WHY DO WE NEED DETAILED CITATIONS?

Science represents a collective and cumulative endeavor, as illustrated by the often-used metaphor “standing on the shoulders of giants.” Every scientific work draws to some extent on previous studies (McInnis and Symes Reference McInnis and Symes1988, 388). Citations are a cornerstone of scientific research because they allow us to engage with these studies—that is, to align ourselves within a discipline, a field, or a research tradition; to claim that our arguments are valid because they are supported by other studies; and to criticize other scholars’ claims (Burbules Reference Burbules2015, 717–19). Whether or not we agree with these researchers, citations are our way of accrediting their endeavors (Merton Reference Merton1996, 290).

Equally important, citations serve the practical purpose of ensuring honesty and transparency. As Dunleavy (Reference Dunleavy2017) elucidated, citations “guide readers seeking to follow the author’s extended chain of reasoning” by allowing them “to understand and ‘replay’ the intellectual journey involved.” They enable readers to analyze the evidence as the author has done rather than to take for granted the author’s claims.

For this to happen, citations should allow readers to easily track down any cited sources as well as particular segments within those sources. In references to primary sources, citations already are expected to be as specific as possible. When citing an archival document, we are trained to provide readers with enough information to identify not only the archive where the document is stored but also the box within the archive, the folder within the box, and even the document within the folder (using details such as the document’s author, recipient, and date).Footnote ²

Likewise, when citing information from a secondary source, we should enable readers to look up both the cited book or article and the chapter, section, page, figure, table, or note where the cited information occurs. When only an author’s surname and year of publication are cited, readers might have to hunt through an entire book or article to find the relevant passage, which could be the equivalent of rummaging through an entire archival collection in search of a specific document. Even worse, general references might dissuade readers from pursuing the citation, thereby closing off the possibility of a broader discussion among readers, researchers, and sources.

That references in any discipline tend to be general often is the result of long-standing conventions. Citation styles vary across fields and reflect specific epistemological norms and traditions (McInnis and Symes Reference McInnis and Symes1988, 393). Some citation practices in the social sciences, including parenthetical and author/date formats, discourage the use of location identifiers. Their main function is to acknowledge the existence of entire works and allow readers to find them (using the list of references at the end of the paper). These practices draw on epistemologies associated with the “hard” sciences, in which an article’s findings matter more than the language used to convey them; therefore, location details are deemed less important (Connors Reference Connors1999, 238–41; Madigan, Johnson, and Linton Reference Madigan, Johnson and Linton1995, 433).

Because of their training, social scientists who internalize these norms may not feel obliged to cite precisely unless a direct quotation is involved. Some scholars misunderstand the requirement to include page numbers when citing direct quotations as an obligation to exclude identifiers when there is no direct quotation. This can result in cases of reviewers and copyeditors removing detailed citations originally placed by authors (Rekdal Reference Rekdal2014, 572). A recent guidebook demonstrates this misperception, concluding that “If you are paraphrasing an idea but not directly quoting another author’s words, your in-text citations will only need to make reference to the author and year of publication. In such cases, it is not customary to include the page numbers” (Galvan and Galvan Reference Galvan and Galvan2017, 147; my emphasis).

This indeed may be the custom in some disciplines, but it is a custom that should change in political science in keeping with current trends of transparency and scientific rigor. To be sure, there is nothing categorically wrong with general citations. As discussed previously, citations serve different functions, and location identifiers are not essential for all of them. We may want to refer to the overarching theme of a paper, which could be difficult to reduce to a specific page range (Pears and Shields Reference Pears and Shields2010, 4). Otherwise, we may be expected to show that we are familiar with relevant works in our field to establish our authority within it (Burbules Reference Burbules2015, 718–19; Madigan, Johnson, and Linton Reference Madigan, Johnson and Linton1995, 432–33). When we need to demonstrate only that a certain paper about a certain topic exists and is known to us, a general citation might suffice.

That said, detailed citations can be helpful even when we refer to a paper’s broad argument. As Henige contends (Reference Henige2006, 105), although nondetailed citations may be used “for background, comparison, or affiliation,” authors typically refer to “specific information and arguments” that can be located within a distinct part of the source. A paper’s key argument is often spelled out at least once throughout the paper. References identifying this part’s location would be more precise and useful than general references by allowing readers to verify whether the cited paper actually makes the argument paraphrased by the citing author.

In other words, we may use general citations when we want to show that a certain work is about a certain topic; however, if we want to claim that the work makes a certain argument, adding a detailed citation is advisable. Used like this, detailed citations may deter authors from citing sources they have not read “to be sure they had everything covered” and from proclaiming that these sources make arguments they do not actually make (Jensen Reference Jensen2017, 74), which often is based on the cited paper’s title, abstract, or reference in a third source.

Instituting detailed citations as a disciplinary norm could only strengthen our field by encouraging writers to be more precise and by allowing readers and researchers to engage more deeply with our scholarship. A general citation may satisfy the “acknowledging past works” criterion; however, when used to support arguments, it may hamper replication and therefore fail to address current transparency expectations. Unfortunately, detailed citations are still not part of the transparency revolution in political science, as discussed in the next section.

STATE OF THE DISCIPLINE: GENERAL CITATIONS AS POLITICAL SCIENCE CONVENTION

More than a decade ago, Moravcsik (Reference Moravcsik2010, 30) lamented that “Current practice in some areas of political science permits citations to be imprecise, vague, and secondary. […] It is still acceptable, even common, for citations to lack page numbers or any other reference to any specific elements.” Trachtenberg (Reference Trachtenberg2015, 13), who originally was trained as an historian, was equally critical toward political scientists, who often cite “a long list of books and articles, without including particular page numbers in those texts that a reader could go to see whether they provided real support for the point in question.” An empirical analysis of recent political science publications shows that these reprimands bear repeating today.

To illustrate the extent to which these observations remain true, I created a database that surveys contemporary citation practices in political science. I began by gathering all author–date citations (N = 26,276) in articles that were published throughout 2019 (N = 256) in five leading political science and international relations journals. I manually identified and coded all of the citations in these journals to ascertain how frequently detailed versus general citations were used. The online appendix includes the full database and explains how citations were extracted and variables coded (Grossman Reference Grossman2021).Footnote ³

The analysis demonstrates that almost 90% of all surveyed citations were general. Table 1 shows that slightly more than 10% of all citations published in 2019 in these journals included identifying details beyond the author’s name and year of publication. In 38 articles (14.8%), not a single reference was detailed; another 70 articles (27.3%) had three or fewer detailed references.

Table 1

Detailed Citations, by Journal

Note: Confidence intervals (CI) are in parentheses.

The low number of detailed citations is a cause for concern. Even if we accept that certain general citations are excusable (if not always recommended), it is difficult to believe that nine out of 10 citations in an article refer to an idea that cannot be located within the cited text. The desirable quantity of precise citations is not “one-size-fits-all” because it varies according to the function of the references and the nature of cited sources in every paper. However, we should generally strive for as many location identifiers as possible (Donovan Reference Donovan2008, 75). Using the best-performing journal in the sample as the baseline, it seems that having at least 20% detailed citations should not be an insurmountable goal; however, all of the other journals indicate considerably lower numbers.

This variation among journals in the sample suggests that a journal’s style preferences or epistemological traditions (Connors Reference Connors1999, 240–41) may affect the share of detailed citations. However, there also is notable variation among articles published in the same journal. One possible explanation for this variation is related to the number of words, citations, and referenced items in a manuscript. Because fewer identifiers mean fewer words—and considering journals’ word limits—authors may sacrifice references or parts of them to allow space for other elements (Gerring and Cojocaru Reference Gerring and Cojocaru2016, 7). Likewise, a higher number of citations in an article may indicate authors who are preoccupied with acknowledging their sources; in turn, this emphasis on transparency may translate into more precise referencing.

To test these assumptions, I used a linear regression model. Specifically, I examined the relationship between the share of detailed citations (out of all the citations in an article) and (1) the article’s total length; (2) its length relative to the journal’s official word limit (i.e., the extent to which the article conforms to or exceeds this limit); (3) its citations-per-word ratio; and (4) the length of the list of references relative to the article’s total length. I also tested an additional variable: (5) the first author’s affiliation with a specific country’s institution, which may indicate that detailed citations are affected by regional- or country-specific conventions.

As shown in table 2, the overall length of an article and especially its citation-per-word ratio have a significant effect on the share of detailed citations. In other words, the longer an article is and the more citations it has, the more detailed its citations are likely to be, ceteris paribus. This finding may suggest that authors who generally write in more detail and are careful to substantiate their arguments with references tend to cite more precisely.

Table 2

Possible Determinants of Detailed Citations

Notes: Standard errors are in parentheses. ***p<0.01, **p<0.05, *p<0.1.

However, when the list of references occupied a larger share of the article’s real estate, the number of detailed citations tended to decrease. This finding may reflect a quantity/quality bargain, in which authors perceive citing more sources to be more rewarding than citing them in detail. Because the size of reference lists in the social sciences is constantly increasing (Nicolaisen and Frandsen Reference Nicolaisen and Frandsen2021, 263), the share of detailed citations may decrease even more.

Nonetheless, the extent to which an article complied with a journal’s word limit was only weakly associated with the share of precise citations. Such a weak relationship may suggest that whereas authors who write longer articles cite more precisely, they have a more pressing reason to exceed word limits, such as fleshing out their argument or citing additional sources. Finally, the authors’ country of affiliation was not found to be significantly related to the share of precise citations, which implies that citing imprecisely is not a country-specific norm but rather a general one.Footnote ⁴

Further empirical analyses of citation practices using different automated or manual coding methods can unveil additional causes for the abundance of general citations. That the journals with the lowest share of detailed citations in my sample also are those that tend to publish more quantitative research suggests a possible relationship between the methodology used in an article and its share of detailed citations. This relationship can be explored by analyzing more qualitatively oriented journals, which are underrepresented in my sample.

Moreover, counting the share of direct quotations among all of the detailed citations in an article may help us to better understand the norm of using general citations for paraphrasing. Controlling for the location of general citations within an article can indicate how many are found in literature reviews—which, as noted previously, may sometimes be less detrimental to transparency than general citations in other contexts. Relatedly, books—more than articles—usually present several arguments, which require more specific referencing (Burbules Reference Burbules2015, 720). Distinguishing between references to books and articles may reveal whether authors actually cite the former more precisely than the latter.

HOW TO INCREASE DETAILED CITATIONS IN POLITICAL SCIENCE

The widespread adoption of transparency practices by political scientists indicates the discipline’s collective commitment to rigor and accuracy. Building on these norms and emphasizing the potential contribution and low costs of detailed citations can precipitate their spread within the discipline. To attain this goal, we should consider ways to incentivize those who cite precisely and then facilitate their work. The following suggestions are a starting point for this discussion.

First, as my sample suggests, the longer an article is and the more citations it contains, the more detailed citations it is likely to boast. Relaxing or expanding word limits and permitting longer articles, therefore, may increase the number of such citations. As Gerring and Cojocaru (Reference Gerring and Cojocaru2016, 5–6) proposed, an article “should be as long as it needs to be.” Editors and reviewers should decide, on a per-article basis, whether a piece should be shortened or expanded. Because most articles today are published and read electronically, allowing longer articles would not entail significant printing costs and may motivate authors to include in their work more citations in general and more precise citations in particular.

Another possible approach is to exclude citations and lists of references from an article’s word count (Gaikwad, Herrera, and Mickey Reference Gaikwad, Herrera and Mickey2019, 14). As my sample shows, there may be a tradeoff between the size of reference lists and the share of detailed citations. Permitting longer reference lists may eliminate this tradeoff. However, allowing more words or references also may result in citation inflation, name dropping, excessive self-citation, and the strategic citation of every potential referee. Therefore, these methods should be approached with caution.

Regardless of whether we allow longer articles, the transparent citation of secondary sources is not likely to become commonplace unless journals attach greater importance to this issue when considering manuscripts for publication (Henige Reference Henige2006, 106). Among other measures, journals can specifically stipulate that detailed citations are preferred over general citations; request reviewers to evaluate the transparency of citations in manuscripts; and consider this issue when deciding whether a manuscript is ready for publication.

Journals also may substitute in-text citations with references in footnotes (Trachtenberg Reference Trachtenberg2015, 14), in which more details can be included without “bulking up the parenthetical citation” (Connors Reference Connors1999, 239). In my database, the two journals that use footnotes instead of in-text citations have the highest share of detailed citations. Tellingly, World Politics, whose transparency of citations far surpasses those of other journals in the sample, is also the only journal whose instructions for reviewers emphasize “including page references for published work.”Footnote ⁵

Finally, promoting the value and payoff of detailed citations may start at an earlier phase (Leatham Reference Leatham2015, 267). Methods and writing courses in political science departments can underscore this topic as part of the general discussion of replication and reproducibility to strengthen transparency norms among the next generation of scholars.

CONCLUSION: WHITHER DETAILED CITATIONS?

This article highlights the political (and social) science norm of citing secondary sources by providing only general information, and it proposes ways to make these citations more useful, specific, and transparent. My proposal centers on calling for more precise citations in scholarly writing by referring to specific locations within sources as the new norm in political science. However, referencing pages—which currently is the most common method of citing precisely—may be feasible only for the short term.

In the information age, an increasing number of secondary sources, including electronic books, are unpaginated or have page numbers that may change with time or across reading platforms and file types (Baron Reference Baron2015, 33–34). Articles in websites, blogs, and online magazines seldom have location identifiers. Scientific articles are published online long before they are assigned final page numbers. Authors consulting an article’s HTML version are incapable of referring to pagination. Audiobooks and podcasts may pose an entire new set of challenges. Thus, as time passes, references to hard-copy page numbers are gradually becoming obsolete.

These trends should not exempt us from citing transparently. Rather, we should discuss and devise new standards for establishing location. Fortunately, the shift toward digital sources, which makes it easier to find specific parts of a source using digital searches, also makes it easier to cite these parts. One possible avenue is to refer to section headings or to quote several words from the original text so that readers can digitally find the cited passage (see Dunleavy, Park, and Taylor Reference Dunleavy, Park and Taylor2018, “Reusing materials from this book”).

Another solution is to use hyperlinked references leading to the exact location in the source where the cited information occurs. Active citations, which are “linked via hypertext to a reproduction or transcription of some part of the source” (Moravcsik Reference Moravcsik2010, 31), are progressively becoming helpful as we shift to electronic reading. Online annotation tools display the cited excerpts as an additional layer alongside the passages in the paper that cite or paraphrase them, which saves readers from having to find and access the cited source (Karcher and Weber Reference Karcher and Weber2019, 2–4). Digital tools that enable transcribing, annotating, and sharing excerpts of audio sources (e.g., Papandrew Reference Papandrew2020) may facilitate the precise citation of nontextual materials.

Regardless of the methods we choose, we can and should advance more transparent citation norms. That the majority of citations in political science are general is an anomaly that hardly can be justified in light of our otherwise methodologically rigorous discipline. Changing these norms will depend on the goodwill of editors, reviewers, and teachers as well as their capacity to encourage and incentivize scholars and students who use higher transparency standards. We can only benefit from such a change.