Hello to everybody, I have read that in some cases the same variable could be consider either a random or a fixed effect at the same time. But do not have any example for this. Could someone explain me when is this the case? When there is a justification to considered a variable in fixed effect term of the model and also in the random part? I am quite interested on that. I am thinking in the next example: Imagine we are interested in knowing if the exam marks of pupils depends on their Sex. So, you choose two different schools to check for this, but School 1 is only for boys and School 2 is only for girls. Within each school you sample many classes. So, we would have ExamMarks as response variable. SchoolSex as explanatory variable (in this given example would be the same School or Sex), with two levels SchoolBoys and SchoolGirls Class as explanatory variable, with many levels (8 per school), coded as Class1, Class 2....up to Class16. The model would be (lme4): lmer(exammarks ~ Schoolsex + (1|Class), data=data) So, we treat class as a random factor. Should we consider in this case, the random part of the model as a nested random and specify it as (1|Class|Schoolsex) ? In that case the model would be: lmer(ExamMarks ~ Schoolsex + (1|Class|Schoolsex), data=data) Thank you very much, Julia
random intercept or nested random?
2 messages · Julia Chacon Labella, Thierry Onkelinx
1 day later
Dear Julia, I wrote a blogpost on this: https://www.muscardinus.be/2017/08/fixed-and-random/ Best regards, ir. Thierry Onkelinx Statisticus / Statistician Vlaamse Overheid / Government of Flanders INSTITUUT VOOR NATUUR- EN BOSONDERZOEK / RESEARCH INSTITUTE FOR NATURE AND FOREST Team Biometrie & Kwaliteitszorg / Team Biometrics & Quality Assurance thierry.onkelinx at inbo.be Havenlaan 88 bus 73, 1000 Brussel www.inbo.be /////////////////////////////////////////////////////////////////////////////////////////// To call in the statistician after the experiment is done may be no more than asking him to perform a post-mortem examination: he may be able to say what the experiment died of. ~ Sir Ronald Aylmer Fisher The plural of anecdote is not data. ~ Roger Brinner The combination of some data and an aching desire for an answer does not ensure that a reasonable answer can be extracted from a given body of data. ~ John Tukey /////////////////////////////////////////////////////////////////////////////////////////// 2018-02-17 11:54 GMT+01:00 Julia Chacon Labella <juliachacon at gmail.com>:
Hello to everybody,
I have read that in some cases the same variable could be consider either a
random or a fixed effect at the same time. But do not have any example for
this. Could someone explain me when is this the case? When there is a
justification to considered a variable in fixed effect term of the model
and also in the random part? I am quite interested on that.
I am thinking in the next example:
Imagine we are interested in knowing if the exam marks of pupils depends on
their Sex. So, you choose two different schools to check for this, but
School 1 is only for boys and School 2 is only for girls. Within each
school you sample many classes.
So, we would have ExamMarks as response variable.
SchoolSex as explanatory variable (in this given example would be the same
School or Sex), with two levels SchoolBoys and SchoolGirls
Class as explanatory variable, with many levels (8 per school), coded as
Class1, Class 2....up to Class16.
The model would be (lme4):
lmer(exammarks ~ Schoolsex + (1|Class), data=data)
So, we treat class as a random factor.
Should we consider in this case, the random part of the model as a nested
random and specify it as (1|Class|Schoolsex) ?
In that case the model would be:
lmer(ExamMarks ~ Schoolsex + (1|Class|Schoolsex), data=data)
Thank you very much,
Julia
[[alternative HTML version deleted]]
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models