Display blank responses as blank instead of as <variable>

What I need to do is choose two people at random from six that participants enter at beginning of the experiment. They fill out further questionnaires about just these two, and not all six.

The way I'm solving that problem (unless you have a better suggestion), is by having six fill-in-the-blank variables that are presented in random order in the first questionnaire. From the participant's perspective, they are all exactly the same. That way I can have MediaLab always inserting fill-in-the-blank information from the same variable in later questionnaires, but I avoid order effects by having the people entered in a random order.

Here's my question. I want to make sure participants enter a new person in each of the six blanks. I'd like to display what they've already entered, but, since the variables are being presented randomly, there are always blank variables. MediaLab presents those as <variablename>, but I'd just like them to be blank. Is there any way to do that?

I found a way to do this, but it's not the best. What seems to work is to have the variables presented at the beginning, have the experimenter enter a comma for each one, and then go to the instructions, etc., and have participants enter new values for the variables. The commas present as blanks, so participants only see what they have entered.

One problem is that since you're presenting variables twice, MediaLab pops up a warning every time you run it. Also, it relies on the experimenter to do everything right. It leaves you with two sets of variables in your data file, one that's all missing data, and the other one is what you care about. So, it works, but I'd still love to hear if there is a more elegant solution!

Would you prefer to have all six blanks on one screen at the beginning?

It's fine if all six are on the screen at the same time, but they need to be presented in random order as far as actually filling in the variables is concerned. We're trying to avoid order effects in the order that participants think of or enter the people they're going to answer questions about.

I've attached a revision of a previous custom item that was created to provide multiple fill in the blanks on one page. In this revision, the blanks are presented in a random order. The javascript file (random_order.js) included in the folder is generic and can accompany any HTML file. So it could be used to randomly order pretty much anything in custom items or other HTML files. Credit for the javascript code goes to [URL="http://www.javascriptkit.com"]www.javascriptkit.com[/URL].

This looks like it might be workable. I guess I'll have to learn some Java!

Here is a follow up sample that randomizes the order of scale response options. It doesn't have an impact on the response data--just the order in which the options are presented. Note that the custom item requires no special parameters such as the (v6) used in the example above. That's because only one response value is collected. Note also that this sample includes an accompanying .que file.

[I]Note this sample was revised on November 18th, 2008 to include "values" for each of the radio options (see discussion below).[/I]

Thanks for working on a solution for me. I especially appreciate the sample questionnaire file. I just have a few follow-up questions:

(1) How is data recorded for this custom item? The data files produced by this configuration don't record numerical data (the best I can tell). For the two sample questions, only the word "on" is recorded in the dataset. Am I missing something here?

(2) Am I able to reformat the look of the random_scale_options.htm file (font, background color, etc.) without it effecting how the random_order.js file works? I have limited knowledge of html and zero knowledge of java, so I have no idea how much I can fiddle with the html file without screwing things up.

Thanks for your help.

I've reposted it--give it another try now. In the code for each radio button, there was no "value" set. I added "value = 1" "value = 2" etc. to the appropriate options and now the values write to the data file.

And yes, you can alter the look of the HTML in any way you like. At worst, you can always go back to the original but in general, formatting will not affect the functionality.