Independent T-test charts: X-axis range not identical
Frantisek Rysanek
Dear SOFA developers,
call this a feature request if you will... I'm revisiting some statistical packages after a long time and I have to praise SOFA for ease of use. The data import from CSV makes perfect sense (compared to e.g. the PSPP, which cannot import column heads as variable names) and the "independent T-test" appears to be the closest to what I need for the "application at hand". In its output, there are two graphs. Each is a histogram with the gaussian curve added on top - bravo. I'd like to suggest some improvements, possibly as configurable options:
1) have the X-axis per partial chart (= per group) configurable to be either centered on mean value per group, or made identical for both (all) groups, so that you can compare the the means and distributions visually along the horizontal scale. I recall the old StatGraphics for Windows 3.11 used to have such an output, back in the day when I was at school...
2) have a configurable confidence level (for the unlikely case that 5% does not fit all)
3) have ticks for the boundaries of the confidence interval visible in the chart(s) ? which implies that the X-axis would have to be scaled to span the whole confidence interval (or the per-group confidence intervals combined). This would be a killer feature in the visual output, for educational value if nothing else. To produce such a chart, at the moment I have to resort to gnuplot or macsyma or some such programming (or manual illustration in Inkscape).
Thanks for your attention...
Frank Rysanek
Grant Paton-Simpson
Hi Frantisek,
Sorry for the delay - lots for me to think about. I'll probably have some time around Christmas to have a good look and I will presumably report back with more detail then.
All the best,
Grant
SOFA Statistics
--
---
You received this message because you are subscribed to the Google Groups "sofastatistics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sofastatistic...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sofastatistics/fa8cd69c-423f-43e5-a15c-342e5ac18ee7n%40googlegroups.com.
Grant Paton-Simpson
Hi Frantisek,
I've had a good look into implementing item 3) code-wise and it is something I might do at some point. I've formally added it to my backlog for possible inclusion. It would involve calculating the bins for the total population first and force-feeding those to the individual histogram chart functions. The main challenges would be around wiring up all the different places in the interface so people can opt in or out of that approach. Given I'm of a mixed mind about adding more options I am going to hold off for now at least while I focus on ironing the last bugs out of 1.5.4 ready for a release. Main changes for 1.5.4: some important bug fixes + much better results for users of MS SQL Server thanks to a major upgrade and rewrite of the underlying code for SQL Server database connections.
Thanks for the suggestions and I hope 2021 goes well for you
(better than 2020 at least!).
All the best,
Grant (New Zealand)