I too would like to know. I have two theories, one is that reflected signals from the wing in the turn contribute to a sort of quadrantal error that is not present in straight flight, the second is that the top of the loop aerial acquires a signal in the turn which it otherwise would not, a sort of night effect. Neither theory really attaracts me and I've never found anyone who could explain it without blathering. Luckily the ATPL syllabus steers quietly away from the subject!