README file: Confidence bands for a distribution function based on the Berk - Jones and reversed Berk - Jones statistics 1. These C-programs were developed by Leah Jager starting from C-programs written by Art Owen at Stanford. They implement the confidence bands discussed in the (forthcoming) paper: Jager, L. and Wellner, J. A. (2004). A new goodness of fit test: the reversed Berk-Jones statistic. Technical Report No. xxx, Department of Statistics, University of Washington. We are very grateful to Art Owen for sharing his C-programs with us. We have modified his programs in several respects so that the resulting bands have accurate / correct coverage probabilities. We take full responsibility for any errors resulting from the changes made. 2. Steps for producing plots of the empirical distribution function and (1-alpha)% confidence bands for the edf: (1) Put data in a file "data.dat" separated by spaces. Format should be: d1 d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12 d13 d14 d15 d16 d17 d18 d19 d20 d21 d22 ... (2) Open either "bj.c" (for bands based on the Berk-Jones statistic) or "reversed.c" (for bands based on the reversed Berk-Jones statistic) and insert proper values for N, outputFileName, and ALPHA in lines 50-52. N is the number of data points (must be 2 <= N <= 1000 for bj.c) (must be 3 <= N <= 1000 for reversed.c) outputFileName is the file where you want the confidence bands written ALPHA is your confidence level (must be 1e-8 <= ALPHA < 1) (3) Compile and run "reversed.c" or "bj.c": gcc -lm bj.c -o bj bj . (4) In R or Splus, run the script in "plot.txt". This can be done by typing source("plot.txt"). In this file, you will have to designate the pathways to the files containing the data and the confidence bands. (5) The confidence bands corresponding to either the Berk-Jones statistic (Owen) or the reversed Berk-Jones statistic (Jager and Wellner) should appear! Written by Leah Jager and Jon Wellner, January 2004.