Investigating disciplinary differences in the relationships between citations and downloads

Several studies have examined the relationships between citation and download data. Some have also analyzed disciplinary differences in the relationships by comparing a few subject areas or a few journals. To gain a deeper understanding of the disciplinary differences, we carried out a comprehensive study investigating the issue in five disciplines of science, engineering, medicine, social sciences, and humanities. We used a systematic method to select fields and journals ensuring a very broad spectrum and balanced representation of various academic fields. A total of 69 fields and 150 journals were included. We collected citation and download data for these journals from China Academic Journal Network Publishing Database, the largest Chinese academic journal database in the world. We manually filtered out non-research papers such as book reviews and editorials. We analyzed the relationships both at the journal and the paper level. The study found that social sciences and humanities are different from science, engineering, and medicine and that the pattern of differences are consistent across all measures studied. Social sciences and humanities have higher correlations between citations and downloads, higher correlations between downloads per paper and Journal Impact Factor, and higher download-to-citation ratios. The disciplinary differences mean that the accuracy or utility of download data in measuring the impact are higher in social sciences and humanities and that download data in those disciplines reflect or measure a broader impact, much more than the impact in citing authors.     

Succinct effect or informative effect: the relationship between title length and the number of citations

Previous studies produced mixed results with respect to the correlation between title length and number of citations. This research aims to provide a new explanation for this controversy by examining one of the largest pools of papers to date, containing over 300,000 economics papers spanning a much longer time period (1956–2012). The results show that correlation between title length and the number of citations is negative between 1956 and 2000, but becomes positive after 2000, when online searches became the predominant method for literature retrieval. Moreover, heterogeneity analyses show that longer titles are especially crucial for papers with relatively lower influence, which researchers can typically only access using online searches.     

Comparison of downloads,citations and readership data for two information systems journals

In our article we compare downloads from ScienceDirect, citations from Scopus and readership data from the social reference management system Mendeley for articles from two information systems journals (“Journal of Strategic Information Systems” and “Information and Management”) published between 2002 and 2011. Our study shows a medium to high correlation between downloads and citations (Spearman r = 0.77/0.76) and between downloads and readership data (Spearman r = 0.73/0.66). The correlation between readership data and citations, however, was only medium-sized (Spearman r = 0.51/0.59). These results suggest that there is at least “some” difference between the two usage measures and the citation impact of the analysed information systems articles. As expected, downloads and citations have different obsolescence characteristics. While the highest number of downloads are usually made in the publication year and immediately afterwards, it takes several years until the citation maximum is reached. Furthermore, there was a re-increase in the downloads in later years which might be an indication that citations also have an effect on downloads to some degree.     

Relationship between downloads and citations at journal and paper levels,and the influence of language

Download indicators are of major potential interest because the great quantity of readily available download data means that any statistical inferences drawn from them will be of robust significance. We study the relationship between citation and downloads at the journal and paper levels, and the influence of language on that relationship. The data used were taken from the Scopus (citations) and ScienceDirect (downloads) databases. The results showed that downloads have limited utility as predictors of citation since it is in the early years when any correlations have the least significance. The relationship between downloads and citation also differs from one discipline to another. The relationship at the paper level is considerably weaker than at the journal level. This may be indicative of the number of downloads depending largely on the diffusion of the journal. In francophone regions, downloading from journals is proportionately less than citations to those same journals. There seems to be a part of citations to non-English-language journals which is invisible to Scopus. This makes the number of downloads proportionately greater than that of citations, leading to a lack of correlation between downloads and citations in that class of journal.     

Does the arXiv lead to higher citations and reduced publisher downloads for mathematics articles?

An analysis of 2, 765 articles published in four math journals from 1997 to 2005 indicate that articles deposited in the arXiv received 35% more citations on average than non-deposited articles (an advantage of about 1.1 citations per article), and that this difference was most pronounced for highly-cited articles. Open Access, Early View, and Quality Differential were examined as three non-exclusive postulates for explaining the citation advantage. There was little support for a universal Open Access explanation, and no empirical support for Early View. There was some inferential support for a Quality Differential brought about by more highly-citable articles being deposited in the arXiv. In spite of their citation advantage, arXiv-deposited articles received 23% fewer downloads from the publisher’s website (about 10 fewer downloads per article) in all but the most recent two years after publication. The data suggest that arXiv and the publisher’s website may be fulfilling distinct functional needs of the reader.     

Revisiting the relationship between downloads and citations: a perspective from papers with different citation patterns in the case of the Lancet

Scientometrics - Downloads and citations of papers are regarded as significant metrics of the academic level of researchers and the influence of journals. Besides, exploring the association between...     

The log-linear relation between patent citations and patent value

This paper reports the results of an analysis of patent citation and patent renewal data, advancing a log-linear relation between patent citations and patent value. A complementary analysis of firms’ patent portfolios confirms that modelling the relation between citations and firm value benefits from the adoption of the log-linear form.     

A probe into 66 factors which are possibly associated with the number of citations an article received

Scientometrics - The number of citations has been widely used for scientific evaluation of publications and even institutions and individual scientists. However, there is as yet no satisfactory...     

Towards understanding the relation between citations and research quality in software engineering studies

The importance of achieving high quality in research practice has been highlighted in different disciplines. At the same time, citations are utilized to measure the impact of academic researchers and institutions. One open question is whether the quality in the reporting of research is related to scientific impact, which would be desired. In this exploratory study we aim to: (1) Investigate how consistently a scoring rubric for rigor and relevance has been used to assess research quality of software engineering studies; (2) Explore the relationship between rigor, relevance and citation count. Through backward snowball sampling we identified 718 primary studies assessed through the scoring rubric. We utilized cluster analysis and conditional inference tree to explore the relationship between quality in the reporting of research (represented by rigor and relevance) and scientiometrics (represented by normalized citations). The results show that only rigor is related to studies’ normalized citations. Besides that, confounding factors are likely to influence the number of citations. The results also suggest that the scoring rubric is not applied the same way by all studies, and one of the likely reasons is because it was found to be too abstract and in need to be further refined. Our findings could be used as a basis to further understand the relation between the quality in the reporting of research and scientific impact, and foster new discussions on how to fairly acknowledge studies for performing well with respect to the emphasized research quality. Furthermore, we highlighted the need to further improve the scoring rubric.     

Comparison of number of citations to full original articles versus brief reports

Most biomedical journals accept original research articles in the form of “brief reports”. We compared the citations to full papers versus brief reports in a sample of journals on Infectious Diseases, Clinical Microbiology, and Antimicrobial Agents. Brief reports were cited less often than full-size articles [regression coefficient: 10.94 (95 % CI: 5.19, 16.69)] even after adjustment for the journal’s impact factor. Our findings may influence decisions of editors and authors regarding brief reports.     

Comparison of scientific impact expressed by the number of citations in different fields of science

Summary Citation distributions for 1992, 1994, 1996, 1997, 1999, and 2001, which were published in the 2004 report of the National Science Foundation, USA, are analyzed. It is shown that the ratio of the total number of citations of any two broad fields of science remains close to constant over the analyzed years. Based on this observation, normalization of total numbers of citations with respect to the number of citations in mathematics is suggested as a tool for comparing scientific impact expressed by the number of citations in different fields of science.     

Relation of title length of journal articles to number of authors

The great importance of titles being highly informative is almost unanimously accepted in literature, assuming that the more informative titles are, the more effectively they serve their functions. The most common measure of title informativeness has been the number of substantive words included in it, and one of the factors which might be associated with it is the number of authors. The present study attempted to test, in a large group of journals fromdifferent areas, and over six decades, the hypothesis that a paper signed by a larger number of authors will have more substantive words in its title. Large samples of original research papers were drawn from each decade year of fourteen leading journals. For each paper, the number of substantive words in the title was correlated with the number of authors. Findings indicate a difference between the scientific journals on the one hand, and the social sciences and humanities journals on the other. A moderate positive correlation was found in most scientific journals (excluding mathematics) for many periods. In the social sciences journals, and to a greater extent, in the humanities journals, a significant positive correlation was limited to only a few periods, while the rest showed a very low correlation, or even a negative correlation. The different findings for the sciences may be somehow associated with their higher rate of multiple authorship.Paper presented at the Fourth International Conference on Bibliometrics, Informetrics and Scientometrics in Berlin (Germany), September 11–15, 1993.     

Reassessing the N-rays reference network: The role of self citations and negative citations

In his article networks of Scientific Papers,Price argued that the N-rays reference network exhibits characteristics one would expect for a cumulative and rapidly developing research area. Although subsequent researchers have questioned Price's characterization of the N-rays network, there have been no replications of Price's work for either the N-rays literature or for any other literature. We reexamine the N-rays reference network, this time distinguishing negative citations and self citations from other citations. Although previous studies of negative and self citations show they are relatively infrequent in scientific literatures, we find that both are very prominent in the N-rays literature. In addition, we show that self citations comprise most of the recency effect observed in the N-rays reference network, and that the high level of self citations in the N-rays literature results primarily from the character of the journal that published the majority of the N-rays papers. Our findings therefore support those who have been skeptical about Price's claim that the N-rays reference graph exemplifies basic characteristics of the structure of scientific literatures.     

Performance and its relation with productivity in Lotkaian systems

In general information production processes (IPPs), we define productivity as the total number of sources but we present a choice of seven possible definitions of performance: the mean or median number of items per source, the fraction of sources with a certain minimum number of items, the h-, g-, R- and h_w-index. We give an overview of the literature on different types of IPPs and each time we interpret “performance” in these concrete cases. Examples are found in informetrics (including webometrics and scientometrics), linguistics, econometrics and demography. In Lotkaian IPPs we study these interpretations of “performance” in function of the productivity in these IPPs. We show that the mean and median number of items per source as well as the fraction of sources with a certain minimum number of items are increasing functions of the productivity if and only if the Lotkaian exponent is decreasing in function of the productivity. We show that this property implies that the g-, R- and h_w-indices are increasing functions of the productivity and, finally, we show that this property implies that the h-index is an increasing function of productivity. We conclude that the h-index is the indicator which shows best the increasing relation between productivity and performance.     

The distribution of the uncitedness factor and its functional relation with the impact factor

The uncitedness factor of a journal is its fraction of uncited articles. Given a set of journals (e.g. in a field) we can determine the rank-order distribution of these uncitedness factors. Hereby we use the Central Limit Theorem which is valid for uncitedness factors since it are fractions, hence averages. A similar result was proved earlier for the impact factors of a set of journals. Here we combine the two rank-order distributions, hereby eliminating the rank, yielding the functional relation between the impact factor and the uncitedness factor. It is proved that the decreasing relation has an S-shape: first convex, then concave and that the inflection point is in the point (μ′, μ) where μ is the average of the impact factors and μ′ is the average of the uncitedness factors.     

Progressive nucleation mechanism for the growth behavior of items and its application to cumulative papers and citations of individual authors

A general expression based on the concepts of the progressive nucleation mechanism is proposed in the form [Formula: see text] to describe the growth behavior of items in an individual system and a collective of systems. In the above relation, α(t) is the ratio of items N(t) at time t to the maximum number C of possible items for the system, Θ is the corresponding time constant and q is the exponent. The above relation is then used to analyze: (1) the growth behavior of cumulative number N(t) of papers published by individual authors and cumulative citations L(t) of N(t) papers of an author as a function of citation duration t, and (2) the relationship between cumulative citations L(t) of papers and cumulative number N(t) of papers. The proposed approach predicts that: (1) the fraction of items produced by successive systems is additive, (2) the cumulative fraction α(sum)(t) of maximum number of sites is the sum of contributions of fractions of maximum number of items produced by different systems, and (3) the values of time constants Θ and exponent q increase with the addition of fraction of items produced by subsequent systems, but their values are the lowest for individual systems. The approach is applied to explain the growth behavior of cumulative N(t) papers and L(t) citations of four selected Polish professors.     

On the ferroelectric dc nonlinear dielectric permittivity and its relation with the macroscopic polarization

Journal of Materials Science - The relation between the nonlinear dielectric permittivity $$\varepsilon '(E)$$ and the macroscopic polarization P(E) is obtained for ferroelectrics considering a...     

Modeling the coevolution between citations and coauthorship of scientific papers

Scientometrics - Collaborations and citations within scientific research grow simultaneously and interact dynamically. Modelling the coevolution between them helps to study many phenomena that can...