PSPP-BUG: [bug #59111] Examine: Spreadlevel plot does not accept float t values / Power Estimation

[Friedrich Beckmann]

URL:
  <https://savannah.gnu.org/bugs/?59111>

                 Summary: Examine: Spreadlevel plot does not accept float t
values / Power Estimation
                 Project: PSPP
            Submitted by: beckmanf
            Submitted on: Sun 13 Sep 2020 10:58:05 PM UTC
                Category: Syntax Parser
                Severity: 5 - Average
                  Status: None
             Assigned to: None
             Open/Closed: Open
                 Release: None
         Discussion Lock: Any
                  Effort: 0.00

    _______________________________________________________

Details:

While doing the examine plot subdialog I noticed two problems with the
spreadlevel plots.

a) The syntax parser only accepts integer value for the t value. This does not
allow to compute square root (0.5) or reciprocal square root (-0.5) scalings
of the data although this seems to be a reasonable thing to do.

see: http://www.unige.ch/ses/sococ/cl/spss/eda/spreadlevelplot.html

b) I think there is a difference in the understanding between SPSS and PSPP
with respect to the spreadlevel plot, but I am not sure. Here is the SPSS
description:


SPREADLEVEL(n). Spread-versus-level plot with the Test of Homogeneity of
Variance table. If the keyword appears alone, the natural logs of the
interquartile ranges are plotted against the natural logs of the medians for
all cells. If a power for transforming the data (n) is given, the IQR and
median of the transformed data are plotted. If 0 is specified for n, a natural
log transformation of the data is done.


see:
https://www.ibm.com/support/knowledgecenter/SSLVMB_sub/statistics_reference_project_ddita/spss/base/syn_examine_plot.html

The gui has a selection "Power Estimation" and that refers to the mode: "If
the keyword appears alone, the natural logs of the interquartile ranges are
plotted against the natural logs of the medians for all cells."

Is "Power Estimation" different from SPREADLEVEL(0) ? In the PSPP docs this is
the same. But aren't the natural logs of the interquartile ranges different
from interquartile ranges of transformed data?




    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?59111>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/