rtdists (version 0.11-5)

speed_acc: Speed-Accuracy Data from Wagenmakers, Ratcliff, Gomez, & McKoon (2008, Experiment 1)

Description

Responses and response times from an experiment in which instruction manipulated speed and accuracy between blocks. This data was also analyzed by Heathcote and Love (2012) who were the first to use the 17 participants also included here.

Usage

speed_acc

Format

A data.frame with 31,522 obs. and 9 variables:

id

participant id

block

block number

condition

accuracy for blocks with accuracy instructions; speed for blocks with speed instruction

stim

unique identifier of stimulus, stimuli are nested in frequency conditions

stim_cat

category of stimulus, either word or non-word

frequency

"high frequency word", "low frequency word", "very low frequency word", or non-words derived from the first three categories

response

word, nonword, or not interpretable response (error, i.e., pushed a button, but not the right one and also not the one next to the right button)

rt

response time in seconds

censor

boolean indicating whether or not a response should be eliminated prior to analysis; uninterpretable response, too fast response (<180 ms), too slow response (>3 sec)

Details

The data excludes the practice blocks but includes all trials. Variable censor can be used for excluding all trials also excluded from the papers using it namely uninterpretable response, too fast response (<180 ms), too slow response (>3 sec). Heathcote and Love (2012, p. 7) describe the data as follows:

We fit the LBA and LNR models to data from Wagenmaker et al.'s (2008) experiment one, where participants made decisions about whether a string of letters constituted a word. These lexical decisions were made about four types of stimuli, non-words (nw) and high-frequency (hf), low-frequency (lf), and very low-frequency (vlf) words. Participants made decisions either under speed or accuracy emphasis instructions in different experimental blocks. Accuracy blocks were preceded by the message "Try to respond accurately" and "ERROR" was displayed after each wrong response. Speed blocks were preceded by the message "Try to respond accurately" and "TOO SLOW" was displayed after each response slower than 0.75 s.We report analyses of data from 17 participants (31,412 data points) in their Experiment 1, including the 15 participants analyzed in Wagenmakers et al. (2008) and two extras (we thank Eric-Jan Wagenmakers for supplying this data).

References

Heathcote, A., & Love, J. (2012). Linear deterministic accumulator models of simple choice. Frontiers in Psychology: Cognitive Science, 3, 292. doi:10.3389/fpsyg.2012.00292

Examples

# NOT RUN {
data(speed_acc)
str(speed_acc)

# remove excluded trials:
speed_acc <- droplevels(speed_acc[!speed_acc$censor,])

# new factors for obtaining values as in Table 1, Wagenmakers et al. (2008, p. 152)
speed_acc$freq <- with(speed_acc, 
                       factor(ifelse(stim_cat == "nonword", "nonword", 
                                     as.character(frequency)), 
                              levels = c("high", "low", "very_low", "nonword")))
# corr = correct (0 = correct, 1 = error)
speed_acc$corr <- with(speed_acc, 1-as.numeric(stim_cat == response))

str(speed_acc)

## aggregated RTs:
aggregate(rt ~ condition + freq + corr, speed_acc, mean)
## Error Rate:
aggregate(corr ~ condition + freq + corr, speed_acc, mean)

# }