I am setting up a number of performance evaluation surveys for my staff in a small medical office of 20 employees.

Not every question is relevant to every staff member. For example, the billing department may interact with a medical assistant in one way, but a clinician in the practice may interact in a different way, and thus need different questions. I'd like to divide up the questions based on participant attributes like "general staff", "clinician", "manager", "billing staff", etc.

I've read in the forums here that there are two ways to do this:

1) Token attributes.
2) Panel integration.

I have not yet read about these, but before I do, I wanted to check if there were any other ways to accomplish my goal. It would be great if I could tag individual questions to different attributes, but I don't see how that is possible.

Are there any other ways to accomplish my goal?

If you don't want tokens, you can ask the respondent what their role is and then filter following questions/groups by the answer to that question.

Interesting, thanks for your response!

So tokens would be the way to do this if I wanted to pre-define what questions the participants see?

Yes, or prefill a hidden (via CSS) question with the URL.

- www.limesurvey.org/manual/URL_fields/en#...using_GET_parameters

I would go with tokens. You can add custom attributes to the token table and you can access these attributes with Expression Manager in relevance equations to show/hide questions that are relevant or irrelevant for the respective person or role.

So you should definitely look into tokens and Expression manager.

I think panel integration makes only sense if you can't / don't want to use tokens and want to pass a variable through the survey URL into Limesurvey.

One other question that may be related:

I had planned to set up separate question groups for the different groups of staff, thinking that would be important. The manual talks all about what question groups are, but doesn't explain WHY you would want to use them. That's my biggest beef with this manual in general. It explains the technology but not how it benefits you.

Do I need separate question groups if I have other means of separating questions based on participant attributes? Why might I want to use separate question groups?

Well, groups are groups of questions. Pretty simple. Why would you use groups? For structuring your survey. For example if you have blocks of questions that belong logically together (you can randomize these groups, you can hide specific groups, etc.). Then there is another reason: There are 3 modes on how to display the questions in your survey: all in one, group by group, question by question.

I think it doesn't make sense to list benefits of groups, because they are pretty obvious and they also might be different for each an everyone. It is only you that can decide how you want to structure the survey.

There is one problem that I see with groups in Limesurvey: Besides the logical grouping it is also used to create a kind of "page" concept. And those two concepts might go against each other.

Ideally you plan your questionnaire, the questions you want to ask and who you want to ask them.

First of all, what you have to do is think about your questionnaire. Ideally create it offline in word first. Then think which question should be ask to who.

If you don't want to show question by question, then you most probably need to use the groups to create "pages", rather than using them for logical grouping.

At the end of the day, the question why you want to use groups is not easily answered and it always depends on what you want to do.

I assume you are not doing this for a living, but rather for the first time? Then the best thing is to think about your questionnaire first, the questions you need to ask, etc. Then you can start thinking about how you will implement it online.

Because only when you know what you want/need you can really plan for the final structure and logic.

Thank you for your response, this is helpful.

So I should think about question groups in terms of participants seeing them on different "pages". That's helpful. I'm not so interested in having separate pages. Ideally I'd like all the questions to appear on one page, but I'd like to have different questions visible to participants with different attributes.

This is a performance evaluation for staff in my office. It's more or less written already, but I'm trying to adapt it to this electronic format and make it more robust in the process.

Here's my wish: Have a single list of questions, like a "bank", and send a link out to participants such that they will only see the questions that are relevant to them (sorted based on their attributes). I'd like to be able to sort the data based on attributes as well so I can see, for example, how my billing staff rated a particular employee on their "communication skills". If I could export this into Excel for tables, graphs, etc, even better.

Thank you, holch.

I've taken a look at tokens, as well as the EM. Wow... this is all very confusing. I wish the manual were better at walking you through the basics. It seems to jump right to advanced applications of the concepts while skipping over basic syntax, where to enter stuff, etc.

It seems like the simplest approach I've found would be to start with a "role" question, and then use the answer to hide/unhide subsequent groups of questions designed for different employee classes. That way I wouldn't have to enter relevance equations for each and every question.

I can't seem to get the syntax right though, maybe you could help me.

The "role" question is in a group by its self, and is coded "r01". The answers are "A1" through "A6" depending on the role. The answer for 'doctor' is "A1". So if I want the 'doctor' question group to be shown, it would seem I need to enter something like "r01==A1" into the relevance equation box for the 'doctor' question group. The problem is I can't get this to work, and I can't figure out exactly what the syntax is supposed to be. For example, is it:

"r01==A1"
"'r01'=='A1'"
"r01"=="A1"
etc... I have tried several combos, and I can't get Limesurvey to hide the question group if I enter a different response to the preview survey role question, such as 'nurse practitioner' (A2).

You don't need quotes around the variable name (the question code), but you do need them around the value (the answer code):

I've made quite a bit of progress with this info. I now have a survey that:

1) Asks the staff their role.
2) Asks the category to which the person being reviewed belongs.
3) Presents options for who the reviewee is to be based on the response to #2, and hides the rest.
4) Presents review question groups based on #1 and #2.

This is why I was curious about why you would want to group questions. Yes- it puts the questions on separate pages-- but it also allows you to neatly show/hide questions based on previous answers-- but only if those previous answers were on a preceding page (ie, in a different group!).

but only if those previous answers were on a preceding page (ie, in a different group!).

That is not really true. You can have questions on the same page, that are hidden, until you click a certain response for a question. This question can be on the same page. But of course you can't hide a group based on a answer within that same group.

I figured there was probably a way to do that, just more complicated than my understanding.

I posted on another issue, but I think I figured it out. I was having an issue with this statement: r01 == "A1" AND r02 == "A3" OR "A4". It wasn't giving an error, but it also wasn't hiding the desired group.

I discovered I need brackets around the AND statement to make the OR portion work independently of the part prior to the AND.

Example: r01 == "A1" AND (r02 == "A3" OR "A4")

Nope, that didn't work. I'm having all kinds of problems with this syntax.

Here is an example:

Group relevance expression: r01 == "A1" AND (r02 == "A3" OR "A5")

If I select r01 = A1, and r02 = A2, this group still shows.

If I place the brackets like this:

r01 == "A1" AND r02 == ("A3" OR "A5")

LS returns an empty page (no questions, but with a "next" box), but since the fields are mandatory, it says you must still complete them. Weird.

If I eliminate the "OR" statement all together, it works. So that seems to be where the problem is.

UPDATE... trial and error seems to have worked.

r01 == "A1" AND r02 == "A3 OR A5"

Putting the quotes around the grouped answers did the trick. For now!

UPDATE... not so fast. That didn't work. But this did:

r01 == "A1" AND (r02 == "A3" OR r02 == "A5")

I'm including all these updates to make my process explicit for anyone else who might be going through the same educational process I am. Would be nice if the manual were helpful here! I've learned a little bit by reading up on Boolean expressions.