LOOP in R: Error: variable lengths differ-CodePudding

I tried to build this Loop so that I can test two outcomes at the same time. However, it produced an error message: "Error in model.frame.default(formula = ~outcome centered.predictor1 : variable lengths differ (found for 'centered.predictor1')"

But when I tested each outcome separately, the code (without loop) didn't produce errors.

Thanks in advance for your help!

n1 = rnorm(n = 2000, mean = 0, sd = 1)
n2 = rnorm(n = 2000, mean = 0, sd = 1)
Z_familism = rnorm(n = 2000, mean = 0, sd = 1)
Z_avoidance = rnorm(n = 2000, mean = 0, sd = 1)
Country = rnorm(n = 2000, mean = 0, sd = 1)
Z_anxiety = rnorm(n = 2000, mean = 0, sd = 1)
data01<-data.frame(n1,n2,Z_familism,Z_avoidance,Country,Z_anxiety)


outcome<-c('n1', 'n2')

for (n in outcome){
  
  rsa.data<-data.frame(predictor1=data01$Z_familism,
                       predictor2=data01$Z_avoidance,
                       nest=as.factor(data01$Country),
                       control=data01$Z_anxiety,
                       multilevel=data01$Country,
                       outcome=data01[n]) 
 
  
  rsa.data <- within.data.frame(rsa.data, {
    centered.predictor1 <- predictor1 - 0    #Center predictor 1
    centered.predictor2 <- predictor2 - 0  #Center predictor 2
    squared.predictor1  <- centered.predictor1* centered.predictor1 #Create squared term
    squared.predictor2  <- centered.predictor2* centered.predictor2 #Create squared term
    interaction           <- centered.predictor1* centered.predictor2 #Create interaction term
   
  })
  
  mlm.model <- lme(outcome ~ centered.predictor1 centered.predictor2    squared.predictor1   interaction  squared.predictor2 control,
                   data = rsa.data,
                   random = ~ 1|multilevel, # Replace "nesting.variable" with the name of your nesting variable 
                   na.action = "na.omit")
  summary(mlm.model) #View Model
  intervals(mlm.model, which = "fixed")
  vcov(mlm.model) #View covariance of model
  
}

CodePudding user response：

The problem is when you create the rsa.data dataframe inside the loop, specifically with the outcome column. Instead of data01[n] which returns a dataframe, you should use data01[, n], which returns a numeric vector. That way all your data has the same length.

rsa.data<-data.frame(predictor1=data01$Z_familism,
                     predictor2=data01$Z_avoidance,
                     nest=as.factor(data01$Country),
                     control=data01$Z_anxiety,
                     multilevel=data01$Country,
                     outcome=data01[, n])