boot.cfa {cfa} | R Documentation |
Check if the configurations detected are stable. This is done by counting the configurations found to be significant in the CFAs performed.
boot.cfa(configmatrix, cntvector, runs=100, bonferroni=F, sig.limit=0.05)
configmatrix |
Dataframe with the variables to be analyzed |
cntvector |
Vector of counts (1 if the data are not aggregated |
runs |
Number of runs and therefore samples to be generated. |
bonferroni |
Use Bonferroni-adjustment in determining if the configuration is considered significant |
sig.limit |
Limit of p which cause the configuration to be counted as "signifcant" |
The repeated use of tests of significance is obviously problematic. The result is therfore to be interpreted in a strictly heuristic way.
Row names |
Configuration |
cnt.antitype |
Number of cases where n<expected |
cnt.type |
Number of cases where n>expected |
pct.types |
Percentage of cases where n>expected |
cnt.sig |
Number of cases where the p(z) for the configuration was considered significant |
pct.cnt.sig |
Percentage of cases where the p(z) for the configuration was considered significant |
The program is implemented in R itself rather than a
compiled library and therefore slow. It repeatedly calls cfa
.
In addition the number of runs must be high to give a reliable result.
The default of 100 is the absolute mininum. So a run of the program under
realistic conditions may take from minutes to hours.
Stefan Funke <funke@attglobal.net>
Lautsch E., von Weber S. (1995) Methoden und Anwendungen der Konfigurationsfrequenzanalyse (KFA) Beltz Verlagsunion (1995)
library(cfa) data(cfadat) boot.cfa(cfadat[c("gender","married","children")],cfadat["count"])