Hi, Once you get one failure, this ipso facto disproves the hypothesis of 100% success. Probably you want a confidence interval for the probability of failure. There are many methods for confidence intervals for proportions. Since the proportion of failures is low, the method based on Poisson approximation would probably be a good choice. Good luck, David At 03:11 AM 12/2/2008, you wrote: >Dear Allstat, > >Your help are much appreciated on the following question: > >Experiment: Showing a very dark and a very bright dot to the >participants asking them to identify the bright one. > >Hypothesis: It was expected that all participants identify the correct dot. > >Four groups were involved in this experiment: > >Correct answers in group 1: 99 out of 100 > >Correct answers in group 2: 98 out of 100 > >Correct answers in group 3: 86 out of 100 > >Correct answers in group 4: 82 out of 100 > >Would you please advise how I could test my hypothesis i.e. the >difference between observed (99, 98, 86, 82) and expected (100). > >Thank you very much for your support and assistance in this matter. > >All the best, >Kaveh David M. Zucker Department of Statistics, Hebrew University Mt. Scopus, Jerusalem ISRAEL Phone: +972-2-5881291, FAX: +972-2-588-3549 E-mail: [log in to unmask]