Last week I have been to a place where there was a discussion going on about attribute data. peopel beleive that attributs data is overlooked because of the variable data specailly in the auto industry.

I assume you are referring to variables data versus attribute data. Variables data means that I have an actual number - the bolt is 3.285 inches long. Attribute data would include the bolt has an acceptable finish, versus an unacceptable finish (whatever the operational definition of that may be).

With attributes data, you do commonly run into go-no go information. It met specification or not. In terms of finish, that may require considerable judgement by a person, and that judgement may vary from person to person. Generally, we can all agree on the length of a bolt (as long as we use the same methodology, and measure from the same point to point).

Generally it is not a good idea to convert variables data into attributed. For example, if the bolt has a specification on its length, we can convert it to acceptable or not-acceptable. But we lose a lot of information. Was it just a little out of specification, or a lot? Against which limit - too big or too small. Statistically, we also need a much bigger sample when dealing with attribute data rather than the actual variables data.

thing is that if we are using variable data more often & avoiding the attribute data.

thing is that if we are using variable data more often & avoiding the attribute data.What attribute data do you feel is being avoided?