You are not currently logged in.
Access your personal account or get JSTOR access through your library or other institution:
If You Use a Screen ReaderThis content is available through Read Online (Free) program, which relies on page scans. Since scans are not currently available to screen readers, please contact JSTOR User Support for access. We'll provide a PDF copy for your screen reader.
Uncertainty, Entropy, Variance and the Effect of Partial Information
James V. Zidek and Constance van Eeden
Lecture Notes-Monograph Series
Vol. 42, Mathematical Statistics and Applications: Festschrift for Constance van Eeden (2003), pp. 155-167
Published by: Institute of Mathematical Statistics
Stable URL: http://www.jstor.org/stable/4356236
Page Count: 13
Since scans are not currently available to screen readers, please contact JSTOR User Support for access. We'll provide a PDF copy for your screen reader.
Preview not available
Uncertainty about the value of an unmeasured real random variable Y is commonly represented by either the entropy or variance of its distribution. If it becomes known that Y lies in a subset A of the support of Y's distribution, one might expect uncertainty about Y to decrease. In other words, one might expect the entropy and variance of Y's conditional distribution given Y ∈ A to be less than their counterparts for the unconditional distribution. Going further it might be conjectured that the uncertainty about Y would be greater given the knowledge that Y ∈ B as compared with Y ∈ A ⊂ B. We do not know whether these conjectures are correct. However, we give sufficient conditions in certain cases where they are true. In particular, when Y is normally distributed we can make considerable progress. For example, we show in the case that A = [a, b] and Y normally distributed with mean η and variance 1, that the variance of the conditional distribution of Y given that a ≤ Y ≤ b is less than that of the unconditional distribution, thereby confirming our intuitive reasoning in this case. This last example also shows that for this exponential family the variance is less than 1 for all a < b and all η--a result that is not known among the experts on exponential families we consulted.
Lecture Notes-Monograph Series © 2003 Institute of Mathematical Statistics