Function compares graphically the readings of two age readers and calculates 2 chi-square statistics for tests of symmetry.
compare2(readings, usecols = c(1,2), twovsone = TRUE, plot.summary = TRUE,
barplot = TRUE, chi = TRUE, pool.criterion = 1, cont.cor = TRUE,
correct = "Yates", first.name = "Reader A",second.name = "Reader B")Separate lists with tables of various statistics associated with the method.
dataframe or matrix containing the readings by Reader 1 and those by Reader 2.
columns of the dataframe or matrix corresponding to the readings of Reader 1 and those of Reader 2. Default=c(1,2).
logical for whether first type of graph is produced.
logical for whether summary table is put on first graph.
logical for whether barplot of frequency of disagreements is drawn.
logical for whether 2 chi-square tests are performed.
used to collapse pairs where the expected number of observations is < pooling criterion (default is 1).
logical for whether a continuity correction should be used in 1st chisquare test.
character for whether "Yates" or "Edwards" continuity correction should be done (if cont.cor=TRUE).
character string describing the first reader or the first aging method. The default is to specify "Reader A".
character string describing the second reader or the second aging method. The default is to specify "Reader B".
John Hoenig, Virginia Institute of Marine Science, 18 December 2012. hoenig@vims.edu
This function can plot the number of readings of age j by reader 2 versus the number of readings of age i by reader 1 (if twovsone=TRUE). Optionally, it will add the number of readings above, on, and below the 45 degree line (if plot.summary=TRUE). The function can make a histogram of the differences in readings (if barplot=TRUE). Finally, the program can calculate 2 chi-square test statistics for tests of the null hypothesis that the two readers are interchangeable vs the alternative that there are systematic differences between readers (if chi=TRUE). The tests are tests of symmetry (Evans and Hoenig, 1998). If cont.cor=T, then correction for continuity is applied to the McNemar-like chi-square test statistic; the default is to apply the Yates correction but if correct="Edwards" is specified then the correction for continuity is 1.0 instead of 0.5.
Evans, G.T. and J.M. Hoenig. 1998. Viewing and Testing Symmetry in Contingency Tables, with Application to Fish Ages. Biometrics 54:620-629.).
data(sbotos)
compare2(readings=sbotos,usecols=c(1,2),twovsone=TRUE,plot.summary=TRUE,
barplot=FALSE,chi=TRUE,pool.criterion=1,cont.cor=TRUE,correct="Yates",
first.name="Reader A",second.name="Reader B")
Run the code above in your browser using DataLab