’s what I’m trying to accomplish. As far as I can see, no matter which path is taken, it’s always moving forwards, never revisiting any node twice

Hi @salty Yes, that’s correct. In this case, the respondent is moving backwards in the form when going from 15 to 4. You will need to adjust this so the respondent continues to move forward in the form. 

I’ve had 3 team members look at this, no-one can figure out how to make it work. It seems as though one cannot handle multiple long edge cases one after the other without breaking the seemingly arbitrary numbering system.

It doesn’t seem to make any sense to assign questions as a series of numbers where there are branches

Envision a path with two branches

1 → (A1 → A2 → A3) OR (B1→ B2 → B3) 

Which should one give higher numbers: The A-series or the B-series? How should one make that decision? If it’s arbitrary, then what sense does it make for the numbering system to decide what can branch to where? Doesn’t this just introduce complexity for no reason?

I don’t want to switch from typeform, but I can’t see any alternative when I can’t do basic things, like use branching logic, without it becoming a logic puzzle

Is this what Question groups are for?

@salty - try reordering your questions/numbers along the lines in the image below… 

this way your logic path for the edge case should work easily AND you are not going ‘backwards’ 

here’s an example of a fairly complicated supplement selection quiz that has about 4k logic rules buried inside…. there’s a red line along a ‘short logic path’ for those who were pregnant (at the bottom) but there were a lot ‘edge cases’ as you called them in the logic. I had ONE spot where i sent things backwards by one question and it took me about 18 hours to debug. … make it big to see the various paths.. 

duplicate your quiz and try the logic/question reordering to see if it works out.. 

des

I ended up using question groups. It solves the arbitrary numbering issue while giving you higher level abstractions so debugging is less painful. The drawback is that it creates an extra click for every ‘edge case’. The problem with trying to mess around with question ordering is that every time you move numbers around it changes the other numbers e.g. if move the sixth question to position 2, 2 becomes 3, 4 becomes 5. That’s fine if you need to move one or two things, but it becomes nightmarish if you need to do something more extensive. One slip up, and everything goes to poop.

Apologies for the wrong use of terminology, I see now that ‘edge case’ was a poor way to put it. I meant “special case which requires questions only relevant to it”.

@salty Glad to hear question groups worked for you! I know it’s not the easiest when you have a one-off case that messes up the entire logic jump setup, but hopefully the question groups help in the meantime.