Hi everyone. Context: Two groups (A and B) performed 30 treatment sessions. Each session has 4 periods. However, the wo groups did the intervention 6-months apart, with no overlap. The model fit of both models is equal, but theoretically, one should make more sense. Q1: Is m2 or m22 that captures this the best? Q2: the fact that I have fixed effect for group makes the previous question irrelevant somehow? Thanks! m2 <- lmer(dp ~ session + period + group + (1 | id) + (1 | date / session / period), data = dat_long ) summary(m22) m22 <- lmer(dp ~ session + period + group + (1 | id) + (1 | session / date / period), data = dat_long ) summary(m22)
Question about random-effects
4 messages · Jorge Teixeira, Thierry Onkelinx
2 days later
Dear Jorge, It is more clear when you write the nested random effects explicitly instead of the shorthand. m2 has (1 | date) + (1 | date:session) + (1 | date:session:periode) m22 has (1 | session) + (1 | date:session) + (1 | date:session:periode) Both models have the same fit but a different parameterization. See my blog post on this topic https://www.muscardinus.be/2017/07/lme4-random-effects/ 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 /////////////////////////////////////////////////////////////////////////////////////////// <https://www.inbo.be> Op za 4 feb. 2023 om 13:36 schreef Jorge Teixeira < jorgemmtteixeira at gmail.com>:
Hi everyone.
Context: Two groups (A and B) performed 30 treatment sessions. Each session
has 4 periods.
However, the wo groups did the intervention 6-months apart, with no
overlap.
The model fit of both models is equal, but theoretically, one should make
more sense.
Q1: Is m2 or m22 that captures this the best?
Q2: the fact that I have fixed effect for group makes the previous question
irrelevant somehow?
Thanks!
m2 <- lmer(dp ~ session + period + group + (1 | id) + (1 | date / session
/ period), data = dat_long )
summary(m22)
m22 <- lmer(dp ~ session + period + group + (1 | id) + (1 | session / date
/ period), data = dat_long )
summary(m22)
[[alternative HTML version deleted]]
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
1 day later
Dear Thierry Onkelin, thanks for the reply. Do you have an opinion about which model best represents the context I have described? That's my main doubt. Thank you. Thierry Onkelinx <thierry.onkelinx at inbo.be> escreveu no dia segunda, 6/02/2023 ?(s) 20:07:
Dear Jorge, It is more clear when you write the nested random effects explicitly instead of the shorthand. m2 has (1 | date) + (1 | date:session) + (1 | date:session:periode) m22 has (1 | session) + (1 | date:session) + (1 | date:session:periode) Both models have the same fit but a different parameterization. See my blog post on this topic https://www.muscardinus.be/2017/07/lme4-random-effects/ 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 /////////////////////////////////////////////////////////////////////////////////////////// <https://www.inbo.be> Op za 4 feb. 2023 om 13:36 schreef Jorge Teixeira < jorgemmtteixeira at gmail.com>:
Hi everyone.
Context: Two groups (A and B) performed 30 treatment sessions. Each
session
has 4 periods.
However, the wo groups did the intervention 6-months apart, with no
overlap.
The model fit of both models is equal, but theoretically, one should make
more sense.
Q1: Is m2 or m22 that captures this the best?
Q2: the fact that I have fixed effect for group makes the previous
question
irrelevant somehow?
Thanks!
m2 <- lmer(dp ~ session + period + group + (1 | id) + (1 | date / session
/ period), data = dat_long )
summary(m22)
m22 <- lmer(dp ~ session + period + group + (1 | id) + (1 | session / date
/ period), data = dat_long )
summary(m22)
[[alternative HTML version deleted]]
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
Dear Jorge, It is hard to give good advice without thorough insight in your case. I recommend you consult a local statistician. 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 /////////////////////////////////////////////////////////////////////////////////////////// <https://www.inbo.be> Op di 7 feb. 2023 om 22:04 schreef Jorge Teixeira < jorgemmtteixeira at gmail.com>:
Dear Thierry Onkelin, thanks for the reply. Do you have an opinion about which model best represents the context I have described? That's my main doubt. Thank you. Thierry Onkelinx <thierry.onkelinx at inbo.be> escreveu no dia segunda, 6/02/2023 ?(s) 20:07:
Dear Jorge, It is more clear when you write the nested random effects explicitly instead of the shorthand. m2 has (1 | date) + (1 | date:session) + (1 | date:session:periode) m22 has (1 | session) + (1 | date:session) + (1 | date:session:periode) Both models have the same fit but a different parameterization. See my blog post on this topic https://www.muscardinus.be/2017/07/lme4-random-effects/ 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 /////////////////////////////////////////////////////////////////////////////////////////// <https://www.inbo.be> Op za 4 feb. 2023 om 13:36 schreef Jorge Teixeira < jorgemmtteixeira at gmail.com>:
Hi everyone.
Context: Two groups (A and B) performed 30 treatment sessions. Each
session
has 4 periods.
However, the wo groups did the intervention 6-months apart, with no
overlap.
The model fit of both models is equal, but theoretically, one should make
more sense.
Q1: Is m2 or m22 that captures this the best?
Q2: the fact that I have fixed effect for group makes the previous
question
irrelevant somehow?
Thanks!
m2 <- lmer(dp ~ session + period + group + (1 | id) + (1 | date /
session
/ period), data = dat_long )
summary(m22)
m22 <- lmer(dp ~ session + period + group + (1 | id) + (1 | session /
date
/ period), data = dat_long )
summary(m22)
[[alternative HTML version deleted]]
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models