Error in lav_lavaan_step11_estoptim(lavdata = lavdata, lavmodel = lavmodel, : lavdata@nlevels > 1L is not TRUE

[zillur rahman]

Hi,

I was trying to use sem as below:

mediation_model3 <- '
  #direct effects
  MVPA ~ a1 * BIP_PRS + Sex + Age + PC1 + PC2 + PC3 + PC4 + PC5 + PC6 + PC7 + PC8 + PC9 + PC10
  #LST ~ a2 * BIP_PRS + Sex + Age + PC1 + PC2 + PC3 + PC4 + PC5 + PC6 + PC7 + PC8 + PC9 + PC10
  Healthy_Food ~ a3 * BIP_PRS + Sex + Age + PC1 + PC2 + PC3 + PC4 + PC5 + PC6 + PC7 + PC8 + PC9 + PC10
  Meat ~ a4 * BIP_PRS + Sex + Age + PC1 + PC2 + PC3 + PC4 + PC5 + PC6 + PC7 + PC8 + PC9 + PC10
  BMI ~ c * BIP_PRS + b1 * MVPA + b3 * Healthy_Food + b4 * Meat
 
  #indirect effect (a * b)
  indirect1 := a1  * b1
  #indirect2 := a2 * b2
  indirect3 := a3 * b3
  indirect4 := a4 * b4
  #total effect (c + indirect)
  total := c + indirect1 + indirect3 + indirect4
 
  #MVPA ~~ LST
  MVPA ~~ Healthy_Food
  MVPA ~~ Meat
  #LST ~~ Healthy_Food
  #LST ~~ Meat
  Healthy_Food ~~ Meat
'
setwd('Desktop/zillur/work/body-brain/physical_activity/sem/')
df1=fread('processed_data1.csv',data.table = F)
df2 = df1[,-1]

mediation_results4 <- sem(mediation_model3, data = df2, cluster = "Meat",
                          verbose = TRUE, optim.method = "em", em.iter.max = 9000000,
                          em.fx.tol = 1e-08, em.dx.tol = 1e-04)
summary(mediation_results4, standardized = TRUE, fit.measures = TRUE)

Gives me the above mentioned error:

> mediation_results4 <- sem(mediation_model3, data = df2, cluster = c("MVPA","Healthy_Food","Meat"),
+                           verbose = TRUE, optim.method = "em", em.iter.max = 900000,
+                           em.fx.tol = 1e-08, em.dx.tol = 1e-04)
lavoptions         ... done.
lavdata            ... done.
  Number of observations                        270288
  Number of clusters [MVPA]                          2
  Number of clusters [Healthy_Food]             270288
  Number of clusters [Meat]                          2
lavpartable        ... done.
lavsamplestats     ... done.
lavh1              ... start:
lavh1              ... done.
lavpartable bounds ... done.
lavstart           ... done.
lavmodel           ... done.
lavoptim           ... start:
Error in lav_lavaan_step11_estoptim(lavdata = lavdata, lavmodel = lavmodel,  :
  lavdata@nlevels > 1L is not TRUE

[Terrence Jorgensen]

The lavaan model syntax you provided does not specify a multilevel SEM.

https://lavaan.ugent.be/tutorial/multilevel.html

If you only want cluster-robust SEs and tests, then you shouldn't be specifying the EM-algorithm options.

Terrence D. Jorgensen    (he, him, his)
Assistant Professor, Methods and Statistics
Research Institute for Child Development and Education, the University of Amsterdam
http://www.uva.nl/profile/t.d.jorgensen