Q1220987
11-24-2009, 10:58 PM
Hello all,
I am presently trying to write up the results for my clinical study.
I have n=203 non-clinical group and n=5 (was 6 but outlier removed) clinical group. I am comparing them on 2 DV's.
Obviously I had problems all over the place due to very unequal n, then unequal variance. The larger SD is in the small group too!!
So.... I weighting cases to correct for scale and proportion. The choice to weight cases rather than deleted cases was made as unequal n has resulted from the nature of the respective populations from which the samples came.
I made the weighting calcs by:
1. to get the total population of people aged between 13 and 17 in Australia (1,743,547) and get a sample ratio for the non- clinical sample (203) = 0.0001164
2. take the total population again and use a figure of 19% (13-17 reporting MH probs in study quoted in study 1) and get sample ration for clinical sample (12) = 0.0000362
3. I then used these figure to calculate integrated weights as per instruction from http://www.spsstools.net/Tutorials/WEIGHTING.pdf making final weight for non-clinical = 0.88995365 and clinical = 2.861617415
I did this to address the unequal n problem.
However, I still have unequal variance. As I am using SPSS and conducting a One way ANOVA (x2) with 1 DV and 1IV with 2 groups, no "equal variance not assumed" statistic is offered, nor can I conduct any post hoc tests as my IV only has 2 levels !!!!!!!!! I know that I could run it in MANOVA with 2 DV's at once, but there are even more assumptions for MANOVA!
I am wondering if anyone out there can make a comment on my weighting rationale and on what to do about reporting F with unequal variance.
I have attached a word document showing the SPSS output and an excel document with the data.
I appreciate any help
Thanks
Rachel :D
I am presently trying to write up the results for my clinical study.
I have n=203 non-clinical group and n=5 (was 6 but outlier removed) clinical group. I am comparing them on 2 DV's.
Obviously I had problems all over the place due to very unequal n, then unequal variance. The larger SD is in the small group too!!
So.... I weighting cases to correct for scale and proportion. The choice to weight cases rather than deleted cases was made as unequal n has resulted from the nature of the respective populations from which the samples came.
I made the weighting calcs by:
1. to get the total population of people aged between 13 and 17 in Australia (1,743,547) and get a sample ratio for the non- clinical sample (203) = 0.0001164
2. take the total population again and use a figure of 19% (13-17 reporting MH probs in study quoted in study 1) and get sample ration for clinical sample (12) = 0.0000362
3. I then used these figure to calculate integrated weights as per instruction from http://www.spsstools.net/Tutorials/WEIGHTING.pdf making final weight for non-clinical = 0.88995365 and clinical = 2.861617415
I did this to address the unequal n problem.
However, I still have unequal variance. As I am using SPSS and conducting a One way ANOVA (x2) with 1 DV and 1IV with 2 groups, no "equal variance not assumed" statistic is offered, nor can I conduct any post hoc tests as my IV only has 2 levels !!!!!!!!! I know that I could run it in MANOVA with 2 DV's at once, but there are even more assumptions for MANOVA!
I am wondering if anyone out there can make a comment on my weighting rationale and on what to do about reporting F with unequal variance.
I have attached a word document showing the SPSS output and an excel document with the data.
I appreciate any help
Thanks
Rachel :D