ImitSim: Simulation of Imitation by Bureaucrrats

In jjbrehm/BadApple: Functions to simulate bureaucracies

Description

ImitSim() runs an R version of the imitation simulation as published in Working, Shirking, and Sabotage (1997). The original function was written in APL11, which hardly anyone uses. This is the core function for the simulation in the book.

Usage

ImitSim(Replications, NumBurs=10, MaxIter=10, supervision="Relative", posprefs=FALSE,
omniscient=FALSE, Tolerance=-99, Std=-99, Punishment=-99, SupObsParms=c(-99,-99),
PrefParms=c(-99,-99), ResponseParms=c(-99,-99), BurObsParms=c(-99,-99),
quiet=FALSE, debug=FALSE

Arguments

Replications	Number (integer) of replications of each run.
NumBurs	Number (integer) of Bureaucrats.
MaxIter	Number (integer) of iterations for each replication to run.
supervision	"Relative" (default) for supervisor to identify deviants based on low-end deviations from responses by other bureaucrats, "Fixed" for the supervisor to identify deviants based on a floor of minimal performance.
posprefs	Logical, FALSE (default) if the preferences of the bureaucrats are not constrained to be positive. TRUE if the preferences of the bureacrats must be positive.
omniscient	Logical, FALSE (default) if the supervisor cannot see everyone, TRUE if she can.
Tolerance	Number (float) of standard deviations away from mean that the supervisor will regard as being deviant. Only makes sense if the "supervision is "Relative".
Std	Number (float) in range (-1,1) for supervisor to regard as the floor for performance. Only makes sense if the supervision is "Fixed".
Punishment	Number (float) subtracted from each deviant bureaucrat.
SupObsParms	Vector (mean,sd) for the observability of the supervisor. Can be redunddant with omniscient, which sets all bureaucrats as visiible (i.e., equivalent to SupObsPamrs=c(1,0)
PrefParms	Vector (mean,sd) for the preferences of the bureaucrat.
ResponseParms	Vector (mean,sd) for the responses of the bureaucrats.
BurObsParms	Vector (mean,sd) for the observability of each bureaucrat to one another.
debug	If TRUE, prints debugging information.
quiet	If TRUE, prints no startup information (but will print debugging information, if requested).

Details

This might be the place for text about the way the function will work.

Value

ExecSummary	Returns a (tidy-ready) dataframe that contains: Replication, Iteration, SupUtilMean, SupUtilSD, PrefMean, PrefSD, ResponseMean, ResponseSD, SupObsMean, SupObsSD, BurObsMean, BurObsSD, Tolerance (or Std), Punishment, Connectivity, FinalResponseMean, FinalResponseSD, BurUtilMean, BurUtilSD
version	2.02
NumBurs	Number of bureaucrats (default=10)
MaxIter	Number of iterations per replication (default=10)
supervision	The mode of supervision ("Relative", "Fixed")
posprefs	Are the preferences constrained to be positive?
omniscient	Is the supervisor omniscient? (I.e., can see performance of all her subordinates?)
Performance	Matrix of performance of each bureaucrat in each iteration of every replication, (Replications*MaxIter*Numbers) x (replication, iteration, (vector of responses)))
ObstyRecord	Matrix of observability for each replication, (MaxIter*Numburs) x Numburss)
SeenRecord	Matrix of observability of each bureaucrat by supervisor in each iteration of every replication, (Replication*MaxIter*NumBurs) x (replication, iteration, (vector of seen bureaucrats))

Note

Makes use of the igraph library. You should have it installed.

Author(s)

John Brehm

References

Brehm, John and Scott Gates. 1997. Working, Shirking, and Sabotage. Ann Arbor, MI

Examples

# To replicate Table 3.1 with relative supervision, allowing for preferences to be (-1,1)
wss <- ImitSim(1000)
hist(wss$ExecSummary$SupUtilMean)

jjbrehm/BadApple documentation built on Dec. 27, 2020, 3:30 a.m.