Log in
Register|
KR ★ India, 2009-03-09 07:47 (6317 d 19:14 ago) Posting: # 3346 Views: 4,139 |
|
|
Hi, Can we do two way crossover study in two groups genderwise i.e. first group includes only males and second group includes only females and also both gropus are unbalanced i.e. 67% males and 33% females out of 100% subjects? If yes, then which effects should be included in anova model? The study is for USFDA. Regards, KR |
|
JPL ☆ Vienna, 2009-03-10 15:47 (6316 d 11:14 ago) @ KR Posting: # 3349 Views: 3,438 |
|
|
Hi KR, why don't you just randomize for sex and perform a single X-over study with two periods? Regards, JPL |
|
Ohlbe ★★★ France, 2009-03-10 17:19 (6316 d 09:42 ago) @ JPL Posting: # 3350 Views: 3,282 |
|
|
Dear JPL, One classical difficulty is the need to host separately male and female subjects. A number of CROs have a single housing area. Regards Ohlbe |
|
martin ★★ Austria, 2009-03-11 00:11 (6316 d 02:49 ago) @ KR Posting: # 3351 Views: 3,415 |
|
|
dear KR ! you may set up your study as a randomized complete block design. In a randomized complete block design, experimental subjects are first divided into homogeneous blocks (i.e. males; females) before they are randomly assigned to a treatment group (i.e. sequence of your cross-over setting). Randomized complete block designs are used to split an experiment into a number of “mini-experiments” to increase precision and/or take account of some natural structure of the experiment. in an randomized complete block design all treatments (i.e. sequences in your design) occur in each block (i.e. sex in your design). for this reason you have to randomize subjects to sequences stratified by sex. you can model your data using an ANOVA model taking the block factor sex into account - in addition to the other commonly used factors for modelling data from a cross-over design. hope this helps martin |
Ing. Helmut Schütz