[Feature Request]: Increase consistency of scatterplots

[PerPalmgren]

Description

No response

Purpose

No response

Use-case

No response

Is your feature request related to a problem?

In statistics it is common practice to insert the dependent variable on the Y axis and the independent variable on the X- axis. Of course, in correlations it does not matter but in regressions it does. However, often the variable inserted at the TOP in a correlation analysis is the dependent variable. Thus, it would be good with coherency when running JASP. Right now, when running a scatter plot for the Descriptive modules when I insert the dependent variable on the TOP (see picture) it ends up in the results on the X-ais (see picture). This makes my students slightly confused. For clarification of pictures, in Swedish “Hopphöjd” means jump height and is = dependent variable while “Benstyrka” is leg strength and is = independent variable.

Describe the solution you would like

I would suggest that the scatter plot function in the Descriptive module would work on the same premises as in the Regression module (more specifically correlation), thus when running a scatter plot the variable inserted at the TOP (see picture) ends up on the Y-axis (picture).

Describe alternatives that you have considered

No response

Additional context

[PerPalmgren]

Hi, Will this be possible to rectify and implement for the 0.17 version? I guess it is not a substantial change. All the best Per

[Kucharssim]

Yes, I will take a look into it by then!

[PerPalmgren]

Hi, Is this feature incorporated now in the new version?

[juliuspfadt]

Simon knows this better than me, but since the project it is assigned to is 0.17, I doubt it is in 0.16.2, nor will it be in 0.16.3. I apologize for the delay.

[PerPalmgren]

So, it is sure it will not be in 0.16.3?😢

[PerPalmgren]

It seems like in JASP 0.16.3 this function has also disappeared in the correlation module😥. I really think it would be beneficial if in ALL scatter plots the inserted top variable always ends up on the Y-axis (so the dependent variable gets accurately placed). Kind regards Per

[Kucharssim]

It seems like in JASP 0.16.3 this function has also disappeared in the correlation module😥

Can you clarify? We certainly did not remove any functionality...

[PerPalmgren]

Sorry I was wrong. It seems to work now. In the Regression module (more specifically correlation), thus when running a scatter plot the variable inserted at the TOP (see picture) ends up on the Y-axis. But as I stated above it would be nice to have this feature for the scatter plots in the descriptive as well.

[Kucharssim]

Yes, I am looking into it right now, but I am not sure whether we already fixed it, or whether I don't understand the issue... It seems that the two analyses show the same variables on the x and y axis:

1. Descriptives

2. Correlation

Now I think we could do a better job at making sure that the two plots look exactly identical, but I don't see an issue with the flipped axes... Or is there some setting I am missing to select? Thanks!

[PerPalmgren]

It is not the case when I run it under customizable plots (scatter plots) in the descriptive module (see below).

[Kucharssim]

Ah, ok, thanks! But then this behaviour is again the same between descriptives and correlations, where selecting the display pairwise give you the plots one by one instead of arranged in a matrix:

1. Descriptives

2. Correlation

When I think about it, it does make sense to me to display those plots as they are: My implicit thinking is that the first and second variable entered in the box are associated with the x- and y- axis of the scatter plot, respectively. So I am not entirely convinced by the argument that this is "how it's done", as that probably depends on personal preferences.

I agree though that it can be confusing when displayed next to each other within one analysis:

So I think we need to change the plots in both descriptives and correlations. Perhaps we can also reuse the code to make the plots identical.

[PerPalmgren]

Now it is me who do not really follow. I would argue in regressions in statistics the dependent variable is always put on the Y axis. So, I think it would be most logic to always let the variable inserted first (on the top) to end up on the y axis. Of course in simple correlations it does not matter but if the second step to correlation is a regression (very, very common) it would make sense to use the same. By many experience many researchers use the descriptive plots to investigate assumptions. Question! In you example is contGamma your outcome variable and contNormal your predictor (in a presumtive simple regression)?

[Kucharssim]

Now it is me who do not really follow.

Sorry for the confusion!

I would argue in regressions in statistics the dependent variable is always put on the Y axis.

Yes, we agree on this.

So, I think it would be most logic to always let the variable inserted first (on the top) to end up on the y axis.

This is where we disagree: I don't see any reason to assume that the first variable put into the variables box in descriptives or correlations is the dependent variable. This is something people may find different preferences for.

Question! In you example is contGamma your outcome variable and contNormal your predictor (in a presumtive simple regression)?

All screenshots I posted were made by entering contNormal first and contGamma second. So if you consider the first variable "outcome", then contNormal is the outcome and contGamma the predictor. I don't really think it's that productive to think about a "predictor" and a "outcome" in descriptives or correlations though. What I am focusing on here is that the axes should be consistent between the correlation matrix plots and the pairwise scatter plots both in descriptives and correlations. I will make sure that in this example contNormal would end up on the y-axis in all plots, which I think is exactly what you wanted in the first place :)

[PerPalmgren]

Yes man! I think we have a consensus and we understand each other 😜👍. Thanks🙏

[juliuspfadt]

is this closable?

[PerPalmgren]

I still think it would be of value if all all scatter plots (both in descriptive and correlation module) used the default that the variable inserted on the top (in the variable box) ALWAYS ended up on the Y-axis in the plots.

[PerPalmgren]

What heppened with this issue?

[Kucharssim]

Hi @PerPalmgren,

sorry this has not been dealt with yet. I have decided to tackle inconsistency between out plots more fundamentally (https://github.com/jasp-stats/jaspGraphs/pull/72), I plan to tackle this particular issue once that is done, but it does take more effort than just flipping the axes in one analysis. Thanks for your patience.

[PerPalmgren]

I understand completely. Thank you for not letting this pass under the radar. All the best Per

Per J. Palmgren | DC, FRCC, MMedEd, PhD | University Lecturer | Department for Learning, Informatics, Management and Ethics | Karolinska Institutet | Tomtebodavägen 18A | 171 77 Stockholm | Tel. + 46 8 524 85 294

---

Karolinska Institutet - a medical university From: Simon Kucharsky @.> Sent: den 24 maj 2023 09:56 To: jasp-stats/jasp-issues @.> Cc: Per Palmgren @.>; Mention @.> Subject: Re: [jasp-stats/jasp-issues] [Feature Request]: Increase consistency of scatterplots (Issue #1627)

Hi @PerPalmgrenhttps://github.com/PerPalmgren,

sorry this has not been dealt with yet. I have decided to tackle inconsistency between out plots more fundamentally (jasp-stats/jaspGraphs#72https://github.com/jasp-stats/jaspGraphs/pull/72), I plan to tackle this particular issue once that is done, but it does take more effort than just flipping the axes in one analysis. Thanks for your patience.

- Reply to this email directly, view it on GitHubhttps://github.com/jasp-stats/jasp-issues/issues/1627#issuecomment-1560627422, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AS5YGHU5DNY5EBUEL5MWHU3XHW5HTANCNFSM5O4FRJ7Q. You are receiving this because you were mentioned.Message ID: @.**@.>>

När du skickar e-post till Karolinska Institutet (KI) innebär detta att KI kommer att behandla dina personuppgifter. Här finns information om hur KI behandlar personuppgifterhttps://ki.se/medarbetare/integritetsskyddspolicy.

Sending email to Karolinska Institutet (KI) will result in KI processing your personal data. You can read more about KI's processing of personal data herehttps://ki.se/en/staff/data-protection-policy.

[PerPalmgren]

@Kucharssim Dear Simon, Just wanted to check in on this? All the best Per😎

[tomtomme]

this is being worked on here https://github.com/jasp-stats/jaspGraphs/pull/100