sna (version 2.4)

event2dichot: Convert an Observed Event Matrix to a Dichotomous matrix

Description

Given one or more valued adjacency matrices (possibly derived from observed interaction ``events''), event2dichot returns dichotomized equivalents.

Usage

event2dichot(m, method="quantile", thresh=0.5, leq=FALSE)

Arguments

m

one or more (valued) input graphs.

method

one of ``quantile,'' ``rquantile,'' ``cquantile,'' ``mean,'' ``rmean,'' ``cmean,'' ``absolute,'' ``rank,'' ``rrank,'' or ``crank''.

thresh

dichotomization thresholds for ranks or quantiles.

leq

boolean indicating whether values less than or equal to the threshold should be taken as existing edges; the alternative is to use values strictly greater than the threshold.

Value

The dichotomized data matrix (or matrices)

Details

The methods used for choosing dichotomization thresholds are as follows:

1. quantile: specified quantile over the distribution of all edge values

2. rquantile: specified quantile by row

3. cquantile: specified quantile by column

4. mean: grand mean

5. rmean: row mean

6. cmean: column mean

7. absolute: the value of thresh itself

8. rank: specified rank over the distribution of all edge values

9. rrank: specified rank by row

10. crank: specified rank by column

Note that when a quantile, rank, or value is said to be ``specified,'' this refers to the value of thresh.

References

Wasserman, S. and Faust, K. (1994). Social Network Analysis: Methods and Applications. Cambridge: Cambridge University Press.

Examples

# NOT RUN {
#Draw a matrix of normal values
n<-matrix(rnorm(25),nrow=5,ncol=5)

#Dichotomize by the mean value
event2dichot(n,"mean")

#Dichotomize by the 0.95 quantile
event2dichot(n,"quantile",0.95)

# }