epi.psi {epiR}

Proportional similarity index

Description

Compute proportional similarity index.

Usage

epi.psi(dat, itno = 99, conf.level = 0.95)

Arguments

dat

a data frame providing details of the distributions to be compared (in columns). The first column (either a character of factor) lists the levels of each distribution. Additional columns list the number of events for each factor level for each distribution to be compared.

itno

scalar, numeric defining the number of bootstrap simulations to be run to generate a confidence interval around the proportional similarity index estimate.

conf.level

scalar, numeric defining the magnitude of the returned confidence interval for each proportional similarity index estimate.

Details

The proportional similarity or Czekanowski index is an objective and simple measure of the area of intersection between two non-parametric frequency distributions (Feinsinger et al. 1981). PIS values range from 1 for identical frequency distributions to 0 for distributions with no common types. Bootstrap confidence intervals for this measure are estimated based on the approach developed by Garrett et al. (2007).

Value

A five column data frame listing: v1 the name of the reference column, v2 the name of the comparison column, est the estimated proportional similarity index, lower the lower bound of the estimated proportional similarity index, and upper the upper bound of the estimated proportional similarity index.

References

Feinsinger P, Spears EE, Poole RW (1981) A simple measure of niche breadth. Ecology 62: 27 - 32.

Garrett N, Devane M, Hudson J, Nicol C, Ball A, Klena J, Scholes P, Baker M, Gilpin B, Savill M (2007) Statistical comparison of Campylobacter jejuni subtypes from human cases and environmental sources. Journal of Applied Microbiology 103: 2113 - 2121. DOI: 10.1111/j.1365-2672.2007.03437.x.

Mullner P, Collins-Emerson J, Midwinter A, Carter P, Spencer S, van der Logt P, Hathaway S, French NP (2010). Molecular epidemiology of Campylobacter jejuni in a geographically isolated country with a uniquely structured poultry industry. Applied Environmental Microbiology 76: 2145 - 2154. DOI: 10.1128/AEM.00862-09.

Rosef O, Kapperud G, Lauwers S, Gondrosen B (1985) Serotyping of Campylobacter jejuni, Campylobacter coli, and Campylobacter laridis from domestic and wild animals. Applied and Environmental Microbiology, 49: 1507 - 1510.

Examples

## EXAMPLE 1:
## A cross-sectional study of Australian thoroughbred race horses was 
## carried out. The sampling frame for this study comprised all horses 
## registered with Racing Australia in 2017 -- 2018. A random sample of horses
## was selected from the sampling frame and the owners of each horse 
## invited to take part in the study. Counts of source population horses
## and study population horses are provided below. How well did the geographic
## distribution of study population horses match the source population?    

state <- c("NSW","VIC","QLD","WA","SA","TAS","NT","Abroad")
srcp <- c(11372,10722,7371,4200,2445,1029,510,101)
stup <- c(622,603,259,105,102,37,22,0)
dat.df01 <- data.frame(state, srcp, stup)

epi.psi(dat.df01, itno = 99, conf.level = 0.95)

## The proportional similarity index for these data was 0.88 (95% CI 0.86 to 
## 0.90). We conclude that the distribution of sampled horses by state 
## was consistent with the distribution of the source population by state. 

## Not run: 
## Compare the relative frequencies of the source and study populations
## by state graphically:
library(ggplot2)

dat.df01$psrcp <- dat.df01$srcp / sum(dat.df01$srcp)
dat.df01$pstup <- dat.df01$stup / sum(dat.df01$stup)
dat.df01 <- dat.df01[sort.list(dat.df01$psrcp),]
dat.df01$state <- factor(dat.df01$state, levels = dat.df01$state)

## Data frame for ggplot2:
gdat.df01 <- data.frame(state = rep(dat.df01$state, times = 2),
   pop = c(rep("Source", times = nrow(dat.df01)), 
      rep("Study", times = nrow(dat.df01))),
   pfreq = c(dat.df01$psrcp, dat.df01$pstup))
gdat.df01$state <- factor(gdat.df01$state, levels = dat.df01$state)

## Bar chart of relative frequencies by state faceted by population:
ggplot(data = gdat.df01, aes(x = state, y = pfreq)) +
  geom_bar(stat = "identity", position = position_dodge(), color = "grey") + 
  facet_grid(~ pop) +
  scale_x_discrete(name = "State") +
  scale_y_continuous(limits = c(0,0.50), name = "Proportion")

## End(Not run)