Dot plot of means with confidence intervals as ellipses

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

Dot plot of means with confidence intervals as ellipses

Kirill Orlov
Can SPSS (particularly, vers. 25) draw a 2D dot plot (showing means by 2 variables, as dots) with ellipses around the dots, expressing the confidence intervals or st. errors for the means?

A more general question - how does one draw ellipses around points in GPL?
Reply | Threaded
Open this post in threaded view
|

Re: Dot plot of means with confidence intervals as ellipses

Andy W
In terms of easy out of the box I don't think there is anything. I have code drawing ellipses, https://andrewpwheeler.com/2014/03/04/visualizing-multi-level-data-using-ellipses/.

I'm sure you know more about the different potential intervals Kirill than I do for this stuff, but I have some of Rick Wicklin's blog posts on the topic saved as reference, see https://blogs.sas.com/content/iml/2014/07/23/prediction-ellipses-from-covariance.html or more generally https://blogs.sas.com/content/?s=ellipse.

Slightly off topic, but one thing I am surprised no one has done is that the grammer of graphics should have mappable parameters for rotation and eccentricity (same as for color/size). I believe the underlying vector glyphs people use typically have those as parameters in the svg. That way you wouldn't need to draw a circle/ellipses as many connected lines, you could have a true vector representation. And then you could other cool stuff, such as a scatterplot with the point glyphs oriented to the mean of its group.
Andy W
apwheele@gmail.com
http://andrewpwheeler.wordpress.com/
Reply | Threaded
Open this post in threaded view
|

Re: Dot plot of means with confidence intervals as ellipses

Kirill Orlov
Thank you, Andy! I will read this through.
Reply | Threaded
Open this post in threaded view
|

Re: Dot plot of means with confidence intervals as ellipses

Kirill Orlov
In reply to this post by Andy W
Andy, thank you very much.
Your instructions and the examples in the link really helped.

That said, SPSS should introduce a special element for ellipse, so one can directly specify in GPL its location and both radii.