PARTICIPANT: And that knowledge would narrow things down a lot.
CORK: Let’s have one more [question], and then we have to make a switchover in technology here.
PIAZZA: Right … Why don’t you start now, and I’ll just take any final questions here?
BANKS: Going to Jesse’s point about the many possible ways through the graph, it would seem that one would probably go a long way down the road towards documenting things adequately if you had … if you treated each module as a vertex in the graph, and then the paths linking modules in the graph, going from one to the other. One has one kind of documentation for a flow, and I think you would want to have pairwise documentation for pairs of modules. And I don’t know if you need to go through the exercise and effort of documenting all possible paths when getting connections between modules of questions could carry you a very long way down the road.
PIAZZA: Right, right. And we have to find some way, like that, that will give a close enough approximation. But partly it’s just to understand the major, just the overall view of the instrument. It’s not the case that we need a full flowchart of everything because that gets very complicated. And a lot depends on the design of the instrument. If it’s really a modular design, then of course it’s a lot easier because you just go from one to the other. Then, within those, recursively, you have the same setup.
DOYLE: The other part of the issue, of using just a flowchart, is that it’s critical to fill out the words. And once you try to fit all of the words in a flowchart, you’ve taken up a lot of the visual real estate with the words, and you can’t get a lot of your paths in there. So what you wind up lots of times is a visual of the flow and then the text of the questions. Which, if you’ve got these in some nice coordinated software package …
PIAZZA: In synch, right …
DOYLE: In synch, then it’s probably all right. But that’s where it needs to go. but it just can’t be a simple flowchart. Because there’s too many words to put into one box.
PIAZZA: Actually, TADEQ does some of that, where you go through and then click on the side; you can select the view and get the text of the question. But, as you see in the examples I’ve presented, sometimes the text of the question—as in the instrument—isn’t all that informative because it’s conditional text or it has all of these fills. So it’s really hard to know, for a particular interview, what really was done.