Urgent help to plots would be appreciated

Comments

• Dmartin427

  January 19

  May I recommend going into Descriptive Statistics --> Customizable Plots --> Boxplots (in colorblind color palette) to address all of your concerns in one fell swoop?
• andersony3k

  January 19

  FYI. This graphic describes the elements of a box plot: https://cdn1.byjus.com/wp-content/uploads/2020/10/Box-Plot-and-Whisker-Plot-1.png

  R
• vinschger

  January 20 edited January 20

  @Dmartin427 and @andersony3k

  Thanks to both of you. This helped a lot!

  Do you think I should do a feature request on the github page for enabling "customized plots" (color set, min/max of y-scale, rotating the labels for the x-axis) for the t-test section? Currently it's indeed only available for the descriptive statistics section, but would make sense to have it in all plots?
• vinschger

  January 20

  @andersony3k thank you for the link. in your graphic the whiskers are showing the minimum and maximim values (or am I wrong)? But according to JASP, see screenshot above, they would show the SE?! I assume this is not the same, or am I wrong. Furthermore as far as I know there are boxplots that show (additionally?) the mean and not (only?) the median. I know this are basic statistics knowledge and if this here should not be the apprpriate forum to ask this please feel free to let me know.
• EJ

  January 20

  Also, you might want to check out our raincloud plot functionality, which offers a lot of customization.

  https://jasp-stats.org/2024/07/26/raincloud-plots-in-jasp-incredibly-useful-surprisingly-easy/

  There is an accompanying 1.5 hr video on our YouTube channel...

  EJ
• andersony3k

  January 20

  @vinschger

  Usually, and in JASP: If the data maximum is not greater than 'the 75th percentile plus (1.5 times the interquartile range)', then the upper whisker extends to the data maximum. Otherwise, the upper whisker extends to 'the 75th percentile plus (1.5 times the interquartile range)'. Likewise, if the data minimum is not less than 'the 25th percentile minus (1.5 times the interquartile range)', the lower whisker extends down to the data minimum. Otherwise, the lower whisker extends to the 'the 25th percentile minus (1.5 times the interquartile range)'.

  The box's center line is the median.

  You will sometimes see the mean added, in the form of a large dot, but that is not usual (and I don't think JASP'S box plots include that option). As referred to by @EJ, above, you can use raincloud plots (in Descriptives) to present a violin (rotated, smoothed histogram) along with either a box plot or a mean with SE bars.

  If you absolutely need to show both means and medians together--especially if you're a beginner on a tight deadline--I would suggest creating two separate graphs and then presenting them side-by-side: (a) a rain cloud graph configured to show the point cloud, mean, SE, and smoothed histogram, and (b) a box-plot graph.

  R
• vinschger

  January 20

  Thanks to all of you. I used the plot in the t-test section because it provides a view where paired cases are shown with connecting lines. As far as I understand, such a figure is not easily or directly achievable in the descriptive plots section. Afterwards, I modified the colors to grayscale and adjusted the overlapping x-axis labels in PowerPoint.

  The plots in the t-test section lack color editing. Wouldn’t it make sense to add this feature for all plots in JABS? Additionally, adjusting the y- and x-axes (e.g., the ability to define min/max values for the y-axis and set tick scale/intervals) would be useful.

  Would it make sense to submit these suggestions as feature requests in the GitHub forum?

  The feature request to include the p-value directly in the figure has already been submitted there.

  
  

  

  
  
• EJ

  January 20

  Did you notice the graph editor? For most graphs, "edit image" (activated upon clicking the small black triangle on top of each figure) allows you to define the ranges and ticks on the axes.

  Cheers,

  EJ
• vinschger

  January 21 edited January 21

  @EJ

  1) yes, I know. but when changing the limits in the edit something strange happens to the raincloud plots, see screenshot. or is this expected behavior?

  2) Furthermore, color changes are only possible in the descriptive plots, not in the accompanying plots to the t-test. Should I submit this as a feature request?

  3) As far as I have understood tick intervals cannot be adjusted in the edit plot settings.

  4) Isn't there a lot of redundancy in such a figure (showing single cases, box plots and raincloud plots) or is this a combination that you often use in your publications?

  5) The median value is shown with a really thick line, especially in my first screenshot, see above. Any reason for this from a statistical point of view?