Related forum topic: http://www.limesurvey.org/en/community-services/forums/can-i-do-this-with-limesurvey/80879-reset-answers-in-a-table-question,-using-a-button-and-javascript

Can people look at this and give some feedback if you want this or not?

Have the ability to insert reset button for question without code everytime with java it's a really nice feature!

For single choice : add 'No answer' item and your go.

this difference between NULL and no answer: it made so many problems with sliders...

I know sociologist consider it as a scientific point, but I'm pretty sure that epistemologist would just find here another point to question the scientificity of sociology :-p (Would Karl Popper have supported a reset button? ^^ )

I mean: I can understand that in a cognitive point of view, not clicking on any answer is not the same than clicking on "no answer". But, clicking on an answer and then clicking on reset, IMHO, it's just not the same at all than not clicking: NULL means that participant clicked on nothing, not that he clicked on reset. So in a scientific point of view, filling the field with "NULL" when people click on "reset" is wrong. Indeed, in a scientific point of view, it would be needed to have 3 values: NULL, No Answer, Reset. Then, it would be pretty easy to prove that "Reset" is logically equivalent to "No Answer".

Whatever, debugging sociology empirical paradigm is not my job. We already did such a reset button for sliders. We can do something similar for other questions. Indeed, it can be done in JavaScript with a workaround.

Thinking about it again, to repeat myself but using the "LimeSurvey language": I'd say that (a) a non-mandatory question with a "NA" button is scientifically better than (b) a non-mandatory question with no "NA" answer but with a reset button.

In case (a), if a participant doesn't touch anything, null is recorded. If the participant wants to cancel his choice "NA" is recorded, else, the selected value is recorded.

In case (b), if a participant doesn't touch anything, null is recorded. If he resets his choice, null is recorded, else, the selected value is recorded.

So, in (b), less information are recorded, and two actions are fused. This fusion between those two cases "probably" has a huge impact on the anchoring of participants (https://en.wikipedia.org/wiki/Anchoring): I bet my two cents that when a reset button is available rather than a "No Answer" item, people are more likely to choose an answer, even when it's not really relevant to them. So the final set of data will be less relevant in general: like counting the vote intention of people who indeed will just not vote at all the day of the election. Maybe if more "No Answer" option would have been used in the American Election surveys, the polls would have been more relevant.

Anchoring is a cognitive bias, putting a cognitive bias in the middle of an experiment is creating an experimental bias. The "RESET" button create a dependent variable that is not controlled.

Just the epistemological point of view of PHP coder on the scientificity of a reset button.

@LouisGac : for (a) : you mean : non-mandatory + hide no-answer ? Right ?

If yes :
case (a) : null is recorded => '' is recorded. Null is reserved for 'not seen'/'hidden by condition' (if able in the DB).

Else : i'm totally OK.