Thank you! About the first option: I get to the point of using the chart creation feature of the pivot table editor, but there is only an option for a "bar" plot. If I use that I get the mean, upper CI and lower CI as separate bars, instead of as an error bar plot. Am I supposed to do something else with the "CE" (what is that?)? :)
About the second option: I am not entirely sure how to do it with OMS, but it is small enough that I can copy/paste all entries into a new dataset. How do I create an errorbar plot with that data though? It looks like I can't really select the error bar position using Chart Builder.
Thanks a lot!
------------------------------
Enzo De Sena
------------------------------
Original Message:
Sent: Mon May 23, 2022 05:48 PM
From: Jon Peck
Subject: Error bar plot with bootstrap-based 95% confidence intervals on SPSS
I can think of two possibilities, neither of which is direct.
One would be to open the table in the pivot table editor and use its chart creation feature to get the bootstrap chart. It would probably require some editing in the CE afterwards.
The other approach would be to use OMS to capture that table as a dataset, and then use the Chart Builder on those data to create the chart.
--
Original Message:
Sent: 5/23/2022 5:12:00 PM
From: Rick Marcantonio
Subject: RE: Error bar plot with bootstrap-based 95% confidence intervals on SPSS
Good question.
I'm sorry, but there is no non-trivial way that I can think of.
We should consider this as an enhancement for future versions.
------------------------------
Rick Marcantonio
Quality Assurance
IBM
Original Message:
Sent: Mon May 23, 2022 02:21 PM
From: Enzo De Sena
Subject: Error bar plot with bootstrap-based 95% confidence intervals on SPSS
I have 10 groups of non-normal data, and I would like to generate a plot with the mean of each group and associated the 95% confidence interval (CI) using bootstrapping.
I can generate the plot using "Chart Builder" -> "Simple Error Bar" but I believe that this assumes normal data (i.e. ), right? I see no option there to use bootstrapping to estimate the 95% CIs.
I can calculate the bootstrap-based 95% CI using Descriptive Analysis->Explore, but only in table-based form. Some plots can be generated too, but it looks like they are only box plots (and I need the means).
How do I put the two things together?
------------------------------
Ozne
------------------------------
#SPSSStatistics