jslefche
commented
4 years ago This may because of a new change in the emmeans package. I wont have time to look at it this week but if you can do some digging, that would be appreciated
Jonathan S. Lefcheck, Ph.D. Tennenbaum Coordinating Scientist MarineGEO: https://marinegeo.si.edu/ Smithsonian Institution Phone: +1 (443) 482-2443 www.jonlefcheck.nethttp://www.jonlefcheck.net
From: Leonardo R. Jorgemailto:notifications@github.com Sent: Wednesday, September 16, 2020 11:33 AM To: jslefche/piecewiseSEMmailto:piecewiseSEM@noreply.github.com Cc: Subscribedmailto:subscribed@noreply.github.com Subject: [jslefche/piecewiseSEM] Model fits but summary and coefs throw errors (#214)
External Email - Exercise Caution
Really excited to find out about this package. Being able to fit a SEM with mixed effects is quite amazing.
However, I'm getting some trouble fitting (an admittedly quite complex) model, and I'm lost on how to figure what could be the issue. I managed to get the psem function to work, running the code below:
SEM.L1 <- psem(
glmer(A3_Endangered ~ A1_LogN.2000. + A6_PNGbible + (1|Language), data = Ind.cl,
family = "binomial", control = glmerControl(optimizer = "bobyqa")),lmer(as.numeric(B1_Urbanization) ~ as.numeric(B4_ParentJob) + (1|Language), REML = F, data = Ind.cl),
lmer(as.numeric(B4_ParentJob) ~ B3_ParentEdu + (1|Language), REML = F, data = Ind.cl),
lmer(as.numeric(C1_ParentMaxFluency) ~ A1_LogN.2000. + A3_Endangered + (1|Language), REML = F, data = Ind.cl),
lmer(as.numeric(C2_HomeLang) ~ C3_SameLang + as.numeric(C1_ParentMaxFluency) + B3_ParentEdu + (1|Language), REML = F, data = Ind.cl),
glmer(C3_SameLang ~ A1_LogN.2000. + A3_Endangered + B3_ParentEdu + (1|Language), data = Ind.cl,
family = "binomial", control = glmerControl(optimizer = "bobyqa")),lmer(D4_TradSkills ~ B3_ParentEdu + as.numeric(B1_Urbanization) + (1|Language), REML = F, data = Ind.cl),
lmer(D5_ModSkills ~ B3_ParentEdu + as.numeric(B1_Urbanization) + (1|Language), REML = F, data = Ind.cl), #Leaving the connection from
traditional skills out, just because it's very difficult to decide on a sign for that arrow, and the
other arrows arriving into these probably already account for possible correlations between these?
lmer(L1_BodyParts ~ A1_LogN.2000. + A3_Endangered + A6_PNGbible + as.numeric(B1_Urbanization) +
B3_ParentEdu + as.numeric(C1_ParentMaxFluency) + as.numeric(C2_HomeLang) + C3_SameLang +
D4_TradSkills + D5_ModSkills + (1|Language), data = Ind.cl, REML = T),glmer(cbind(L1_BodyParts,24-L1_BodyParts) ~ A1_LogN.2000. + A3_Endangered + A6_PNGbible +
as.numeric(B1_Urbanization) + B3_ParentEdu + as.numeric(C1_ParentMaxFluency) + as.numeric(C2_HomeLang) + C3_SameLang +
D4_TradSkills + D5_ModSkills + (1|Language), data = Ind.cl,
family = "binomial", control = glmerControl(optimizer = "bobyqa")),
data = Ind.cl
)
(Don't mind the commented out glmer... that was my first try at fitting the effects on the main response variable, but I think the fact it is a set of 24 binary values for each observation messes up even more with the fit and the model doesn't work at all)
When trying to run summary or coefs I get the following (different) errors. I couldn't find info on them (and they are not very clear):
summary(SEM.L1)
| | 0%Error
in list2env(data) : first argument must be a named list
coefs(SEM.L1)
Error in emmeans::emmeans(model, specs = v, data = data) :
No variable named A3_Endangered in the reference grid
However, just printing the object does show the fitted SEMs and data summary, so I have no idea what the issue actually is.
If you need a version of the data to work on, I can provide it privately somehow. Thank you in advance!
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fjslefche%2FpiecewiseSEM%2Fissues%2F214&data=02%7C01%7Clefcheckj%40si.edu%7C03d9f0170c4b4183534408d85a55e7d0%7C989b5e2a14e44efe93b78cdd5fc5d11c%7C0%7C0%7C637358672300126989&sdata=QQ7hFV5c%2F4n5TlJhRBrltLJfqgAjWR2QTOt6fSVCBIA%3D&reserved=0, or unsubscribehttps://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAAR4AV6CR3SUJMRL6MTBNUTSGDLFXANCNFSM4RPAPMFQ&data=02%7C01%7Clefcheckj%40si.edu%7C03d9f0170c4b4183534408d85a55e7d0%7C989b5e2a14e44efe93b78cdd5fc5d11c%7C0%7C0%7C637358672300136986&sdata=1ATwnZPAfNnMxCxIfUMGPpKVZaGLxxZDkljjUhEC22w%3D&reserved=0.
leorjorge
Really excited to find out about this package. Being able to fit a SEM with mixed effects is quite amazing.
However, I'm getting some trouble fitting (an admittedly quite complex) model, and I'm lost on how to figure what could be the issue. I managed to get the
psemfunction to work, running the code below:(Don't mind the commented out glmer... that was my first try at fitting the effects on the main response variable, but I think the fact it is a set of 24 binary values for each observation messes up even more with the fit and the model doesn't work at all)
When trying to run
summaryorcoefsI get the following (different) errors. I couldn't find info on them (and they are not very clear):However, just printing the object does show the fitted SEMs and data summary, so I have no idea what the issue actually is.
If you need a version of the data to work on, I can provide it privately somehow. Thank you in advance!