Description Usage Arguments Details Value Author(s) References See Also Examples
strat
estimates statistical strategic models using maximum likelihood
estimation.
1 |
formulae |
Nested List. Formulas for utilities and private information variances for each outcome and player player. This list has a specific structure. Each element of the top-level list is an outcome; these do not need to be named as the name of the outcome variable can be determined from the formulas. Each outcome is a list with elements named for each players; these elements must be named. Each player is a list with two elements. The first is the formula describing the utility function of the player for that outcome; the second is the variance of the private information error term for that (player, outcome). The private information element is optional, and is assumed to be 0 (meaning, no private information error term) if not present. If an outcome, player element is missing, the utility of that player for that outcome is assumed to be 0. |
tree |
Character. Name of model to be estimated. |
tNodes |
Character. Name of dependent variable associated with each terminal node. |
alpha |
Named list. Variance of agent error for each action. Each element must have the same name as its action. |
data |
An optional data frame, list or environment (or object
coercible by |
subset |
an optional vector specifying a subset of observations to be used in the fitting process. |
weights |
an optional vector of weights to be used in the fitting process. |
na.action |
Character or function. A function which indicates
what should happen when the data contain |
model |
Logical. If |
start |
Numeric. Initial values for optimizer. |
method |
Character. Optimization method to use. Default is
"BFGS". See |
... |
Further arguments to pass to |
strat
is a general function to estimate statistical strategic
models with binary choices. Each extensive game form must be implemented by adding a class
that extends "StratModel"
and adding a
prAction
method to calculate the probabilities of each
action for that particular game form. Future plans are to
allow for arbitrary extensive form games. For a simpler front-end
to strat
, written for the 1-2 extensive form game, see strat_1_2
.
The details of the estimation procedure is taken primarily from Signorino (2003). See the other references for more details on statistical strategic models.
The optim
optimizer is used to find the minimum of the
negative log-likelihood. Since optim
is called by the
mle
function, strat
returns an S4 object that
extends "mle"
.
Currently strat
has the following restrictions
only binary choices are allowed.
errors are distributed normally.
regressor error is not supported.
An object of class StratMLE
.
Jeffrey Arnold, jarnold7 AT mail DOT rochester DOT edu
~put references to the literature/web site here ~
See Also as strat_1_2
, mle
, optim
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 | data(war1800)
## Estimate a Strategic 1-2 Model
## Define the utility functions for each player at each outcome
utils <- list(
list("1" = list(sq ~ - 1 + peaceyrs + s.wt.re1, 1)),
list("1" = list(capit ~ 1, 1),
"2" = list(capit ~ 0, 1)),
list("1" = list(war ~ -1 + balanc, 1),
"2" = list(war ~ balanc, 1))
)
## Define the structure of the Tree
nodes <- list(a0 = StratActionNode("1", 0),
a1 = StratTerminalNode("sq", 0),
a2 = StratActionNode("1", 0),
a3 = StratTerminalNode("capit", 0),
a4 = StratTerminalNode("war", 0))
gameTree <- StratTree(nodes,
c(NA, "a0", "a0", "a1", "a2", "a2"))
quux <- strat(utils, tree="1-2", data=war1800,
tNodes = c("sq", "capit", "war"),
model=TRUE)
## Summary of the model
summary(quux)
## Other methods inherited from 'mle'
print(quux)
vcov(quux)
ll <- logLik(quux)
AIC(ll)
BIC(ll)
## Note confint() and profile() are very slow for any non-trivial
## strategic model
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.