Functions to transform MPT models.
Description
(Helper) functions that takes an MPT model file and transforms it into a representation in the contextfree language of MPT models LBMPT (Purdy & Batchelder, 2009) or takes the representation in LBMPT and returns the model equations.
Usage
1 2 3 4  make.mpt.cf(model.filename, restrictions.filename = NULL,
model.type = c("easy", "eqn"), treewise = FALSE)
lbmpt.to.mpt(model.list, outfile = NULL, category.names= TRUE)

Arguments
model.filename 
A character 
restrictions.filename 

model.type 
Character vector specifying whether the model file is formatted in the 
treewise 
logical. Should the model be concatenated to one tree before transforming to LBMPT? Default is 
model.list 
A list of character 
outfile 
Name of the file the model equation should be saved in (in easy format). If 
category.names 
logical. Should category names (e.g., "category 1") be printed at the end of each line? 
Details
Purdy and Batchelder (2009) provide a new way of how binary multinomial processing tree (MPT) models can be represented, a context free language called LBMPT. This function takes a model file that consists of the equations defining a model (e.g., http://www.psychologie.unifreiburg.de/Members/singmann/R/mptinr/modelfile) and returns a character vector representing this model in LBMPT.
There are three important things to know about this function:
1. LBMPT distinguishes between observable categories (C) and parameters (theta). As MPTinR allows parameters to have any name that is legal for a variable in R (with the only restriction that parameters should NOT start with hank
) the LBMPT representation of the model's parameters can also consist of any name that is a legal variable name in R. To distinguish parameters from categories, categories are represented as integers (i.e., numbers) (a number is not a legal variable name in R). Furthermore, as legal variable names may end with a number, concatenating parameters and categories into one string could lead to an ambiguous representation of the model. Therefore, the returned representation in LBMPT is a character vector with each element representing either a parameter (any legal variable name in R) or a category (an integer).
Theta: Names that are legal variable names in R.
C: Integers.
2. If a model consists of more than n > 1 trees, this function per default concatenates the trees into a single binary MPT model by adding n1 parameters (named hank.join.x
with x
be a integer starting at 1; see Wu, Myung & Batchelder, 2010). This can be turned off, by setting treewise
to TRUE.
3. It is absolutely necessary that the representation of the model via equations in the model file exactly maps on the structure of the binary tree. In other words, equations in the model file can NOT be simplified in any way. The equations in the model file may only consist of the parameters and their negations (e.g. d
and (1d)
). Simplifications and aggregations in the model file (e.g., from u * (1u) + u * (1u)
to 2*u*(1u
) will lead to erroneous results! Similarly, reparameterizations for inequality constraints (which can be done by fit.mpt
) can not be represented in LBMPT.
Value
make.mpt.cf
: A character vector with each element representing either a parameter or a category (categories are represented by integer numbers). In case of multiple trees and treewise = TRUE
, a list of such vectors.
lbmpt.to.mpt
: Either prints the model to the screen or returns nothing and saves the model equations in the specified file.
Note
It is absolutely necessary that the model file exactly maps on the structure of the binary tree. See Details.
Author(s)
Henrik Singmann (make.mpt.cf)
Quentin Gronau and Franz Dietrich (lbmpt.to.mpt, using a function from Akhil S Bhel, LinearizeNestedList)
References
Purdy, B. P., & Batchelder, W. H. (2009). A contextfree language for binary multinomial processing tree models. Journal of Mathematical Psychology, 53, 547561.
Riefer, D. M., & Batchelder, W. H. (1988). Multinomial modeling and the measurement of cognitive processes. Psychological Review, 95, 318339.
Wu, H., Myung, J., I., & Batchelder, William, H. (2010). Minimum description length model selection of multinomial processing tree models. Psychonomic Bulletin & Review, 17, 275286.
See Also
get.mpt.fia
and link{bmpt.fia}
for functions calling make.mpt.cf
to obtain the FIA of a MPT model.
See fit.mpt
for the main function of MPTinR which also calls make.mpt.cf
for obtaining the FIA.
prepare.mpt.fia
will provide the code needed for obtaining the Minimum Description Length of a MPT model using Matlab (Wu, Myung & Batchelder, 2010) and calls make.mpt.cf
for obtaining the LBMPT representation.
Examples
1 2 3 4 5 6 7  model2 < system.file("extdata", "rb.fig2.model", package = "MPTinR")
make.mpt.cf(model2)
make.mpt.cf(model2, treewise = TRUE)
lbmpt.to.mpt(make.mpt.cf(model2, treewise = TRUE))
