Braun Tibor, Glänzel Wolfgang, Schubert András: Országok, szakterületek, folyóiratok tudománymetriai mutatószámai 1981-1985 (A MTAK Informatikai És Tudományelemzési Sorozata 6., 1992)
Indicators - Statistical reliability of scientometric indicators
A. SCHUBERT, W. GLÄNZEL, T. BRAUN : SCIENTOMETRIC DATAFILES The Attractivity Index {AAl) is defined as the country's share in citations attracted by publications in the given field AAl = -r the country's share in citations attracted by publications in all science fields or, equivalently, the given field's share in citations attracted by the country's publications AAl = the given field's share in citations attracted by all publications of the world Attractivity indices in this study are based on citation counts during 19811985 to papers published in the 1981-85 period. AAl characterizes the relative impact of a country's publications in a given subject field as reflected in the citations they attract. AAl = 1 indicates that the country's citation impact in the given field corresponds precisely to the world average; AAl > 1 reflects higher-thanaverage, AAl < 1 lower-than-average impact. It follows from the definition that any country can have high AAr% in some fields only at the expense of having lower values in others. Statistical reliability of scientometric indicators For estimating the statistical reliability of AI and AAl let us consider the second of the definitions of AI and AAl, respectively. Of the two proportions, that in the numerator is the main source of statistical error. The denominator, based on world aggregate data, is obviously loaded with a much smaller error than the contribution of a single country. That's why the relative error of AI and AAl is taken equal to that of their numerator. What is then needed are the error bounds within which the observed share of a given field in a country's publications (resp. citations) can be considered the estimator of the probability that a randomly drawn item will fall into the field in question. Using the error formula of the binomial distribution the following relations can be obtained: A AI = AI * (l/N - l/S y 1/ 2 , A AAl = AAl * (l/M - l/T )" 1/ 2 , Scien l omelrics 16 (1989) 13
