To Do: Jan 20

[dicook]

• [x] Remove values from axes of experimental design factors, Fig 3, 4, 6, 8, 9, 10. I think it really is better for these to be done as facetted plots, to remove redundant axis labels, too.
• [x] Not clear we need all of the lineup plots. Re-visit this when writing is complete. At the very least they need to be re-organised to be after the experimental design plots.
• [x] Aspect ratio of plots needs to be checked. The aspect ratio of design plots needs to be the same as used in the experiment. The plots in Figure 2 lineup look too short relative to width to me, to be aesthetically pleasing, and I just changed it from height=5 to height=7. I think your lineups used in the study are closer to this aspect ratio. So then the plots in Fig 3 etc need to have the same (approx) aspect ratio. (Patrick: the actual height width setting is 7 * 7)
• [x] #6
• [x] All the experiment components need to be in the repo. Are all the lineups used there? It doesn't look like it. Maybe I'm wrong. (Patrick: all the lineups used in the experiment and the corresponding responses are stored in the visage package. We use them by importing the package. The paper repo only stores the simulation of the conventional tests.)
• [x] Because we are focusing on only the uniform, should the different X distributions in the experimental design be removed?
• [x] Need to use either "Fitted values" or "X_raw" but not both. It is confusing. Decide on one, and use it throughout the plots and text.

[dicook]

• [ ] Grammarly check done on each paragraph

[dicook]

• [x] Caption for figure 14 is changed to reflect what we learn - please check this.
• [ ] Check all figures are referred to, and order of figures does match mention in text

[dicook]

• [x] Percentages in section 4.3 need to be conditional, to match the mosaic plot. They are currently marginal proportions. If we wanted to summarise with marginal proportions we would need to use a fluctuation diagram. I think that it is better to use conditional.

[TengMCing]

• [x] Fix - Title of plots are inconsistent: some use A, B, C, ... and some do not.

[TengMCing]

• [x] Remove lineup ID from the captions. They are meaningless to the reader.

[TengMCing]

• [ ] Fix - Some plots use capital labels. Make them consistent.

[TengMCing]

• [x] Should we add more confidence intervals to plots of power curves?

[srvanderplas]

CIs on curves like that are very prone to the sine illusion, so be careful interpreting them if you do include the intervals.

On Tue, Jan 24, 2023 at 1:40 PM Weihao (Patrick) Li < @.***> wrote:

• Should we add more confidence intervals to plots of power curves?

— Reply to this email directly, view it on GitHub https://github.com/TengMCing/PhD/issues/5#issuecomment-1402492196, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAQHKFPRYUW2ZAOARKIN7QLWUAV3BANCNFSM6AAAAAAUA6IDXM . You are receiving this because you are subscribed to this thread.Message ID: @.***>