search

rcalinjageman / esci

Estimation Statistics with Confidence Intervals
https://rcalinjageman.github.io/esci/
19 stars 10 forks source link

Estimation figure in Jamovi #56

Open maximgalchenko opened 6 months ago

maximgalchenko commented 6 months ago

Hi,

I was really glad to see so great library in jamovi. I have misundestanding with estimation figure for paired median difference test. I have following figure in the Jamovi (paired median difference test): image I have some questions about data (blue dots) part:

  1. I have zero values in the data set. I see dots only on the level, equal to 1 by y axis.
  2. I have not pair (7, 8) by y axis in the data, but on the plot I see pair, that looks like (7, 8). I made some experiments and found that in the data this is 7, 7 pair. It looks like value raised on the plot, but in fact values before and after are equal Please, help me to understand figure. Is there any kind of transformation for raw data? Are data calibrated in any way?
rcalinjageman commented 6 months ago

Interesting. Could you send me the data file as well? I will take a look.

Bob Calin-Jageman Professor, Psychology Neuroscience Program Director 210 Parmer Hall Dominican University 7900 West Division Street River Forest, IL 60305 v: 708.524.6581 f: 708.524.5990 http://dr-bob.homeip.net @.***

On Tue, Apr 2, 2024 at 9:02 AM Maxim Galchenko @.***> wrote:

Hi,

I was really glad to see so great library in jamovi. I have misundestanding with estimation figure for paired median difference test. I have following figure in the Jamovi (paired median difference test): image.png (view on web) https://github.com/rcalinjageman/esci/assets/25666155/fec9d3ca-2d2a-4bf3-a24d-01aaee8a387a I have some questions about data (blue dots) part:

  1. I have zero values in the data set. I see dots only on the level, equal to 1 by y axis.
  2. I have not pair (7, 8) by y axis in the data, but on the plot I see pair, that looks like (7, 8). I made some experiments and found that in the data this is 7, 7 pair. It looks like value raised on the plot, but in fact values before and after are equal Please, help me to understand figure. Is there any kind of transformation for raw data? Are data calibrated in any way?

— Reply to this email directly, view it on GitHub https://github.com/rcalinjageman/esci/issues/56, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAOQWYK6FQAFATRMQRGM2UTY3K27NAVCNFSM6AAAAABFTOUOIWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGIZDANRRGMZTKNA . You are receiving this because you are subscribed to this thread.Message ID: @.***>

maximgalchenko commented 6 months ago

Thank you,

Here are Excel file with data and zipped omv file for Jamovi. On the plot:

  1. For Value, after you can see value, equal to 8, but there is not such values in data.
  2. For both values 0 are not exists on the plot, but exists in the data.

test_data.xlsx test_data_esci.omv.zip

rcalinjageman commented 6 months ago

I see the problem: The data columns in your jamovi file are set to be ordinal, but my code is receiving them from jamovi as nominal. Specifically, my code seems to be picking this up where the lowest value in a column is coded as "level 1" and passed as a 1, then the next lowest as level 2, etc.

I must need to do something to retrieve from jamovi the correct values. It must be possible, because jamovi's t-test function handles this data correctly. I'll look into it.

In the meantime, if you set your columns to continuous (even though that's not conceptually correct), all will be well.

maximgalchenko commented 6 months ago

Thank you! I change types and now have nice plots. Hope it can be fixed for ordinal scale. Your library gives excellent results and this plot is very helpful. Thank you for the work!