Timeline for linear mixed effects models: should I aggregate repeated trials?
Current License: CC BY-SA 4.0
2 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 16, 2021 at 7:59 | comment | added | Nice | Thank you Robert. I see that unaggregated data are ok for you. I can understand that 3 trials can bring more information than 1 trial. I just want to be sure that LME models can deal with this complication and (hopefully) have more power. In other words, unaggregated repeated data in LME models lead to less false negative without increasing false positive discoveries. I think that my skepticism is because of the "1 row, 1 person" rule. This seems the major rule in statistics. Other rules can be violated (e.g. parametric statistics on ordinal data), but the 1-1 rule has not. | |
| Mar 15, 2021 at 10:11 | history | answered | Robert Long | CC BY-SA 4.0 |