help
help > RE: Exchange blocks for unpaired t-test
Mar 28, 2017 11:03 PM | Andrew Zalesky
RE: Exchange blocks for unpaired t-test
Hi Mc,
I'm not sure if I completely follow. The exchange block structure you propose should work without errors, but I'm not sure if it makes sense, in that it will only allow subjects to be permuted within the two groups (but not between groups). For a between-group comparison, permuting within groups will not alter the value for most test statistics.
The GLM for NBS is implemented in much the same way as FSL's Randomise tool.
Andrew
Originally posted by Mchandra pal:
I'm not sure if I completely follow. The exchange block structure you propose should work without errors, but I'm not sure if it makes sense, in that it will only allow subjects to be permuted within the two groups (but not between groups). For a between-group comparison, permuting within groups will not alter the value for most test statistics.
The GLM for NBS is implemented in much the same way as FSL's Randomise tool.
Andrew
Originally posted by Mchandra pal:
Hi Andrew,
I have seen the NBS documentation regarding the exchange blocks in the context of repeated measure design. In the Table 12 of Anderson Winkler's PALM paper in Neuroimage (2014), exchange blocks were also defined in the context of unpaired t-test (no repeated measure). Group variances were not assumed to be the same. If I have 2 and 3 subjects in group1 and 2, the exchange block will be [1;1;2;2;2]. However, NBS shows error message if I supply this exchange block.
Is it possible to supply exchange block in NBS for un-paired t-test? What will be the structure of the exchange block for the above mentioned example?
Thanking you,
Mc
I have seen the NBS documentation regarding the exchange blocks in the context of repeated measure design. In the Table 12 of Anderson Winkler's PALM paper in Neuroimage (2014), exchange blocks were also defined in the context of unpaired t-test (no repeated measure). Group variances were not assumed to be the same. If I have 2 and 3 subjects in group1 and 2, the exchange block will be [1;1;2;2;2]. However, NBS shows error message if I supply this exchange block.
Is it possible to supply exchange block in NBS for un-paired t-test? What will be the structure of the exchange block for the above mentioned example?
Thanking you,
Mc
Threaded View
| Title | Author | Date |
|---|---|---|
| Mchandra pal | Mar 28, 2017 | |
| Andrew Zalesky | Mar 28, 2017 | |