Compare text length across conditions

Compare word and character counts across conditions. I'd inspect a condition that consistently produces longer texts: respondents may be reacting to the extra information as well as the intended manipulation.

Usage

check_length_balance(texts, conditions, threshold = 0.2)

Arguments

texts

Character vector of materials. NA entries are excluded from the statistics and reported separately.

conditions

Condition labels, same length as texts.

threshold

Proportional deviation from the grand mean that triggers a warning (default: 0.20).

Value

A list with by_condition, grand_mean_words, max_deviation, n_missing, n_no_condition, and pass.

Examples

check_length_balance(
  c("Short text here.", "A somewhat longer text appears here now.", "Brief."),
  c("treatment", "treatment", "control")
)
#> ! Length balance: max deviation 72.7% (> 20%)
#> $by_condition
#> # A tibble: 2 × 6
#>   condition     n mean_words sd_words mean_chars deviation
#>   <chr>     <int>      <dbl>    <dbl>      <dbl>     <dbl>
#> 1 control       1          1    NA             6     0.727
#> 2 treatment     2          5     2.83         28     0.364
#> 
#> $grand_mean_words
#> [1] 3.666667
#> 
#> $max_deviation
#> [1] 0.7272727
#> 
#> $n_missing
#> [1] 0
#> 
#> $n_no_condition
#> [1] 0
#> 
#> $pass
#> [1] FALSE
#>