Collaboration networks and scientific impact among behavioral ecologists
Collaboration networks and scientific impact among behavioral ecologists
复制标题
DOI：
10.1093/beheco/arp194
复制
发表时间：
2010-03-01
影响因子：
2.4
通讯作者：
Pike, Thomas W.
中科院分区：
文献类型：
--
作者：
Pike, Thomas W.
量化作者的科学影响对于评估和比较的目的（例如，大学招聘和晋升或授予资助）变得越来越重要。为此，已经提出了一些定量指标，（原则上）允许比较个人的科学质量或影响（Cartwright and McGhee 2005; Cheek et al. 2006; Meho 2007；尽管关于这种方法的潜在缺陷的讨论，参见Garfield 1979； MacRoberts and MacRoberts 1996）。这些通常分为声誉，产量或生产力，影响或影响（Avital和Collopy 2001）。常用的指标包括发表的论文总数，这通常用于衡量基础生产力，可能与获得的资金和研究小组规模等因素呈正相关；被引用的平均或总次数，被认为表明一项研究的科学效用，因此可以用作研究质量和影响的部分指标（Lawani 1986）；论文发表的期刊以及这些期刊的影响参数（例如，Steinpreis et al. 1999）。特别是，出版物生产率和引用频率指标通常用于评估影响力，尽管也有人提出将这两种指标纳入单一指标的其他指标。例如，Hirsch（2005）开发的h指数旨在通过观察发表的论文数量以及他/她的作品被引用的次数来衡量研究人员产出的累积影响。对这些指标的有效评估可能需要了解诸如可见性、引用社区的规模以及融入社会和专业网络的程度等因素（Ward et al. 1992）。然而，这些因素与科学影响之间的关系尚不清楚。在这里，我研究了个人科学影响的变化是如何与科学合作网络（特别是合著网络）的结构相关的，他们是其中的一部分。社会网络（协作网络就是一个例子）是个体的集合，其中每个个体都与其他个体的某个子集熟悉。科学学科内的这种网络结构可能对个人和研究人员群体（如研究小组、部门或大学）的科学产出的质量（和数量）有重要影响，因为有些人会与他人广泛互动，而有些人只与可用个人的很小一部分互动。原则上，这些互动可以采取任何社会联系的形式，比如那些发生在会议上、喝咖啡时或通过电子邮件通信时的互动，但这些很难量化。相反，研究论文的合著最近被用作科学合作的代表，以便成功地构建合作网络（例如，Newman 2001）。这种科学认识的定义似乎是合理的，因为大多数一起写过论文的人会彼此非常了解。此外，这些数据可以方便而准确地从出版物数据库中提取。将共同作者作为合作的指标，一些研究表明，共同作者的数量增加了某些学科中个别文章的被引率（Abt 1984; Smart and Bayer 1986; Katz and Hicks 1997; Glänzel 2002；尽管参见Smart and Bayer 1986； Avkiran 1997; Rousseau 2001; Leimu and Koricheva 2005b）。因此，科学合作被普遍认为可以提高质量和…
Quantifying an author’s scientific impact is becoming increasingly important for evaluation and comparison purposes (eg, for university recruitment and advancement or the award of grants). To this end, a number of quantitative metrics have been proposed that (in principle) allow the comparison of individuals’ scientific quality or impact (Cartwright and McGhee 2005; Cheek et al. 2006; Meho 2007; although for a discussion on the potential pitfalls of this approach, see eg, Garfield 1979; MacRoberts and MacRoberts 1996). These generally fall into the categories of reputation, yield or productivity, and influence or impact (Avital and Collopy 2001). Commonly used metrics include the total number of papers published, which is commonly used to gauge basal productivity and is likely to be positively correlated with factors such as funding obtained and research group size; the mean or total number of citations received, which are assumed to indicate the scientific utility of a study and can thus be used as a partial indicator of a study’s quality and impact (Lawani 1986); and the journals where the papers were published and these journals’ impact parameter (eg, Steinpreis et al. 1999). In particular, publication productivity and measures of citation frequency are commonly used to assess influence, although other metrics have also been proposed that incorporate both measures into a single metric. For example, the h-index, developed by Hirsch (2005), aims to measure the cumulative impact of a researcher’s output by looking at the quantity of papers published along with the number citations his/her work has received. The effective evaluation of such metrics is likely to require an understanding of factors such as visibility, the size of citing community, and the extent of integration into social and professional networks (Ward et al. 1992). However, the relationship between these factors and scientific impact is unclear. Here, I investigate how variation in individual scientific impact is related to the structure of the scientific collaboration network (specifically, the coauthorship network) of which they are a part. A social network (of which a collaboration network is one example) is a collection of individuals, each of whom is acquainted with some subset of the others. The structure of such a network within a scientific discipline may have important implications for the quality (and quantity) of scientific output, both for individuals and groups of researchers (such as research groups, departments, or universities), as some individuals will interact widely with others, whereas some will interact with just a very small subset of the available individuals. These interactions could in principle take the form of any social association, such as those that occur at conferences, over coffee, or via email correspondence, but these are difficult to quantify. Instead, coauthorship of research papers has recently been used as a proxy for scientific collaboration in order to successfully construct collaboration networks (eg, Newman 2001b). This definition of scientific acquaintance seems reasonable because most people who have written a paper together will know one another quite well. Moreover, these data can be readily and accurately extracted from publication databases. Using coauthorship as an indicator of collaborations, several studies have shown that the number of coauthors increases the citation rates of individual articles in some disciplines (Abt 1984; Smart and Bayer 1986; Katz and Hicks 1997; Glänzel 2002; although see Smart and Bayer 1986; Avkiran 1997; Rousseau 2001; Leimu and Koricheva 2005b). Consequently, scientific collaboration is widely assumed to enhance the quality and …