transfer {tfer}

Glass Transfer, Persistence and Recovery Probabilities

Description

Construct a transfer object to simulate the number of glass fragments recovered given the conditions set by the user.

Usage

transfer(
  N = 10000,
  d = 0.5,
  deffect = TRUE,
  lambda = 120,
  Q = 0.05,
  l0 = 0.8,
  u0 = 0.9,
  lstar0 = 0.1,
  ustar0 = 0.15,
  lj = 0.45,
  uj = 0.7,
  lstarj = 0.05,
  ustarj = 0.1,
  lR = 0.5,
  uR = 0.7,
  lt = 1,
  ut = 2,
  r = 0.5,
  timeDist = c("negbin", "cnegbin", "uniform"),
  loop = FALSE
)

Arguments

N

Simulation size

d

The breaker's distance from the window

deffect

Distance effect. deffect = TRUE when distance effect exists. Otherwise deffect = FALSE.

lambda

The average number of glass fragments transferred to the breaker's clothing.

Q

Proportion of high persistence fragments.

l0

Lower bound on the percentage of fragments lost in the first hour

u0

Upper bound on the percentage of fragments lost in the first hour

lstar0

Lower bound on the percentage of high persistence fragments lost in the first hour

ustar0

Upper bound on the percentage of high persistence fragments lost in the first hour

lj

Lower bound on the percentage of fragments lost in the j'th hour

uj

Upper bound on the percentage of fragments lost in the j'th hour

lstarj

Lower bound on the percentage of high persistence fragments lost in the j'th hour

ustarj

Upper bound on the percentage of high persistence fragments lost in the j'th hour

lR

Lower bound on the percentage of fragments expected to be detected in the lab

uR

Upper bound on the percentage of fragments expected to be detected in the lab

lt

Lower bound on time between commission of crime and apprehension of suspect

ut

Upper bound on time between commission of crime and apprehension of suspect

r

Probability r in ti ~ NegBinom(t, r)

timeDist

the distribution for the random amount of time between the commission of the crime and the apprehension of the suspect. There are three choices "negbin", "cnegbin", and "uniform". Before talking about these it should be noted that if lt is equal to ut - then there is no randomness in this calculation. If lt does not equal ut, then the average of these two values is used in the two negative binomial options: "negbin" and "cnegbin". The difference betweeen them is that "cnegbin" is a constrained negative binomial where the allowable times are constrained to be between lt and ut. If "uniform" is selected, then a uniformly distributed random time between lt and ut is used in each iteration.

loop

if TRUE an element by element version of the simulation is used, if FALSE then a (mostly) vectorised element version of the simulation is used. The results from the two methods appear to be almost identical - they won't be the same even with the same seed because of the way the random variates are generated. I (James) believe the vectorised version is faster and better. There was also a small mistake which has been corrected in that the initial set of persistent fragments was not being

Value

a list containing:

results

The simulated values of recovered glass fragments

paramList

Input parameters

The returned object has S3 class types tfer and transfer for backwards compatibility

Author(s)

James Curran and TingYu Huang

References

Curran, J. M., Hicks, T. N. & Buckleton, J. S. (2000). Forensic interpretation of glass evidence. Boca Raton, FL: CRC Press.

Curran, J. M., Triggs, C. M., Buckleton, J. S., Walsh, K. A. J. & Hicks T. N. (January, 1998). Assessing transfer probabilities in a Bayesian interpretation of forensic glass evidence. Science & Justice, 38(1), 15-21.

Examples

library(tfer)

## create a transfer object using default arguments
y = transfer()

## probability table
probs = tprob(y)

## extract the probabilities of recovering 8 to 15
## glass fragments given the user-specified arguments
tprob(y, 8:15)

## produce a summary table for a transfer object
summary(y)

## barplot of probabilities (default)
plot(y)
plot(y)

## barplot of transfer frequencies
plot(y, ptype = "f")

## histogram
plot(y, ptype = "h")