Hello!
I'm working on a survey with two groups. The first are companies and the second are participants.
I want to implement a "matching" logic where companies fill out a survey first to state wich values are important for them. Second the participants fill out the same survey and in the end get a match-rating with the companies. E.g.: company A has 70% of the questions answered like you; company B has 63%.

If some german folks are around: It's meant to be like Wahl-O-Mat.

Is something like this possible?

Thank you!
P.S. Maybe I overread sth but I can't seem to find anything like that. Maybe it's really easy or a plugin exists, but I don't know how to google for my usecase.

Hi,
you did not answer but removed the questions at the beginning. Why?

What is "answered like you"?
What type of question and which scale do you want to use?
If you use a 5-point scale how do you determine "answered like you"? Exactly the same answer? Or a "4" instead a "3" is about 70% of the "same" answer?
Please, explain!

Workflow.
As all companies have to answer the survey before the participants start, you may store their answers in some hidden questions.
I assume that it is an open survey.
Then it should be possible to calculate the values.

Or it should be easy with an ajax call. But this depends on your environment.

Without further knowledge, no more.

Joffm



 

 Hi thank you for your reply!

Regarding the term "answered like you": Basically I have two question types.
1. Dropdown-Menu, where a matching answer should only be counted if exatcly the same answer was given.
2. 5 Point Scale, where partial points should be given if the answers are close to each other also.

But yes: Maybe I have to develop this further. I'm not quite sure what I want at the moment. 

Anyways I just wanted to ask the community if something like this "finding a close match to your answers" already exists. I assume it doesn't.

What I will maybe do now is to query the latest dataset from my survey after a participant completes it via the JSON-RPC API. Then I can work with Python to do the matching (whatever this will mean) and afterwards send an E-Mail to the participant.
Do you think this is a good solution? Or should a maybe have a look at another approach? 

Querying the "latest" dataset is risky. I would incorporate the response ID in the query. It is defined in the {SAVEDID} variable, available in the survey after the first page submission.

I think the "matching" should be possible from within Limesurvey, without an issue. If you have the answers from the company already, you can always create equations that check if a respondent and a company have similar answers.

Here a rough example
Answers of company 1:
 
Answers of company 2:
 
My answers:
 
Result
 

Only four simple equations and a temporary container to store intermediate results.

Algorithm:
Alternativ questions: exact same answer: 100%, otherwise 0%
Rating questions: exact same answer: 100% difference of 1: 50%, otherwise 0%

Now you can improve the result by
refining the algorithm
creating a table to show the answers to each single question
or use pdfReport to create a good looking document (and send it to the respondent),
or...

Joffm