Adding a number thats constant within participants but differs between participants (lv. 2 variable) [loop r, MLM]

Question

I want to calculate a moderation analysis in a multilevel model. I have 20 taskblocks (lvl 1) for 33 participants (lvl 2). For my relevant IV and DV I have already written a loop that works:

for(i in 1:length(participants)) {
  participant.sub <- df[df$participant == participants[i],]
  participant.sub <- participant.sub[!is.na(participant.sub$participant),]
  for (j in 1:20) {
    block.sub <- participant.sub[participant.sub$block == j,]
    data.row <- c(unique(block.sub$participant),
                  j,
                  block.sub$slider_mentaleffort.response[!is.na(block.sub$slider_mentaleffort.response)],
                  block.sub$goal[!is.na(block.sub$goal)],
                  block.sub$sum_correct[!is.na(block.sub$sum_correct)])
    data.row <- data.row[2:6]
    dummy.df <- rbind(dummy.df, data.row) 
  }
}

Now I want to include the moderators. The moderator variables are likert-scales that got answered before the experiment started. Each person has exactly one value on each scale. How do I add those values to the current loop/data frame, so that I can continue with lmer?

I tried this

for(i in 1:length(participants)) {
  participant.sub <- df[df$participant == participants[i],]
  participant.sub <- participant.sub[!is.na(participant.sub$participant),]
  for (j in 1:20) {
    block.sub <- participant.sub[participant.sub$block == j,]
    data.row <- c(unique(block.sub$participant),
                  j,
                  block.sub$slider_mentaleffort.response[!is.na(block.sub$slider_mentaleffort.response)],
                  block.sub$goal[!is.na(block.sub$goal)],
                  block.sub$sum_correct[!is.na(block.sub$sum_correct)],
                  block.sub$slider_effort.response[!is.na(block.sub$slider_effort.response)],
                  block.sub$slider_talent.response[!is.na(block.sub$slider_talent.response)],
                  block.sub$slider_luck.response[!is.na(block.sub$slider_luck.response)],
                  block.sub$slider_taskdiff.response[!is.na(block.sub$slider_taskdiff.response)])
    data.row <- data.row[2:9] 
    dummy.df <- rbind(dummy.df, data.row) 
  }
}

and

for(i in 1:length(participants)) {
  participant.sub <- df[df$participant == participants[i],]
  participant.sub <- participant.sub[!is.na(participant.sub$participant),]
  for (j in 1:20) {
    block.sub <- participant.sub[participant.sub$block == j,]
    data.row <- c(unique(block.sub$participant),
                  j,
                  block.sub$slider_mentaleffort.response[!is.na(block.sub$slider_mentaleffort.response)],
                  block.sub$goal[!is.na(block.sub$goal)],
                  block.sub$sum_correct[!is.na(block.sub$sum_correct)],
                  block.sub$slider_effort.response,
                  block.sub$slider_talent.response,
                  block.sub$slider_luck.response,
                  block.sub$slider_taskdiff.response)
    data.row <- data.row[2:9] 
    dummy.df <- rbind(dummy.df, data.row) 
  }
}

But to no avail.

Answer 1

It is better to use dpylr and slice the data in the rows I wanted, before merging them into one frame, that leaves out the na's

df_attitude <- df %>%
  filter(!is.na(slider_talent.response)) %>%
  select(participant, slider_talent.response, slider_luck.response, slider_effort.response, slider_taskdiff.response)
data <- data %>%
  merge(., df_attitude, by = "participant")

df_effort <- df %>%
  filter(!is.na(slider_mentaleffort.response)) %>%
  select(participant, block, slider_mentaleffort.response, sum_correct)
df_goal <- df %>%
  filter(!is.na(goal)) %>%
  select(participant, block, goal)

It is the more elegant, flexible and shorter way than looping.