Dear all,
ODA paradigm (Optimal data Analysis) can only be applied if the data point to the alternative hypothesis that there is a relationship between variables (Ha)?
Please see the file attached where I explain why I think that .
If data points to Ho, the exact Fisher test (output in ODA) point to Ho. Nevertheless therer are other ODA outputs with negative probabilities (at least mine outputs)!.
Can anyone explain me this , please?
Thank you
Helena
There is no such thing as a negative probability: on the output you uploaded, p0 is better than chance, ESS0.05). The other output reports a directional analysis (note that p
There is no such thing as a negative probability: on the output you uploaded, p0 is better than chance, ESS0.05). The other output reports a directional analysis (note that p
PS:
I present the following discussion in order to prevent confusion concerning the commands I used as indicated in the two outputs that I uploaded.
In UniODA and CTA software an asterisk (*) placed at the beginning of a command line indicates that everything following the asterisk until a semi-colon (;) occurs is to be ignored by the program.
I keep "template files" for ODA programs that I use constantly in a handy directory. These are little "calculator" programs. I used such a template file to analyze the data for which I obtained and uploaded the two output files.
In the non-directional output that I uploaded, a listing of the commands that I used is presented at the beginning of the output. One of these commands is a directional (DIR) hypothesis appropriate for a four-category problem, that has an asterisk as the first character. Thus, this DIR command was ignored. Since there was no other DIR command in the code, the analysis was non-directional.
On every UniODA and MegaODA output the listing indicates all of the active functions in play by the algorithm being run. Included in this listing is the status of the hypothesis--either directional (confirmatory) or non-directional (exploratory). As seen in the directional output, this listing states that a directional hypothesis is being evaluated. And, in the non-directional output, this listing states that a non-directional hypothesis is being evaluated.
RG does *not* respect the actual formatting of uploaded output files, which in reality do *not* wrap as happens when viewing the RG-based files. In order to see the outputs as they actually appear it will be necessary to reduce the font size to prevent output lines from wrapping because the font is too large.
Dear Helena:
After reading the answers of Paul, I do not know what to say.
Thanks for the question and the answers.
Regards,
Andréa
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics