Draft 0.5B1 of a CAMEO scale

In response to numerous requests, we have constructed a draft of a scale for CAMEO that is roughly equivalent to the Goldstein scale for the WEIS event coding scheme. This link will take you to a text file formated so that it can be used in the KEDS_Count [Java] program.

How was this scale constructed?

Just like the original Goldstein scale—in an ad hoc manner on a slow afternoon with a deadline approaching. Well, a little more systematically: I [Schrodt] started by matching up most of the codes for which there were exact equivalents between WEIS and CAMEO, and then filled in from there. Except that's not exactly what I did, because in a few places in the extreme ranges (codes with scale values with magnitude greater than ±7 or thereabouts) I adjusted some of the values upwards or downward a bit to open up space for behaviors that CAMEO differentiates but WEIS didn't.

Sounds very arbitrary. Wouldn't it be better to do a large scale Web-based survey of a diverse, multi-national, multi-disciplinary set of experts?

The IDEA project tried that about five years ago, and the estimates never converged. More generally, we have done a number of tests which indicate that results are not sensitive to small variations in the values of the scales, so investing a lot of effort in this is not a priority.

Are you likely to make incremental adjustments to this scale in the future?

Very likely, once we get some time to look at it more closely.

But basically you don't like it, right? You don't like scaled event data in general, right?


Last update: 23 March 2007