Confidence+Intervals+versus+Credibility+Intervals

Confidence Intervals versus Credibility Intervals (Estimation) (GWU EMSE-271)
Index | Topics (Logical Lectures) | Lectures | Problems | Readings | Nomenclature | Concepts

In general, confidence intervals are calculated for a characteristic (like E[X} or Var[X]. Credibility intervals estimate whether X is in the interval. - EMSE 271, Fall 2009, Slide 102

Confidence Intervals "with a large number of repeated samples, 90% of the calculated confidence intervals would include the true value of the parameter." - [|Wikipedia]

Credibility intervals indicate the probability that the random interval contains the value X with the stated probability.

A confidence interval get narrow as the number of samples is increased. A credibility interval does not change based on inceases in sample sizes.


 * Confidence Interval
 * value +- test statistic * error estimate
 * "Some variation of: C.I. = PE ± CV (SE)" - PAD 202, Fall 2008, Lecture 7
 * Credibility Interval
 * LB {disribution} INV (nth percentile)
 * UB {distribution} INV (1-nth percentile)


 * Sources:**
 * EMSE 271, Fall 2009 (Slides 102-104)
 * PAD 202, Dylan Conger, Research Methods and Applied Statistics, The Trachtenberg School of Public Policy and Administration, George Washington University, Fall 2008