UDC 519.254 PACS 02.50.Ng 06.20.Dk 07.05.Kf A Method for Statistical Comparison of Histograms S. I. Bityukov∗, N. V. Krasnikov†, A. N. Nikitenko‡, V. V. Smirnova∗ ∗ Institute for High Energy Physics 1, pl. Nauki, Protvino, Moscow region, Russia, 142281 † INR RAS 7a, prospekt 60-letiya Oktyabrya, Moscow, Russia, 117312 ‡ Imperial College Sci., Tech. & Med, London, UK § Joint Institute for Nuclear Research 6, Joliot-Curie str., Dubna, Moscow region, Russia, 141980 The problem of the testing the hypothesis that two histograms are drawn from the same distribution is a very important problem in many scientific researches. <...> There are several approaches to formalize and resolve this problem. <...> We propose an approach for testing the hypothesis that two realizations of the random variables in the form of histograms are taken from the same statistical population (i.e. two histograms are drawn from the same distribution). <...> The approach is based on the notion “significance of deviation”, which has a distribution close to standard normal distribution if both histograms are drawn from the same distribution. <...> This approach allows to estimate the statistical difference between two histograms using multi-dimensional test statistics. <...> The distinguishability of histograms is estimated with the help of the construction a number of clones (rehistograms) of the observed histograms. <...> The approach considered in the paper allows to perform the comparison of histograms with a test more powerful, in the cases considered, than those that use only one test statistic. <...> Also, the probability of correct decision is used as an estimate of the quality of the decision about the distinguishability of histograms. <...> Let the experimental facility register the flow of events during two independent time intervals [t1, t2] and [t3, t4]. <...> Events from first time interval belong to statistical population of events G1, events from second time interval belong to statistical population of events G2. <...> If facility (beam, detectors, data acquisition system, .) is in norm during both time intervals then the properties of events, registered in the facility during time interval [t1, t2], is the same as the properties of events, registered in the facility during time interval [t3, t4], i.e. G1 = G2. <...> If facility is out of norm during one of time intervals then the properties of events from statistical population G1 differ from the properThe test of the hypothesis <...>