Testing parallelism among the profiles after a certain time period /
The typical initial null hypothesis in a repeated measures design is no interaction between all treatment levels and all time periods, i.e., parallel profiles. A decision to fail to reject this hypothesis leads to testing possible secondary hypotheses such as no time trends, i.e., flat profiles, an...
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
1998.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=737708701&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | The typical initial null hypothesis in a repeated measures design is no interaction between all treatment levels and all time periods, i.e., parallel profiles. A decision to fail to reject this hypothesis leads to testing possible secondary hypotheses such as no time trends, i.e., flat profiles, and no treatment differences, i.e., coincidental profiles. On the other hand, a decision to reject the initial null hypothesis and reach a conclusion of nonparallel profiles across all time periods leads to conducting some different secondary tests. One such test of interest is the test of parallel profiles after a certain time period. Currently, there are two methods that test this claim. The first is referred to as the throw away method since it discards data that is taken before the time period of interest. The second method is referred to as the drop test since it compares the full model, one that uses all of the data, with the reduced model, one that uses only data occurring before the time period of interest. Since there are some disadvantages of these two methods, a new procedure is introduced to test the secondary null hypothesis of interest. The test statistics for the three methods are developed under three common variance-covariance assumptions, namely compound symmetry, autoregressive, and unstructured. Additionally, these three methods are compared with respect to calibration, power, and robustness. |
|---|---|
| Item Description: | Vita. "Major Subject: Statistics". |
| Physical Description: | x, 143 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilms Inc. |
| Bibliography: | Includes bibliographical references: pages 133-135. |