I have to agree w/ AA's response in the article.
The sample size was small.. the sample was taken at one specific point in time during the day...
airlines know EXACTLY what their hold time is at any moment for every type of line they maintain (premium, general domestic, etc) and they maintain that information for consideration periods of time.
It is highly unlikely that any random survey of telephone response time is going to even come close to what airlines actually have measured internally... not sure why the survey people don't realize that.