estObserved: Estimate observed data density
In DstarM: Analyze Two Choice Reaction Time Data with the D*M Method

Description Usage Arguments Value Examples

Estimates the density of the observed data by convoluting the estimated decision distributions with the estimated nondecision distributions. If a traditional analysis was run the argument resND can be omitted.

estObserved(
  resDecision,
  resND = NULL,
  data = NULL,
  interpolateND = FALSE,
  tt = NULL
)

`resDecision`	output of `estDstarM`.
`resND`	output of `estND`.
`data`	Optional. If the data used to estimate the decision model is supplied additional fitmeasures are calculated.
`interpolateND`	Logical. If the decision model and nondecision model have been estimated on different time grids, should the rougher time grid be interpolated to match the smaller grid? If FALSE (the default) the decision model will be recalculated on the grid of the nondecision model. This tends to produce better fit values.
`tt`	Optional time grid to recalculate the model densities on. Unused in case of a DstarM analysis.

Returns a list of class DstarM.fitObs that contains:

`obsNorm`	A matrix containing normalized densities of each condition response pair.
`obs`	A matrix containing unnormalized densities of each condition response pair.
`tt`	The time grid used.
`fit`	A list containing the values of the objective function for the total model ($total), for the decision model ($Decision) and for the nondecision distribution(s) ($ND).
`npar`	The number of parameters used in the decision model.
`obsIdx`	A numeric vector containing indices of any not observed condition-response pairs.

# simulate data with three stimuli of different difficulty.
# this implies different drift rates across conditions.
# define a time grid. A more reasonable stepsize is .01; this is just for speed.
tt = seq(0, 5, .1)
pars = c(.8, 2, .5, .5, .5, # condition 1
        .8, 3, .5, .5, .5, # condition 2
        .8, 4, .5, .5, .5) # condition 3
pdfND = dbeta(tt, 10, 30)
# simulate data
lst = simData(n = 3e5, pars = pars, tt = tt, pdfND = pdfND, return.pdf = TRUE)
dat = lst$dat
# define restriction matrix
restr = matrix(1:5, 5, 3)
restr[2, 2:3] = 6:7 # allow drift rates to differ
# fix variance parameters
fixed = matrix(c('sz1', .5, 'sv1', .5), 2, 2)
## Not run: 
# Run D*M analysis
resD = estDstarM(dat = dat, tt = tt, restr = restr, fixed = fixed)
# Estimate nondecision density
resND = estND(resD)
# Estimate observed density
resObs = estObserved(resD, resND)
# plot histograms with overlayed
# densities per condition-response pair
plotObserved(resObserved = resObs, data = dat,
            xlim = c(0, 1))
# plot estimated and true densities
plot(resObs, col = rep(1:3, each = 2), xlim = 0:1)
matlines(tt, lst$pdfNormalized, col = rep(1:3, each = 2), lty = 2)

## End(Not run)