Hello
I ran a study where subjects were presented, individually, two conditions (Cond, 1 & 2) at one week interval. For each condition, there was two measurements (time 1 & 2) of physiological signal (RMSSD): before a sad film and after. here is an example of data.
I have many problems:
1. when trying to use repeated mixed model with time as repeated measurement, I obtain an error message
mixed RMSSD Cond || id,:residuals(un,t(time))
repeated t() values within lowest-level panels
I see why there is a problem but can't solve it adequately as this design is a full crossed one: all subjects take all conditions at all times.
2 my second problem is relative to specifying the correct random part to take into account this repeated measurements. the subject intercept || id: is clear. but for the random slopes, it s unclear for me.
so I hesitate between the first command line mixed RMSSD Cond || id,:residuals(un,t(time)) and something like this : mixed RMSSD Cond || id:Cond || id/time. in this last case, I don't see how the repeated measurement is take into account. I have read many things about longitudinal study and mixed (Rasbeth et all is very difficult to follow). but that remain unclear for me.my interprétation is that in the second command, we allow to consider that individual may exhibit a different slope between the two condition that is subject-condition specific and that individual exhibit their own slope between T1 and T2. this last random effect would allow to test a potential time difference relative to the emotional induction. but i have serious doubt about my interpretation.
Thank you for your help.
carole
I ran a study where subjects were presented, individually, two conditions (Cond, 1 & 2) at one week interval. For each condition, there was two measurements (time 1 & 2) of physiological signal (RMSSD): before a sad film and after. here is an example of data.
Code:
* Example generated by -dataex-. For more info, type help dataex clear Sujet RMSSD time Cond id "1" 69.48 1 1 1 "1" 71.03 2 1 1 "1" 18.84 1 2 1 "1" 14.67 2 2 1 "10" 49.57 1 1 2 "10" 39.81 2 1 2 "10" 46.97 1 2 2 "10" 40.73 2 2 2 "11" 11.63 1 1 3 "11" 13.25 2 1 3 "11" 37.02 1 2 3 "11" 27.17 2 2 3 "12" 38.18 1 1 4 "12" 38.97 2 1 4 "12" 83.79 1 2 4 "12" 71.57 2 2 4 end
1. when trying to use repeated mixed model with time as repeated measurement, I obtain an error message
mixed RMSSD Cond || id,:residuals(un,t(time))
repeated t() values within lowest-level panels
I see why there is a problem but can't solve it adequately as this design is a full crossed one: all subjects take all conditions at all times.
2 my second problem is relative to specifying the correct random part to take into account this repeated measurements. the subject intercept || id: is clear. but for the random slopes, it s unclear for me.
so I hesitate between the first command line mixed RMSSD Cond || id,:residuals(un,t(time)) and something like this : mixed RMSSD Cond || id:Cond || id/time. in this last case, I don't see how the repeated measurement is take into account. I have read many things about longitudinal study and mixed (Rasbeth et all is very difficult to follow). but that remain unclear for me.my interprétation is that in the second command, we allow to consider that individual may exhibit a different slope between the two condition that is subject-condition specific and that individual exhibit their own slope between T1 and T2. this last random effect would allow to test a potential time difference relative to the emotional induction. but i have serious doubt about my interpretation.
Thank you for your help.
carole
Comment