Questions regarding the Data Library "Larks and Owls"
I am very grateful to JASP for developing such a wonderful tool.
I have two questions regarding the Data Library "Larks and Owls"
Q1. When I checked the "Fixed effects estimates", a table including Chronotype (1), Chronotype (2), and TimeOfDay (1) appered. Is there any way to know what exactly these values mean? (In a future version I hope to see it as ChronotypeMorning instead of Chronotype (1).)
Q2. I'm not sure if I correctly understand how to specify contrasts, I think there is a problem with contrast 6. In contrast 6, there are three "1"s and only one "-1". Is it correct? In addition, as for contrast 7, I do not understand why such an assignment would imply that the difference of the effect of task completion time between the Evening and Morning chronotype participants.
Best regards,
Daiichiro Kuroki
Comments
Hi Daiichiro,
I am not sure where I can find this data set in the JASP Data Library -- can you tell me where exactly it is?
E.J.
Hi E.J.,
Thank you for your reply, and sorry for the delay in getting back to you.
You'll see the demo from
Open -> Data Library -> 17. Mixed Models.
I also attach an image file.
Best regards,
Daiichiro
Hi Daiichiro,
Thanks for reporting the issue.
Regarding Q1, the fixed effects estimate summary table shows estimates for the sum coded contrasts -- those are usually not directly interpretable (in comparison to dummy coding) but present a better way of estimating random effect models in the presence of interactions. (I.e. The analysis uses sum contrast encoding for categorical (nominal and ordinal) predictors (R uses dummy encoding by default). This scheme is used for better interpretability of models with interactions. However, the fixed and random effects estimates will differ from those obtained from R with default settings. We advise using the 'Estimated marginal means' section for obtaining mean estimates at individual factor levels. For comparing the mean estimates, use the contrasts option.)
Regarding Q2, there is indeed an error in the 6th contrast: the 4th row should have been -1. I will fix the issue. You can think about the 7th contrast as a difference between the two simple contrasts, i.e., contrast 3 which represents the effect in the evening chronotype and contrast 1 which represents the effect on the morning chronotype. Since we compute the difference, we keep the coding for contrast 3 and multiply the coding for contrast 1 by (-1).
Cheers,
Frantisek
Hi Frantisek,
Thank you very much for your reply. I probably understand what you are saying.
About the 7 the contrast, should I consider the following?
A = ChronotypeEvening and TimeOfDayEvening
B = ChronotypeEvening and TimeOfDayMorning
C = ChronotypeMorning and TimeOfDayEvening
D = ChronotypeMorningand TimeOfDayMorning
Contrast3 = A - B
Contrast1 = C - D
Contrast7 = (A - B) - (C - D) = A - B - C + D
So, A's sign = 1, B's sign = -1, C's sign = -1, D's sign = 1.
By the way, about the plot, is MWCount predicted/estimated values? Is it not the actual observed values?
Cheers,
Daiichiro
Hi Daiichiro,
Yes, that's exactly the way how the contrasts are constructed.
Regarding the plot, the visualized points are the raw data aggregated across the random effects specified in "Background data show" field, e.g., participants in this case.
Cheers
Frantisek
Hi Frantisek,
Thank you very much for your reply.
Regarding the plot, do you mean the actual observed MWCount that are shown in the data sheet are plotted? If so, I feel the current explanation is slightly misleading. The values are not estimates, are they?
To figure out the statistically significant pattern, we create a figure that visualizes mean estimates for the idividual groups in the Plots section.
Cheers,
Daiichiro