Comparison of MED-T and MAD-T interval estimators performance under different conditions for estimating the mean of a positively skewed distribution

---

Thesis Type: Postgraduate

Institution Of The Thesis: Gazi Üniversitesi, Fen Bilimleri Enstitüsü, Turkey

Approval Date: 2013

Student: GÖZDE ÖZÇIRPAN

Supervisor: MELTEM EKİZ

Abstract:

Several researchers proposed various interval estimators for estimating the mean of a positively skewed distribution. Banik and Kibria (2007) compared the MED-T and MAD-T confidence intervals with those proposed by various researchers, under similar simulation conditions. In this study, the best performance of MED-T and MAD-T interval estimators are investigated in terms of various distributions, skewnesses, sample sizes and confidence levels. Towards this aim, simulation studies are made by using Matlab R2007b. In general, MED-T interval estimator gave better results in terms of coverage probabilities of confidence interval. Coverage probabilities for MED-T interval estimator were close to confidence levels for low skewness and small sample sizes. In case of moderately skewness it has been observed that the coverage probabilities has given better results for large sample sizes. MAD-T interval estimator has the narrower interval in terms of the widths of confidence intervals.