topic Re: Oneway Analysis Excluded points are hidden even when not set so in Discussions

Oneway Analysis Excluded points are hidden even when not set so

Hegedus1 — Fri, 09 Jun 2023 00:34:44 GMT

Hi,

JMP 16.

Doing simple oneway for a class I am teaching and in creating a simple oneway graph I would like to exclude certain points but still show them on the graph. In earlier versions of JMP this was a no brainer but it does not seem to work as previous.

The starting graph is very simple

The objective is to exclude the high points in group A and refit.

I select the points and exclude, but DO NOT HIDE.

When I do the fir again and set the axis range to the same range. I copied the axis settings and pasted them to the new graph to ensure the same range is covered. The points are not visible. Previous versions would show them. Is there something changed in JMP 16?

Andy

Re: Oneway Analysis Excluded points are hidden even when not set so

Mark_Bailey — Fri, 11 Jun 2021 12:25:36 GMT

How do you "do the fit again?"

The current behavior is consistent with the intention of the row states. The excluded rows are not entered into the analysis, and so they are not shown in the plot.

Please note that I am not justifying this behavior. I am only explaining it.

I do not have JMP earlier than version 15.2.1 but it also behaves this way.

Re: Oneway Analysis Excluded points are hidden even when not set so

dale_lehman — Fri, 11 Jun 2021 17:01:45 GMT

So, I believe the proper way to do what Hegedus wants is to do the Fit Y by X graph first, then exclude (but not hide) those rows, and then do the pooled T test. The test will ignore the excluded points, but the points will still be in the graph.

Re: Oneway Analysis Excluded points are hidden even when not set so

Hegedus1 — Mon, 14 Jun 2021 13:11:02 GMT

A second test,

Add a column with row number and plot as Bivariate. After plotting you need to rescale the Y axis since the range in truncated, however the points are shown. This is the behavior I expect by the terms hidden and excluded being separate states.

Re: Oneway Analysis Excluded points are hidden even when not set so

Mark_Bailey — Mon, 14 Jun 2021 16:28:01 GMT

We already confirmed the behavior in Bivariate and Oneway. It is known that Oneway handles excluded rows differently than the other platforms launched using the Fit Y by X command.

Re: Oneway Analysis Excluded points are hidden even when not set so

dale_lehman — Mon, 14 Jun 2021 23:12:35 GMT

OK, now I'm officially confused. I just looked at a data set with a clear outlier, using Fit Y by X. When I analyze it and compare the results to what I get if I first exclude (but not hide) the outlier, the results differ but the graphs are the same. Isn't this the way it is supposed to work? So, why is the question saying it did not work that way, and why are you saying that behavior about the Fit Y by X platform treats excluded observations differently? Is it only the Oneway analysis that differs from the rest of the Fit Y by X platform? If so, how does it treat it differently?

Re: Oneway Analysis Excluded points are hidden even when not set so

Mark_Bailey — Tue, 15 Jun 2021 11:58:39 GMT

If you exclude rows in Oneway, the excluded rows should still show in the plot, but if you redo the analysis, they won't show, even though you did not hide them. This behavior is different from the other three platforms launched by the Fit Y by X command.

Re: Oneway Analysis Excluded points are hidden even when not set so

pwenz — Sat, 03 Dec 2022 00:31:23 GMT

I just stumbled on this after using JMP for 25 years. I've always explained that hide and exclude are independent operations, typically you use them together (hence the first "hide and exclude" option under rows menu) but there are times that you'd like independent control. Seems that if you hide but not exclude, it behaves as expected- the point is not on the plot but the t-test includes the hidden point. The converse is not true, if you exclude but do not hide, there is no way to show the point on the plot. I wanted to show the outliers on the plot, but present the statistics on the outlier screened results. Is this really by design or is this a bug?