The influence of references per paper in the SCI to Impact Factors and the Matthew Effect

Abstract

All references data was extracted from the annual volumes of the CD-Edition of Science Citation Index (SCI) and the Web of Science of the Institute for Scientific Information (ISI), the journal citation and self-citation data extracted from the Journal Citation Report (JCR), the self-citing rate and self-cited rate calculated based on the JCR method. To determine the trend of mean value of references per paper throughout 1970–2005, a total number of 10,000 records were randomly chosen for each year of under study, and the mean value of references per paper was calculated. To determine the growth of journals IF a total number of 5,499 journals were chosen in the JCR in 2002 and the same set of journals in the year 2004. To show the trend of journals IF, all journals indexed in the JCR throughout 1999–2005 were extracted and the mean values of their IFs was calculated annually. The study showed that the number of references per paper from 1970 to 2005 has steady increased. It reached from 8.40 in 1970 to 34.63 in 2005, an increase of more than 4 times. The majority of publications (76.17%) were in the form of Journals Article. After articles, Meeting Abstracts (9.46%), Notes (3.90%) and Editorial Material (3.78%) are the most frequented publication forms, respectively. 94.57% of all publications were in English. After English, German (1.50%), Russian (1.48%) and French (1.37%) were the most frequented languages, respectively. The study furthermore showed that there is a significant correlation between the IF and total citation of journals in the JCR, and there is an important hidden correlation between IF and the self-citation of journals. This phenomena causes the elevation of journals IF. The more often a journal is citing other journals, the more often it is also cited (by a factor of 1.5) by others. In consequence the growing percentage of journal self-citation is followed by journal self-citedness, which can be considered as the Matthew Effect. There is a linear correlation between journal self-citing and journal self-cited value, the mean value of self-cited rate always stays higher than the self-citing rate. The mean value of self-cited rate in 2000 was 14% and the mean value of self-citing rate is 6.61%, whereas the mean value of self-cited rate in 2005 was 12% and the mean value of self-citing rate was 7.81%.