Sampling errors versus real changes in magazine audience research