I have 3 different methods of power calculator, but measured it in 2 different conditions and in 6 different positions.
The problem is that Friedman in Jamovi can´t analyze per group. Some that I can do in this situation?
I only can compare between methods, but not separate in escha condition and in each different position.
Friedman Test
Re: Friedman Test
This is not a limitation of jamovi. As discussed previously, the Friedman test is a non-parametric alternative to a repeated-measures one-factor ANOVA. Whether implemented in jamovi or any other software, the Friedman test is not applicable to multi-factor designs.
e.g., https://www.ibm.com/docs/en/spss-statis ... edman-test
The are no standard, broadly accepted, non-parametric methods for analyzing data from designs that can produce interactions. In such situations, people try to make the parametric approach work by transforming the dependent variable, trimming the distributions, etc. Alternatively, one may opt to simply take a piecemeal approach--for example, conducting a bunch of two-condition, non-parametric tests such as Wilcoxon signed rank and/or Mann-Whitney U.
e.g., https://www.ibm.com/docs/en/spss-statis ... edman-test
The are no standard, broadly accepted, non-parametric methods for analyzing data from designs that can produce interactions. In such situations, people try to make the parametric approach work by transforming the dependent variable, trimming the distributions, etc. Alternatively, one may opt to simply take a piecemeal approach--for example, conducting a bunch of two-condition, non-parametric tests such as Wilcoxon signed rank and/or Mann-Whitney U.
Re: Friedman Test
I understand.
I´m trying to do the same with JASP and report different results. Which is the difference?
In addition JASP have option "Optional Grouping factor", posthocs results change when add some variable but global anova result show error. Which is the reason?
https://ibb.co/MCvkbch
https://ibb.co/ZSm5c2w
https://ibb.co/M8TXp4Z
I´m trying to do the same with JASP and report different results. Which is the difference?
In addition JASP have option "Optional Grouping factor", posthocs results change when add some variable but global anova result show error. Which is the reason?
https://ibb.co/MCvkbch
https://ibb.co/ZSm5c2w
https://ibb.co/M8TXp4Z
Re: Friedman Test
Within your repeated-measures ANOVA that also includes a non-repeated measures factor, JASP is letting you choose options that you should not be choosing . This is because Friedman's is only applicable (and only works) when there is a single factor and when that single factor is a repeated-measures factor. Consequently, when you choose the particular JASP options that you should not be choosing, JASP gives you an error message. If, from the beginning, you only include one factor (the repeated-measures factor you call "power") in your analysis, then: You can opt to perform a non-parametric test, you won't be ale to choose options you shouldn't be choosing, and you'll get Friedman's test output.
Re: Friedman Test
Ok I understand.
But why the result doing the same is different for JASP and Jamovi for the same data and the same Durbil-Conover post-hoc?
For example posthoc comparisson for P_B and P_C is p=0.165 for Jamovi and p=0.209 for JASP.
But why the result doing the same is different for JASP and Jamovi for the same data and the same Durbil-Conover post-hoc?
For example posthoc comparisson for P_B and P_C is p=0.165 for Jamovi and p=0.209 for JASP.
Re: Friedman Test
I see that the chi-square result for the Friedman test is identical in jamovi and JASP.
Not so for the post-hoc tests. There is disparity both in the t values and the p values.
Perhaps you could submit bug reports to jamovi and jasp. It appears that some R packages, or some versions of particular R packages produce wrong output, and jamovi and JASP are using different R packages or perhaps different versions of the same R package.
See https://github.com/jasp-stats/jasp-issues/issues/314
Not so for the post-hoc tests. There is disparity both in the t values and the p values.
Perhaps you could submit bug reports to jamovi and jasp. It appears that some R packages, or some versions of particular R packages produce wrong output, and jamovi and JASP are using different R packages or perhaps different versions of the same R package.
See https://github.com/jasp-stats/jasp-issues/issues/314