Statistical methods for astronomical data with upper limits. II - Correlation and regression

Citations Over TimeTop 10% of 1986 papers

Abstract

Statistical methods for calculating correlations and regressions in bivariate censored data where the dependent variable can have upper or lower limits are presented. Cox's regression and the generalization of Kendall's rank correlation coefficient provide significant levels of correlations, and the EM algorithm, under the assumption of normally distributed errors, and its nonparametric analog using the Kaplan-Meier estimator, give estimates for the slope of a regression line. Monte Carlo simulations demonstrate that survival analysis is reliable in determining correlations between luminosities at different bands. Survival analysis is applied to CO emission in infrared galaxies, X-ray emission in radio galaxies, H-alpha emission in cooling cluster cores, and radio emission in Seyfert galaxies.

Related Papers

• → A Catalog of Compact Groups of Galaxies in the SDSS Commissioning Data(2004)80 cited
• → THE VLT SINFONI Mg ii PROGRAM FOR LINE EMITTERS (SIMPLE). II. BACKGROUND QUASARS PROBING $Z\sim 1$ GALACTIC WINDS(2015)65 cited
• → Outflows in the Disks of Active Galaxies(2019)33 cited
• → X-ray and infrared selected AGN. II - Optical spectroscopy(1990)41 cited
• → Key Source Categories in the Galaxy(2005)