Premium
Power calculations for tests of trend with missing observations
Author(s) -
Cabilio Paul,
Tilley Jessica
Publication year - 1999
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/(sici)1099-095x(199911/12)10:6<803::aid-env397>3.0.co;2-o
Subject(s) - missing data , statistics , econometrics , trend analysis , measure (data warehouse) , statistical power , computer science , mathematics , data mining
A record of observations which were to be taken over time at regularly spaced intervals, in fact has one or more observations missing. In this framework, it is wished to test for the presence of trend. Using various trend models, empirical power comparisons are conducted using Spearman and Kendall based statistics which either ignore the gaps in the data, or integrate this information in arriving at a measure of trend. For the cases considered, the statistics which use the information on data gaps generally appear to perform as well as, and in many cases substantially better than, the statistics which ignore the presence of missing observations. Copyright © 1999 John Wiley & Sons, Ltd.