Searches for parasite interactions taking risk factors into account.

This function implements a method to correct for shared risk factors in the search for interactions. It provides the observed chi-square value, a measure of association between two parasites, and simulates bootstrapped data taking risk factors into account.

Usage

chi2Corr(formula, data.obs, namepara1, namepara2, nsimu)

Arguments

formula: a string of characters indicating a symbolic description of the model of shared risk factors to be fitted without any response variable
data.obs: the name of the data set to be used
namepara1: the name of the column giving the status to the first parasite
namepara2: the name of the column giving the status to the second parasite
nsimu: an integer indicating the number of repetitions for the bootstrap simulation

Value

The value returned is a list containing:

formula: the model fitted without any response variable
time: duration in seconds of the simulations
chi2.corr.obs: the Pearson's chi2 statistic calculated on data.obs
dispcoeff: the estimated coefficient of over- (or under-) dispersion, defined as the mean of the bootstrapped values of the corrected chi-square.
pval1: p-value of the corrected chi-square test under the null hypothesis of independence of the two parasites. pval1 was estimated assuming that the corrected chi-square is proportional to a chi-square with one degree of freedom.
pval2: p-value of the corrected chi-square test under the null hypothesis of independence of the two parasites. pval2 was given by the proportion of bootstrapped corrected chi-squares smaller than the observed value (chi2.corr.obs).
tab.th: expected frequencies, ie. the contingency table calculated on the theoretical (bootstrapped) data
tab.obs: observed frequencies, ie. the contingency table calculated on data.obs
chi2.corr.sim: a vector containing the nsimu Pearson's chi2 statistics calculated on simulated data.

The distribution of the bootstrapped corrected chi-squares (an histogram) is also provided.

Note

pval2 is better than pval1 but requires running enough simulations, wich may be long in some cases. pval1 allows working with smaller numbers of simualtions when simulation times are too long.

References

True versus False Parasite Interactions: A Robust Method to Take Risk Factors into Account and Its Application to Feline Viruses. Hellard E., Pontier D., Sauvage F., Poulet H. and Fouchet D. (2012). PLoS ONE 7(1): e29618. doi:10.1371/journal.pone.0029618.

Examples

if (FALSE) { # \dontrun{
  library(Interatrix)
  data(dataInteratrix)
  res1 <- chi2Corr("F1+F2*F3+F4", dataInteratrix, "Parasite1", "Parasite2", 500)
} # }