作者: Riccardo. Gatto
DOI: 10.1080/00949650008812002
关键词:
摘要: In this article we show the effectiveness and accuracy of test statistic based on expnnent saddlepoint approximation for density M-estimators, proposed by Robinson, Ronchetti Young (1999), testing simultaneous hypotheses mean variance a wrapped normal distribution. We base trigonometric method moments estimator Gatto Jammalamadaka (l999b), which admits M-estimator representation necessary test. This has an approximate chi-squared distribution, asympiotically up to second order, high is shown numerical simulations.