Simulate measurement

Simulate measurement of a variable by adding random error. The function allows adding a subject specific error (sub_bias_sd), that varies between subjects.

Usage

simulate_measurement(
  true_val,
  sub_id,
  residual_error_sd,
  mean_bias = 0,
  sub_bias_sd = 0,
  proportional_errors = FALSE
)

Arguments

true_val

A numeric vector of true values.

sub_id

A factor or vector representing subject IDs. Must be the same length as var_true.

residual_error_sd

Standard deviation of residual errors.

mean_bias

Mean bias.

sub_bias_sd

Standard deviation of subject-specific biases.

proportional_errors

Should errors be proportional to true value. If TRUE, errors are added on a log scale, and the result is exponentiated. To add a bias of 10% (true_value * 1.1), the mean_bias should be set to log(1.1). A SD of log(1.2) corresponds to ±20%.

Value

A numeric vector representing simulated measurements.

Examples

df <- simulate_repeated_data(n_sub = 50, n_rep = 5, var_name = "true_co")
df$ref <- simulate_measurement(
  df$true_co,
  df$id,
  mean_bias = log(1.1),
  sub_bias_sd = log(1.2),
  residual_error_sd = log(1.1),
  proportional_errors = TRUE
)
plot(df$true_co, df$ref, col=df$id)
abline(a=0, b=1)