Supplementing the text: the role of appendices in academic papers

Some published academic papers include one or more appendices – additions to the end of the manuscript offering supplementary information beyond that included in the main text. With the move towards on-line publication, publishers of academic journals often now offer to provide supplementary material on-line that is not included in the published paper, but is linked to it through the journal website. (It is also possible to link to data sets that are not hosted by the publishers themselves, but have been uploaded to a public repository.) This can, for example, allow the publisher to make available audio-visual material that is not in a suitable format to include in a printed journal article. This also allows sometimes lengthy supplementary information to be provided to readers, but without including it as appendices to the paper itself. That can be important for a journal with hard copy as production and distribution (postage) costs are linked to page count – so including lengthy supplementary material as appendices in a printed journal means either a thicker and heavier issue – and so a substantive increase in costs – or reducing the number of articles included in an issue.

Appendices have advantages over separate supplementary documents

As Chemistry Education Research and Practice (CERP) is an on-line journal, increasing the length of downloaded papers by including appendices does not significantly increase production costs, and has no implications for the inclusion of accepted articles in issues. With the exception of theme issues, CERP normally includes in each issue all accepted articles where the production process has been completed (which usually means those already published online as accepted articles) regardless of the page length of the articles concerned.

The editorial policy at CERP is to avoid publishing discrete supplementary materials of a form that could be included in the article file (text, diagrams, tables, etc.), but rather to include such supplementary information as appendices, annexes added to the end of the article text. This has an advantage for readers working with downloaded articles, as if they find something of interest that is referred to in the main text as being available in the appendices, it can be immediately accessed in the same document file – rather than having to go on-line and download a separate file. This means the appendix will be part of the same document as the main text and so share its DOI (Digital Object Identifier, the unique reference number that allows on-line materials to be readily located even if their location – URL – changes). If preferred, the downloaded file could be electronically duplicated by a reader so that the appendix can be displayed alongside the relevant part of the main text. (Note, however, that downloaded files should not be used to make copies of articles for other readers – see Appendix 1.)

Including material in the main article file also makes it clearly part of the ‘version of record’ (VOR) of an article. The VOR is the officially published version that represents ‘the’ journal article, rather than the author's manuscript version submitted for peer review or an accepted version that has not yet been subjected to final formatting and proofing. The VOR of a CERP article will include the main manuscript text (often interspersed with figures, tables, etc.), references, and possibly one or more appendices. The appendices are clearly part of the published article, and so part of the intellectual property of the author that has been licensed to the publisher. So if another scholar wishes to refer to something in an article's appendix it is clear they should cite the article to which the appendix belongs – something the author of the article is likely to appreciate given that it has been suggested that “supplemental materials are typically not cited on their own” (Feeney, 2010, p. 29).

Appendices can make articles longer, but when used appropriately this is not a problem for readers. The main text of the article should be self contained so it can be fully understood without referring to the appendices: “all key information should be included in the main text of the manuscript” (Taber et al., 2014, p. 318). Appendices therefore contain relevant, but subsidiary information: a reader may wish to turn to the appendices when these are referred to by the author, but could instead simply choose to carry on reading the main text of the article without taking a detour to the appendices. For many readers, appendices will offer technical material that is of less interest than the main text, but the assumption is that appendices will contain material that will be seen of interest by some readers.

What should be appended?

These guidelines seem clear in principle, but – in practice – decisions as to what needs to be included in the main text of a research article, what would be better appended, or what should not be included at all, are always somewhat a matter for professional judgement. Journal articles need to reflect a balance between two important considerations.

An article in a research journal needs to offer an argument, supported by evidence, for some new contribution to ‘public knowledge’ (a notion which can be considered problematic, but remains useful – see Appendix 2). That is, published articles are considered to make knowledge claims that go beyond the existing canon of (previously) published scholarship. The article needs to therefore include sufficient text to make the argument clearly and to set out the evidential basis for the claim. However, articles should also be succinct as possible to support the reader in following the flow of the argument, and avoiding distractions and diversions. (This is especially important in an international journal where many readers will be accessing the text in an additional language, and indeed many authors will themselves not be writing in their native language.) Sometimes readers access papers for more specific, and so more limited, reasons than to read the entire study (perhaps to learn about the application of a particular research technique) – or at least choose to seek out particular information (the context of the research; the conclusions reached; the research design employed) as the basis for deciding whether they wish to commit time to reading the full paper. As there is nothing to stop any reader skimming an article, or skipping sections they are less interested in, material should not be relegated to an appendix simply to reduce the reading load of a study: the assumption is that material is placed in an appendix because it is considered subsidiary to the basic narrative being offered in the article and so is not necessary to the main thread of the argument made. (The appendices included below are meant to model this structure, although their contents might be considered more suitable for footnotes than appendices.)

Data collected is only as good as data collection instrumentation – so readers may be interested in seeing examples of questionnaires, assessment items, observation schedules, or – when using some forms of research methodology – the development of data coding protocols or the like. Some studies depend upon statistical analyses of data that can be presented with brevity without compromising the quality or strength of the argument. Other studies depend upon the interpretation of extensive qualitative data sets, and may require the presentation of a good deal of supportive quotations from the data to fully illustrate and exemplify the analysis reported. This can lead to a tension between the imperatives for brevity and thoroughness (Pope and Denicolo, 1986). A solution here may be that results and limited illustratory examples are included in a main text, but appendices used to supplement this. So, for example, in an interpretative study where data analysis leads to four or five key themes the results section of the paper reporting the study may be limited to offering descriptions of the themes supported by limited quotes from the data, but an appendix might be used to illustrate the extent to which the data collected offers extensive evidence to support one of the themes reported.

Appendices may contain material related to any aspect of the work reported. The APA Publication manual (American Psychological Association, 2009) used as a reference by many journals in education and the social sciences gives the example of listing in an appendix the canon of studies accessed in a meta-analysis where not all of the studies included in the study will appear in the reference list (which should only contain works directly cited in the article). An article on preparing studies for publication in chemistry education suggests that “examples of course materials developed or evaluated in a study, examples of how data were analysed (e.g. factor analysis), examples of interview transcripts or observation notes to illustrate how the data were coded and analysed, and other materials of these kinds may sometimes be considered suitable for including as appendices (Taber et al., 2014, p. 318).

Using appendices

Appendices appear at the end of the manuscript after the main text. Bibliographic details of any works cited in an appendix need to be included in the article’s list of references. Where there is one Appendix, this should clearly be labelled as ‘Appendix’ and ideally should be given an illustrative name (e.g. ‘Appendix: Codes developed during open-coding of interviews’). Where there are several appendices these should be numbered (e.g. ‘Appendix 1: Instructional text given to the intervention group’; ‘Appendix 2: Instructional text given to the comparison group’). Each appendix should be discrete and self-contained. It would not be appropriate, for example, to include in the same appendix demographic details of a sample population and also information on how questionnaire responses were coded. If the heading needed to describe the contents of an appendix is vague (e.g. ‘Appendix: Additional information on the study’) or compound (e.g. ‘Appendix: Interview schedule and observation schedule and timetable for data collection’) it is likely the contents should be divided into several discrete components – each a separate appendix to the study.

Appendices need to be called out at appropriate points in the text: that is, the author must indicate to readers somewhere in the main text that there is relevant supplementary information in an appendix. If there is no suitable point in the main narrative of the study to refer a reader to an appendix then it has no place in the article and should be omitted (in the same way that any table or figure appearing in an article has to be called out in the text). Moreover, appendices should be numbered according to when they are first called out – as is the practice with figures and tables. (That is a reader following the main text should not meet ‘see Appendix 2’ before they have met ‘see Appendix 1’.)

It may well be that there is good reason to refer to the same appendix at several points in the main text, and so subsequent references to an appendix may be out of order. (That is, a reader may be referred to Appendix 1, then to Appendix 2, then later to Appendix 1 again.) If the different references to the same appendix are pointing to different information within the appendix, this may – but would not necessarily – indicate that the appendix would be better split. So for example, an appendix could present background information in the form of biographical sketches of three teachers participating in a study, and it may be open to question whether these are best presented in the same appendix, or would better be split into different appendices. There are nuanced judgements to be made about such matters. If the study was a multiple case study of the professional work of the three teachers (so that the data from each teacher was analysed separately, and results reported discretely in the findings before a cross-case discussion) then separate appendices may seem more in keeping with the methodological approach, whereas if the data relating to the three teachers had been analysed together and fed into a unitary report of findings then placing the sketches in different appendices seems less appropriate – even if this means the reader is pointed to the same appendix at several points in the main text where the different teachers' backgrounds may be considered to be pertinent to the findings.

Reviewers and editors should be sensitive to the need for such professional judgement in deciding the nature of appendices included in an article. It may be a matter of professional opinion whether some statistical analyses are peripheral to the main theme of a study, and are better relegated to an appendix; or whether some of the ‘thick description’ needed to give the required context for naturalistic case studies can be considered less relevant background and does not need to appear in the main text of a manuscript. Generally authors' choices should be respected, and certainly article length is not by itself a good reason to ask authors to move information from the main body of the report of a study. However, authors should be sensitive to the general points made above, and ensure when preparing their manuscripts that any appendices included are relevant (but not central) to the study; discrete; suitably titled and correctly sequenced.

Appendix 1. Rationale for the restriction on copying and sharing downloaded journal articles

Copies of articles downloaded from journals such as CERP should not be sent to other potential readers, who should instead be sent a link to the article on-line so they can download their own copy. Usually when downloading an article from a journal site the person downloading has agreed to conditions of use which specify the file may not be copied for anyone else. Where a reader has a personal subscription or has paid for an individual article, they have usually only paid for personal use. Even when a reader accesses an article through a library subscription they are not usually allowed to pass copies to other readers (except as part of formally registered study packs). Institutional librarians are often very concerned about articles being shared around after download (even if this might seem harmless when the sharing is only among others entitled to access the journals within an institution) both because of the breech of the legal basis on which articles are provided and because institutional libraries need to accurately monitor journal use when justifying subscriptions and making decisions on which journal subscriptions to renew.

Even though CERP is a free to access journal, downloaded articles are for personal use, and respecting this supports the journal by giving the publisher (the Royal Society of Chemistry, RSC) a realistic sense of the readership of published articles. The RSC is able to offer access to CERP free to all readers due to generous support of the journal by the RSC's Education Division which recognises the educational value of making CERP available to readers such as school teachers, and researchers and lecturers in institutions in developing countries with limited library budgets. Part of the justification for committing resources to CERP in this way is that CERP reaches a large and diverse readership, and the main measure of that is the number of article downloads from the journal site. Readers will therefore help the journal, and support the argument for keeping it free to access, by notifying colleagues of articles likely to be of interest, and sending them links to the articles on-line, but not sending copies of the downloaded files they have been provided for their own personal use.

Appendix 2. The notion of public knowledge in relation to journal articles

The term public knowledge, as used in the main text, is intended to be contrasted with the notion of personal knowledge held by individuals. Public knowledge might be seen as the account found in authoritative texts, or as shared by members of the relevant public community – so in the case of chemistry education, scholars in chemistry education.

In science it is not enough to discover something (i.e. develop new personal knowledge), rather it must be shared with the scientific community through publication. Research journals are the main means of communication of scientific results, and so of adding to public knowledge in scientific fields. Prestigious journals normally have a criterion for accepting an article for publication that it offers some significant new knowledge not previously published, and so adds to public knowledge. So when a researcher undertakes some scholarship or empirical work that leads them to develop new (personal) knowledge that they believe is not already known in the community (in effect, published) they write up for a research journal. If the process of peer review leads to an evaluation that the knowledge claims being made are both sound and represent an original contribution then the article is published and contributes to an expanded public knowledge.

When discussed in these terms the idea of public knowledge seems quite straightforward, and the account in the previous paragraph would probably seem familiar and reasonable to most readers of this editorial. However, the notion of public knowledge may seem quite problematic when viewed from certain perspectives on the nature of knowledge itself. So, for example, a personal constructivist notion of knowledge might require knowledge to have a (personal) knower, and consider that knowledge is a property of mind, and furthermore a mental construction produced by perceptual-cognitive processes (Taber, 2013). From this perspective journal articles do not contain knowledge as such (as knowledge only exists in minds), but rather representations of their authors' knowledge. These representations then have to be interpreted by readers as material for producing their own personal knowledge which will necessary depend upon the interpretive resources each reader has available, and so will potentially be idiosyncratic. A judgement of whether an article submitted for publication offers something not already part of public knowledge is from this standpoint inevitably a matter of interpretation and includes an inherent subjective element (as no two journal reviewers, no matter how expert, will have the ‘same’ notion of the current state of public knowledge in a field).

Academic journals may be the tools for making knowledge publicly accessible, but ‘public knowledge’ (if such an idea is given credence) is distributed (in the sense that it is spread over all the people said to know that knowledge) and manifold (in the sense that there is not a single unique authorised version of the knowledge copied around, but rather a potentially large set of unique personal takes on what that knowledge is). From such a perspective public knowledge as something uniform and shared – completely agreed upon – in a community is a fiction, albeit a useful fiction that can act as a referent when discussing academic norms and processes (see Taber, 2013, chapter 10).

References

1. American Psychological Association, (2009), Publication Manual of the American Psychological Association, 6th edn, Washington DC: American Psychological Association.
2. Feeney P., (2010), DOIs for Journals: Linking and beyond, Information Standards Quarterly, 22(3), 27–31.
3. Pope M. L. and Denicolo P., (1986), Intuitive theories – a researcher's dilemma: some practical methodological implications, Brit. Educ. Res. J., 12(2), 153–166.
4. Taber K. S., (2013), Modelling Learners and Learning in Science Education: Developing representations of concepts, conceptual structure and conceptual change to inform teaching and research, Dordrecht: Springer.
5. Taber K. S., Towns M. H. and Treagust D. F., (2014), Preparing chemistry education research manuscripts for publication, in Bunce D. M. and Cole R. S. (ed.), Tools of Chemical Education Research, Washington DC: American Chemical Society, pp. 299–332.

---