Journal of Official Statistics, Vol.16, No.4, 2000. pp. 363–378
On Variance Estimation for Measures of Change When Samples are Coordinated by the Use of Permanent Random Numbers
A common objective in business surveys is to compare two estimates
the same characteristic taken on two occasions, e.g., the level of production
the same month in two consecutive years, and to judge whether the observed change
is statistically significant or merely subject to random variation.
Business surveys often use samples at separate occasions that are positively
coordinated, i.e., overlapping, in order to increase the precision in estimates
of change over time. Such sample coordination will make
correlated. Some systems used for sample coordination rely on Permanent Random
Numbers (PRNs). However, the use of PRNs brings an additional component of randomness
to the rotation pattern as compared to ordinary panel rotation. This makes the
estimation of the correlation more difficult.
The SAMU system which is used by Statistics Sweden for sample coordination of
business surveys is such a system. The purpose of the present paper is to show
how to estimate the variance for measures of change such as
estimated from two separate SAMU samples.
Keywords:Survey sampling; variance estimation; estimates of change; panel designs; permanent random numbers.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group