Question: should each different variable type have a separate icon?

[asizemore]

I don't want to add more icons without a real need. But, it could be helpful for a user in understanding our variable constraints if they could easily identify if vars are continuous, integers, dates, etc. Right now all numeric vars (including dates) have the same icon. Should we give each type it's own?

We encountered a problem on mbio where for some studies, Age is an integer and in others, Age is a continuous var. That meant that Age worked for particular plots in one study but not another, and it was totally unclear to the user (Dan and me, in this case), what might be the reason for the discrepancy. Has this happened to others?

[dmfalke]

I just want to point out that there are two distinct variable properties: dataShape and type. In the case above, both instances of the age variable have dataShape = continuous, but one has type = number and the other has type = integer. We currently use different icons based on dataShape, with the exception of multiFilter variables.

The current values for dataShape that we support are:

• categorical
• ordinal
• binary

The current values for type that we support are:

• string
• number
• integer
• date
• longitude

Rather than rely on icons to communicate this information, maybe we should include it in a tooltip, parenthetical, or some other means.