sim_practice: Simulate data with a practice effect

In dalejbarr/autocorr: Autocorrelation toolkit for multilevel data

Description

Simulates psychological response time data containing a practice effect, such that the response time decays exponentially at a rate determined by learning_rate until reaching asymptote. Units are milliseconds.

Usage

sim_practice(
  n_subj,
  n_trials,
  within_eff = 10,
  between_eff = 24,
  asymptote = 400,
  sweep = 400,
  learning_rate = -2,
  rslope_sd = 20,
  rasym_sd = 40,
  rsweep_sd = 40,
  rlrate_sd = 0,
  err_sd = 20,
  verbose = FALSE
)

Arguments

n_subj	Number of subjects.
n_trials	Number of trials per subject.
within_eff	Mean effect of within-subject factor, coded by xi.
between_eff	Mean effect of between-subject factor, coded by xj.
asymptote	Asymptotic value.
sweep	The 'sweep' of the decline; i.e., the distance from the asymptote to the starting value.
learning_rate	Speed of the decay.
rslope_sd	By-subject standard deviation for the random slope.
rasym_sd	By-subject standard deviation for the random asymptote.
rsweep_sd	By-subject standard deviation for the random sweep.
rlrate_sd	By-subject standard deviation for the learning rate.
err_sd	Error standard deviation.
verbose	Whether to return random effects in the data frame.

Details

The data contains main effects of the within-subject and between-subject factors, but no interaction. The model also contains four by-subject random effects (within-subject slope, asymptote, sweep, and learning rate). These random effects are all independent from one another. The data-generating model for response y for subject i and observation j is:

yij ~ (g00 + S0i) * wij + g10 + S1i + g11 * bi + (g20 + S2i) * exp(-exp(g30 + S3i) * tij) + eij

where the predictors are:

g00

mean within-subject effect;

S0i

by-subject random slope;

wij

deviation-coded predictor for the within-subject effect;

g10

mean asymptote;

S1i

random asymptote for subject i;

g11

mean between-subject effect;

bi

deviation-coded predictor for the between-subject effect;

g20

mean sweep (distance from starting value to asymptote);

S2i

random sweep for subject i;

g30

mean learning rate;

S3i

random learning rate for subject i;

tij

trial number for subject i, observation j;

eij

random error for subject i, observation j.

Value

A data frame with some or all of the following fields, depending on the value of verbose:

subj_id

Unique subject identifier.

tij

Trial number.

wij

Deviation-coded within-subject predictor.

bi

Deviation-coded between-subject predictor.

y

The response value.

S0i

By-subject random slope.

S1i

By-subject random asymptote.

S2i

By-subject random sweep.

S3i

By-subject random learning rate.

trueval

Response value excluding the within-subject effect/random slope.

dalejbarr/autocorr documentation built on March 27, 2021, 3:03 a.m.