ارسل ملاحظاتك

ارسل ملاحظاتك لنا







Identification Influential Observation for Two Stage LS with Jackknife after Bootstrap

المصدر: مجلة القادسية للعلوم الإدارية والاقتصادية
الناشر: جامعة القادسية - كلية الادارة والاقتصاد
المؤلف الرئيسي: Mohammed, Mohammed A. (Author)
المجلد/العدد: مج21, ع2
محكمة: نعم
الدولة: العراق
التاريخ الميلادي: 2019
الصفحات: 12 - 21
ISSN: 1816-9171
رقم MD: 1234731
نوع المحتوى: بحوث ومقالات
اللغة: الإنجليزية
قواعد المعلومات: EcoLink
مواضيع:
كلمات المؤلف المفتاحية:
2SLS | Outliers | HLP | Influential Observations | Studentized | LMS
رابط المحتوى:
صورة الغلاف QR قانون
حفظ في:
المستخلص: In the literature, there is not significant works than dealing with identifying of influential observations (10) in the two stages least square (2SLS) models. It is a clear that when a set of high leverage points (HLPs) are exist in a data set, the traditional and the robust diagnostic approaches fails to classify correctly the IOs. This issue is because of the masking and swamping affects. In this work, we propose a new robust detection approach based on Modified Cook distance to identify the 10 and to reduce the masking and swamping effects. The proposed method called modified Cook distance based on modified Generalized Studentized residuals denoted by MGt-MCD. In addition we suggested a new cut off point depend on the Jackknife after bootstrap. To evaluate the performance of our proposed method, a popular data set and a Monte Carlo simulation study were used. The results refer to the proposed method have good performance compared with some existing method to identify IOs in the 2SLS.

ISSN: 1816-9171

عناصر مشابهة