Filter only duplicated rows based on condition

In the data below, I want to filter the duplicated participants (Ana, Paul) based on their answer to Q40. Participants should have answered only "NoPlan" or "Plan", but sometimes they've answered both. I want to keep only the rows in each the duplicated participants have answered "Plan"

• The data is:

  ID     Q40   
  <chr>  <chr> 
1 Maria  Plan  
2 Joseph Plan  
3 Cyntia NoPlan
4 Ana    NoPlan   ## first noPlan
5 Paul   Plan  ## first Plan
6 Ana    Plan  ## then, Plan 
7 Paul   NoPlan ## then, noPlan 

• Desired Output:

  ID     Q40   
  <chr>  <chr> 
1 Maria  Plan  
2 Joseph Plan  
3 Cyntia NoPlan
5 Paul   Plan  
6 Ana    Plan  

• I was doing it manually, like this:

df %>% 
  filter(Q40 %in% c('noPlan', 'Plan')) %>% 
  ungroup() %>% 
  distinct()

then I was removing manually the duplicated ones with (slice()) and their corresponding position.

QUESTION

1: How can I dinamically remove the rows only of the duplicated participants in ID (who answered "noPlan" and "plan" at the same time), keeping only their "Plan" row? I'm sure this is easy to do with tidyverse , but I'm struggling with it. I really need a tidyverse approach.

Important note:

1 I don't want to filter all "NoPlan" rows, I only want to filter the "Plan" rows of the participants who have answered both NoPlan AND Plan. Only the duplicated ones, not the whole sample.

2 I don't want to keep only the first row such as with distinc() , I want to be able to set the condition myself (in this case, to filter "NoPlan" from the duplicated participants

3 I cannot use distinct(starwars, pick(contains("color"))) because this would remove "NoPlan" from all participants and I need to filter only the duplicated ones

• data:

structure(list(ID = c("Joseph", "Cyntia", "Paul", "Ana", "Paul", 
"Maria", "Ana"), Q40 = c("Plan", "NoPlan", "Plan", "Plan", "NoPlan", 
"Plan", "NoPlan")), row.names = c(NA, -7L), class = c("tbl_df", 
"tbl", "data.frame"))

?duplicated should aid you here:

dat %>%
    filter(!((duplicated(ID) | duplicated(ID, fromLast=TRUE)) & Q40 == "NoPlan"))

Or in base R speak:

dat[with(dat, !((duplicated(ID) | duplicated(ID, fromLast=TRUE)) & Q40 == "NoPlan")),]

It sounds like you'd prefer a tidyverse answer, but here's a base R solution using the split-operate-recombine approach. You split by ID; in subsets with multiple rows, select the rows with "Plan" in Q40; then use do.call("rbind", ...) to put the remaining rows back together.

dat <- structure(list(ID = c("Joseph", "Cyntia", "Paul", "Ana", "Paul", 
"Maria", "Ana"), Q40 = c("Plan", "NoPlan", "Plan", "Plan", "NoPlan", 
"Plan", "NoPlan")), row.names = c(NA, -7L), class = c("tbl_df", 
"tbl", "data.frame"))

do.call("rbind", lapply(split(dat, dat$ID), function(x) {

  if(nrow(x) > 1) {

    x[x$Q40 == "Plan",]

  } else {

    x

  }

}))

And here's what that produces:

  ID     Q40   
* <chr>  <chr> 
1 Ana    Plan  
2 Cyntia NoPlan
3 Joseph Plan  
4 Maria  Plan  
5 Paul   Plan