**200 pairs of treatment**(0 and 1 for placebo and the real treatment, respectively)

**and disease**(0 and 1 for the absence or presence of the disease).

This way, I have duplicated (copied & pasted) my 100 rows of patients' demographics into 200 rows and have distinguished these two 100-row bundles of data by the variable "treatment" (100 rows with treatment = 0, 100 copied rows with treatment = 1).

Now I need to run a mixed model regression to account for the multilevel design and avoid atomistic fallacy (thanks to Greta).

I have written the following R syntax, but there are some problems: first that the P values are really disappointing (all above 0.8, although they might be correct). Second that I don't know if the code is correct in the first place or not!!

My code is:

Code:

```
(mixed <- lmer(Disease ~ Gender + Age + Demographic3 + Demographic4 + TREATMENT
+ Age*Demographic4 + Age*TREATMENT + Age*Gender + Age*Demographic3 + Demographic4*TREATMENT + Demographic4*Gender
+ Demographic4*Demographic3 + TREATMENT*Gender + TREATMENT*Demographic3 + Gender*Demographic3
+ (Age | PatientID)
+ (Gender | PatientID)
+ (Demographic3 | PatientID)
+ (Demographic4 | PatientID)
, family=binomial(logit), MixedModelData))
```

Code:

`In mer_finalize(ans) : false convergence (8).`

Many thanks.