On Nov. 13, we mailed question naires to 6,000 people drawn at random from our list of active-duty subscribers.

Recipients were asked to mail their answers to an independent firm that machine-tabulated the results to guarantee anonymity. We stopped processing incoming questionnaires Dec. 22.

About 4,000 of the 6,000 people who received questionnaires turned out to be on active duty.

Only responses from active-duty personnel were tabulated. Of those 4,000, 954 responded.

The margin of error in the survey is plus or minus 3 percentage points at the 95 percent confidence interval, meaning there is a 95 percent probability that results of the poll are accurate within 3 percentage points.

Those polled differ from the military as a whole in important ways. They tend to be older, higher in rank and more career-oriented.

Even so, it is perhaps the most representative independent sample possible because of the inherent challenges in polling service members, according to polling experts and military sociologists.

The annual poll has come to be viewed by some as a barometer of the professional career military.