How to get mean for all participants after selecting only a certain number of trials

Question

I have a dataset of 500 trials per participant that I want to sample from in various quantities (i.e. I want to sample the same number of trials from each participant) and then compute the mean for each participant. Instead of doing so, it is creating a file with a one mean for each participant separately for each "num", e.g. if the mean for participant 1 with 125 trials is 426 that will be the whole file, then another file for participant 1 with 150 trials with a single value, and that is what happens for all participants. I was aiming for a single file for 125 with the means for all participants, then another file with the means for 150, etc.

num <- c(125,150,175,200,225,250,275,300,325,350,375,400)

Subset2 <- list()


for (x in 1:12){
  for (j in num){
   Subset2[[x]] <- improb2 %>% group_by(Participant) %>% sample_n(j) %>% summarise(mean = mean(RT))
  
  
}}

Here is a reproducible example:

RT <- sample(200:600, 10000, replace=T)
df <- data.frame(Participant= letters[1:20]) 
df <- as.data.frame(df[rep(seq_len(nrow(df)), each = 500),])

improb2 <- cbind(RT, df)
improb2 <- improb2 %>% rename(Participant = `df[rep(seq_len(nrow(df)), each = 500), ]`)

One of the desired dataframes in subset2 would be something like:

Subset2[[1]]

Participant  mean
   <chr>       <dbl>
 1 P001         475.
 2 P002         403.
 3 P003         481.
 4 P004         393.
 5 P005         376.
 6 P006         402.
 7 P007         497.
 8 P008         372.
 9 P010         341.

Answer 1

This answer uses tidyverse and outputs a list object data where the names are the sample sizes. To access each sample size summary you have to use backticks data$`125` . data$`125` is a tibble object. I made a comment in the output where you can change it to a data.frame object if you need.

library(tidyverse)

num <- c(125, 150, 175, 200, 225, 250, 275, 300, 325, 350, 375, 400)

# create function to sample data by certain size and summarize by mean
get_mean <- function(x, n) { 
  dplyr::group_by(x, Participant) %>% # group by participant
    dplyr::sample_n(n) %>% # randomly sample observations
    dplyr::summarize(mean = mean(RT), # get mean of RT
                     n = n(), # get sample size
                     .groups = "keep") %>% 
    dplyr::ungroup()
# add a pipe to as.data.frame if you don't want a tibble object
}

# create a list object where the names are the sample sizes
data <- lapply(setNames(num, num), function(sample_size) {get_mean(df, n = sample_size)})

head(data$`125`)

 Participant  mean     n
  <chr>       <dbl> <int>
1 V1           20.2   125
2 V10          19.9   125
3 V11          19.8   125
4 V12          20.2   125
5 V2           20.5   125
6 V3           20.0   125

---

Data

I wasn't 100% sure what your dataset looked like, but I believe it looks something like this:

# create fake data for 45 participants with 500 obs per participant
df <- replicate(45, rnorm(500, 20, 4)) %>%
  as.data.frame.matrix() %>% 
  tidyr::pivot_longer(everything(), 
                      names_to = "Participant", # id column
                      values_to = "RT") %>% # value column
  dplyr::arrange(Participant)


head(df) # Participant repeated 500 times, with 500 values in RT
 Participant    RT
  <chr>       <dbl>
1 V1           24.7
2 V1           15.2
3 V1           21.1
4 V1           21.6
5 V1           20.3
6 V1           25.6

If this is a similar structure (long with repeated participant IDs and a single column RT of values) then the above should work.