Writes of Passage: Writing an Empirical Journal Article

LYNN WHITEUniversity of Nebraska—LincolnWrites of Passage: Writing anEmpirical Journal ArticleThis article provides advice about preparingresearch reports for submission to professionaljournals in general and Journal of Marriage andFamily in particular. In addition to workingthrough all the major parts of a research paper,I provide some general advice about writing,editing, and revising. The article is intended tohelp new professionals improve the quality oftheir journal submissions and the likelihood ofsuccessful publication.Writing research articles for professional journalsis an art requiring good research skills, aclear sense of problem, and strong writing andediting skills. Assuming that years of graduateschool have provided good research skills, Ifocus on the other requirements of writing aresearch article. My advice reflects the issues Imost often raise when I review articles and 30years of experience writing (and revising) researcharticles. I review guidelines for the majorsections of the typical empirical research reportand conclude with some suggestions about writingprofessionally. The emphasis is on writingfor Journal of Marriage and Family (JMF), butthe general principles apply across journals andsubstantive areas.WORKING THROUGH A RESEARCH PAPERThe format for a research paper is not set instone. Each research problem is different, andDepartment of Sociology, University of Nebraska—Lincoln,Lincoln, NE 68588-0324 (lwhite3@unl.edu).Key Words: research, theory, writing.the organization of the paper will depend onwhether it is exploratory research rather than theorytesting. In addition, authors have some latitudein developing a personal style. Generally,however, each article needs an introduction, aliterature review, a statement of the problem,description of method, results, and conclusion.The organization of the piece, the titles of varioussections, and the relative weight of these sectionsvary from paper to paper and from journalto journal, but some general guidelines apply toreports of qualitative and quantitative research.AbstractAn abstract should summarize your study. In afew short sentences, it should state the researchhypothesis, the sample, sample size, data used,and the findings. A starting sentence such as‘‘Using data from a national sample of n womeninterviewed by telephone in 2002, we examinethe relationship between x and y’’ will allowyou to squeeze a lot of information into a fewwords. In a bare-bones fashion, without hyperboleor exaggeration, state the findings of thestudy. Examine prior issues of your target journalfor abstract style and be sure to comply withthe maximum length specified by the journal(120 words for JMF).IntroductionThe introduction is critical to capturing thereader’s attention and setting the tone for thepaper. In approximately a single page, it shouldspecify the research question, the data to beused, and the strengths of the design, and itJournal of Marriage and Family 67 (November 2005): 791–798 791

792 Journal of Marriage and Familyshould answer the ‘‘so what?’’ question. Toomany authors wait until p. 13 to tell the readerwhether they have 20 or 10,000 cases. It isimportant to be specific in the introduction aboutwhat you are going to do. Implying that you aregoing to do the definitive test of exchange theoryin the introduction will cause trouble if theactual study is much narrower. By the end ofthe second paragraph, the reader should have anaccurate idea of what is coming. It is not a placefor extensive citations or literature review.Literature ReviewThis section includes a review of both the theoreticalor conceptual framework and of the priorempirical literature relevant to the topic. Thismaterial helps the reader understand the problemand how it fits with prior work. The relativeweight of these two aspects varies from paper topaper, but you generally do need both.It helps to think of the literature review asa funnel: You begin with a general overview ofrelevant theoretical or conceptual ideas thenmove through prior empirical work to a discussionof your paper. Thus, the empirical literaturereview is usually narrower than the theoreticalframework. The literature review should leadinexorably to the research questions or statementof the problem section. By the time you get tothe end of the literature review, the researchquestions should be obvious to the reader becausethey have been the focus of the review.Unlike a dissertation, the literature review ina journal article sets up the particular researchproblem instead of providing a comprehensivereview. You need neither to describe the evolutionof the theoretical perspective you are usingnor to review all the work that has ever beendone on topics related to yours. You do notneed to be exhaustive. There is no extra creditfor listing five citations after every statement orfor dragging Bourdieu into the theory. You onlyneed to cite an article once, not each time it isrelevant to a sentence in your review. Whenyou are on well trodden ground (e.g., genderand housework), cite one or two major reviewsand focus only on what is critical to your paper.Unless you are giving the material a new spin,you do not need to review prior conceptualizing.Important criteria are relevance and balance.Make sure you cover both sides of thestory, for example, work that finds no relationshipas well as work that does.The literature review should support thepaper that actually develops rather than thepaper that was planned. If your data lead youdown an unexpected trail, it may be necessaryto examine new theoretical frameworks andnew literature to try to make sense of the findings.If your original review does not preparethe reader for the analysis actually presented, itshould be omitted—no matter how elegant orwell written.Because so much material goes into a literaturereview, a clear structure is essential. It maybe helpful to introduce it with a statement suchas ‘‘three bodies of research bear on this question:x, y, and z.’’ Then review each in turn witha heading. (Because they provide structure tothe material, use headings freely.)Statement of the ProblemThis section also might be called research questionsor hypotheses, but it is the crux of thepaper. It acts as the link between the literaturereview and the method section by laying out theissues to be examined, introducing the studydesign, and defending the design’s appropriatenessfor the problem. It is both the conclusionof the literature review and the introduction tomethodology. It should not introduce new ideasbut should be a summary of issues you haveraised in the literature review and a transition toyour own study.Hypotheses or research questions should befairly explicit, though they need not be statedformally. The details of this section will dependon the type of research. For quantitative analysis,this should include stating the basic researchquestions, explaining any expectations aboutmediating or moderating influences, and explainingbriefly what control variables will beincluded. For qualitative work, this section willlay out the questions to be explored.The details of your study design go in themethod section (below), but why you madeyour major design decisions goes in the statementof the problem. This section is where youintroduce the strengths and weaknesses of yourdesign and provide preliminary justification foryour decisions. The aim is not only to introduceyour design but also to help your reader reach apositive judgment about your design decisions.Especially if some aspect of your study designis going to raise eyebrows (e.g., you includeonly women or measurement of a key concept

Writing an Empirical Journal Article 793is unusual), introduce the issue and explain ithere. Explain the likely consequences of yourdecision and suggest what the reader shouldmake of this. Thus, you might say,orTo answer these questions, we used the ABCDdata set from 1997. Although this study lacksmeasures of E and F, it does include unique dataon G and H—the focus of this paper.To answer these questions, we conducted in-depthinterviews with 12 welfare mothers living in theSan Diego metropolitan area. Although the sampleis small and geographically restricted, it ....MethodThe method section provides the bare bones ofthe sample, measures used, a statement aboutmissing values, and an analysis plan. In somejournals (e.g., in psychology), this material isprinted in a tiny font in standardized blocks.Although JMF does not do this, the principle isthe same: This is the place for a terse statementof what you did rather than a discussion of whyyou did it. You need to provide enough informationso that others with access to the samedata (an increasingly likely phenomenon) canreplicate your work and get the same results.Data. This section should include sample size,how respondents were selected or recruited, andfrom what population respondents were drawn.Information should be provided about levels ofnonresponse or refusals, description of patternsof nonresponse, any oversampling procedures,and a brief discussion of any biases introducedbecause of the design or nonparticipation rates.In the case of a subsample drawn from a largersample, the criteria for selection and the numberomitted should be specified. Systematic assessmentand reporting of selection bias is necessary.For example, are these families selectiveof those with better parent-child relationships orchildren with higher attendance? The statisticalsophistication of this report will depend onmethod, but all studies need such a section.Either at the end of the data section or at thebeginning of results, provide some descriptionof your sample so the reader understands thecontext of your data. For example, are thesemiddle-aged, middle-class, married, White individualswith adolescent children?Measures. This is a place for simple declarativestatements with minimum justification. Specifythe source of the item or scale, exact questionwording, the answer categories, and how theywere coded. For example, it is sufficient to state,Following Smith and Jones (2002), marital happinessis measured by 5 statements in a 4 ¼strongly agree to 1 ¼ strongly disagree formatwith don’t know coded 2.5. The statements were,I am happy with my husband as a ‘‘friend to dothings with,’’ ‘‘sexual partner,’’ ...For scales, an indicator of reliability and perhapsunidimensionality is important. If yourmeasures are extensively used, you can providea citation and omit extensive information aboutspecific properties such as factor analysis.For qualitative reports, this section is lessterse and stylized, but the information providedis similar: How were the concepts coded, whatreasons do we have for judging the results reliableand valid, and what method was used toanalyze the data (including software, if any)?Missing values. The bane of empirical studies ismissing values. Standards for dealing withmissing data in quantitative studies are increasinglyhigh, but the critical step in all research isa description of the extent of the problem, howyou handled it, and an assessment of the consequencesfor your findings. If the sample has10,000 cases, but you are analyzing only 7,800,we need to know why. If you use an imputationprocess, describe this. The issue is not restrictedto analysis of secondary data sets. Whether youdrop 3 of 25 observations or 2,200 out of10,000 cases, we need to know who was omitted,why, and the consequences of this omissionfor the credibility of the research.Analytic approach. If you are using a novel statisticaltechnique, it is useful to describe thetechnique briefly and give some references. Ifthe method has been used recently in the journal,then this probably should not be more thana paragraph.ResultsThe results section is the most technical part ofa research article, yet it still requires strong writingskills to hold it together. The basic principleis to tell a story and develop the plot as yougo along. Start with basic descriptive findings

794 Journal of Marriage and Family(orient the reader and describe the major characters),test the basic hypothesis, and then elaborateon it. Throughout, you need to keep thereader’s eye on the central thread and explainhow the next procedure will help you follow thatthread. This means reminding the reader of thehypotheses and including summaries and transitionsas you move through this section. Whenyou have completed testing a specific hypothesis,briefly tell the reader whether your suppositionwas supported and what the next step is inmaking sense of the findings. Although the discussionand the integration of the findings intothe literature are reserved for the end of thepaper, summarize as you go along.A major decision that confronts qualitative andquantitative researchers is deciding how much toinclude and where to stop. The best guide to thisshould be your introduction and statement of theproblem, which should have identified the majorquestion and story line. This should help youavoid getting tangled up in interesting (but notcentral) issues such as relationships with controlvariables. If it is not set up in your theory section,it is probably a different paper.Some pointers for quantitative reports.Although specifics will depend on whether youare using regression or structural equations,some problem areas that come up in many reportsare worth noting:ddSignificant or meaningful? In any data set,but especially those with very large samples,substantively unimportant differences may bestatistically significant at the conventional .05level. You can solve some of this problemmechanically by using a more demandingprobability level. At the other end of the spectrum,a very small sample may lack the powerto find that a substantively important effect isstatistically significant. Regardless of samplesize, it is important to consider the substantiveimportance of the findings instead ofrelying on statistical significance. For example,you might calculate effect sizes or reportpredicted scores for those with low and highvalues of key independent variables. Dealand Anderson (1995) provide a good discussionof this and a variety of issues relating toreporting research results.Interaction terms. Make sure that your interactionterms were hypothesized in the statementof the problem and have a reasonabledjustification in theory or prior work. Too ofteninteractions appear to be ad hoc data dredging.Report how many tests you ran (andhow), and how many were significant.Make your text comprehensible to readersindependent of the tables. Not all readers willwant to study the tables, and the report shouldread clearly by itself.Special pointers for qualitative reports. Becausemy own research is quantitative and I amrarely asked to review qualitative work, I referthe reader to three useful articles that will helpwith qualitative reports. In the 1995 specialissue on Paradigms and Designs, Ambert,Adler, Adler, and Detzner (1995) discuss commonerrors in reporting that make qualitative reportshard to evaluate and make reviewers lessinclined to trust them. More recently, Belgrave,Zablotsky, and Guadagno (2002) provide adviceto qualitative researchers who are writingfor (or at least being reviewed by) quantitativeresearchers. Finally, Sarah Matthews’s article inthis issue provides tips on organizing and writingqualitative reports.Summary, Discussion, ConclusionAlthough headings may vary, articles need toend with a summary, a discussion, and a conclusion.The summary does just that: It summarizesthe key findings. This summary section (one ortwo paragraphs) can serve as the basis for theabstract, and many readers will rely on it to getthe gist of your paper.The discussion section links the findings backto theory and prior empirical work, pointing outwhere your results support Smith or contradictJones and where you explore new ground. Theanswer to the ‘‘So what?’’ question raised in theintroduction should appear here. Explain howyour research advances the field and what weknow now that we did not know earlier. The discussionalso assesses the strengths and weaknessesof the study design: How confident canwe be of these findings? Where should we becautious? At this point, you should bring backthe issues you raised in your statement of theproblem, for example, sample restrictions, causalorder, and so on. You need to be honest, but youdo not need to make these issues the major pointof the section. The discussion often includes theubiquitous call for future research. Although it

796 Journal of Marriage and Familyclearly on the first page and should remaincentral and obvious throughout the paper.Although it should go without saying, experiencesuggests it bears repeating: The focusshould be the same in the abstract, the introduction,the literature review, the analysis, andthe conclusion.LengthJournal space is at a premium, and mostjournals have suggested page limits for theentire manuscript (JMF suggests a 30-pagelimit). This translates to something like 20–22pages of text, 4 pages of references, and fourtables. Within this general guideline, a complexpaper with a complex study design may require35 pages, and a narrowly focused paper maybe too long at 25 pages. Although it maybe painful, a succinct presentation will helpyou avoid extensive theory discussions, unnecessaryliterature reviews, and tangentialanalysis.If you are cutting down a larger report (e.g.,a dissertation), start writing from scratch ratherthan trying to save paragraphs or even sentences.Often, the article will be narrower in scopethan the dissertation, so the theory and literaturereview will need to be refocused. In addition,the terse format of a journal articlerequires an entirely different approach than themore discursive and comprehensive writing ofa dissertation.Have Others Read It and Heed Their AdviceWhen your manuscript is ready, ask two orthree colleagues, preferably those with publishingexperience, to read it. Ask them to writea review of it or at least to talk to you about it.Despite how dumb you think their commentsare, pay attention to them. If they did not understandsomething that you think is perfectlyclear, the reviewers probably will not either. Ifthey did not get the point, the reviewers are alsolikely to miss it. Even if you did explain it,obviously you did not explain it well enough.Your colleagues will undoubtedly be morepatient with you and nicer than the reviewers,and they may read it more carefully. Addressingyour colleagues’ critiques is probably just thetip of the iceberg, however, and you should notbe surprised if journal reviewers have differentor more extensive comments.Edit Your Own Work ThoughtfullyLearning to edit your own work is a hard-wonskill. Although there are many guide books, Icontinue to rely on Strunk, White, and Angell’s(2000) The Elements of Style (or any edition).Read it carefully and often until such rules as‘‘Use the active voice,’’ ‘‘Omit needless words,’’and ‘‘Avoid a succession of loose sentences’’ areingrained. The American Psychological Association’s(2001) Publication Manual is also a usefulreference.Subject/verb agreement and correct use ofcommas are simple but critical requirements.More than a few mistakes give the impressionof carelessness that can color the reader’sopinion about your research work. Rememberthat 80% of articles submitted to top journalssuch as JMF are rejected. You want to makeit clear in every way that you are in theother 20%.In addition to checking for technical correctness,scan your work for flabby writing andfor unnecessary paragraphs, sections, andwords. The classic topic sentence remainsa good idea, and a paragraph should hangtogether and cover a single topic. Look forcoherence within a paragraph and examine thetransition from one sentence, one paragraph tothe next.Follow the Guidelines Establishedby the JournalEach journal has guidelines for the text, headings,tables, figures, and references. These maybe included in the first or last issue of eachyear or, increasingly, on the journal’s Website. Following these guidelines carefully isimportant to putting your work in the top20%. In addition to reducing problems at theediting stage, it demonstrates a professionalapproach to writing, and it signals to the editorand reviewers that you have made a thoughtfuldecision about crafting this piece for this journal(as opposed to sending it to a half dozenjournals in succession without revision). JMFhas an unusually helpful Web site. Follow itslinks to advice on table guidelines, informationfor authors, APA style guidelines, and preparationof figures. In addition to these technicalmatters, it is often useful to scan recent issuesof your target journal carefully for style andcontent.

Writing an Empirical Journal Article 797Learn From Your MistakesLearn from editors rather than fight them. If youare lucky enough to have your work copyeditedby a professional or if one of your colleagues isa good copyeditor, do not just make the changesthey suggest, but think about why they madethe changes and make a note to yourself toavoid the error in the future.The same is true with rejected manuscripts.Although it is tempting just to file that fat envelopeof dense criticism because it is too painful toread, you really must consider carefully what youhave been criticized for and figure out how toavoid it in the future. Probably, these are seniorscholars who have given hours of their time toassess your work. If you can avoid defensiveness,their advice should be valuable. In addition, youmay be able to address the concerns and haveyour work accepted in another journal.Revising ManuscriptsSome papers are rejected outright. You may,however, get some version of a ‘‘we’ll be willingto look at it again if you revise it as suggested.’’These phrases may be hard to decode,and you should go over them with a more experiencedcolleague to figure out whether thismeans ‘‘Take care of these few things and publicationis practically guaranteed’’ or ‘‘We arewilling to look at a revision, but we do not havemuch faith in your ability to make all thesechanges and it hardly seems worth your effort.’’You can talk to the editor to help you figure outthis message.Even scholars with 10-page lists of publicationscan probably count on the fingers of onehand the number of manuscripts that have beenaccepted outright without a request for somerevision. This means that revising is a centralskill. Some general advice from someone whodoes not need all five fingers to count manuscriptsthat have been accepted outright: Thefirst reading of the reviews almost always ismore daunting than it really is. Give yourselfa few days to get over your initial dismay, andthen analyze the reviews and the editor’s lettersystematically. I make a photocopy of the reviews,and then put a number in the margin foreach concrete suggestion. After I do this, I considerwhether it is possible to make each changeand, if it is not possible, whether a strongerexplanation and more references will allow meto make a compelling argument for the way Idid it.My experience is that reviewers’ suggestionssubstantially improve my papers. Although somerevision requests are minor, editors seldomask for or are fooled by cosmetic revisions. Revisionsoften require a lot of work—perhapsrebuilding the data set, reanalyzing it, and refocusingthe literature review. The end result is usuallya much better paper. Occasionally, you maycontinue to believe that the way you did it firstwas just as good if not better than the reviewersuggests. In this case, you may have to decidewhether you would rather be right or be published,at least in this journal. Finally, althoughpersistence in revision often pays off, it is vital torecognize when the major critique centers ona fatal flaw that cannot be fixed (e.g., there is toolittle variance on the independent variable, thedata are too old). In this case, the most sensibleresponse is sometimes to chalk this up to experienceand move on to a more fruitful project.If you get a revise and resubmit that leanstoward the encouraging end of the continuum,do it now! Do not wait for the journal to changeeditors, the editor to change her mind, or someoneelse to publish it before you. Regardless ofwhether you decide to revise for the originaljournal or to submit it to another journal, do notput it off. The longer you wait, the harder it willbe to pick it up again, and the less likely youwill be to get to it.Responding to a revise and resubmit includestwo parts, of almost equal weight: the revisedmanuscript and the response to the editor andreviewers. Most new professionals have zerotraining on this aspect of research, so ask seniorcolleagues to let you see some of their responseletters before you start. The response should getthe same amount of careful attention as the revisionitself, so edit it carefully and have someoneelse read it before you send it. This responseshould not repeat all the material changed in therevised manuscript. Instead, it should summarizethe issues and give the page number wherethe new material is found.ConclusionWriting is a craft, and each paper you write willmake the next one easier to write. Nevertheless,even the most senior scholars are familiar withthe revise and resubmit letter, and occasionally,their papers are rejected out of hand. Sometimes,

798 Journal of Marriage and Familya paper one journal rejects will get accepted byanother journal (after revision, maybe even abetter journal), but we all have papers thatbecome permanent residents of the file drawer.It is not the experience of rejection that distinguishesthe successful from the unsuccessfulwriters, but the response to it.Happy writing.NOTEAlan Booth and Gay Kitson provided helpful commentsduring the preparation of this paper.REFERENCESAmbert, A.-M., Adler, P. A., Adler, P., & Detzner,D. F. (1995). Understanding and evaluating qualitativeresearch. Journal of Marriage and the Family,57, 879–893.American Psychological Association. (2001). Publicationmanual of the American PsychologicalAssociation. Washington, DC: Author.Belgrave, L. L., Zablotsky, D., & Guadagno, M. A.(2002). How do we talk to each other? Writingqualitative research for quantitative readers. QualitativeHealth Research, 12, 1427–1439.Deal, J. E., & Anderson, E. R. (1995). Reporting andinterpreting results in family research. Journal ofMarriage and the Family, 57, 1040–1048.Matthews, S. H. (2005). Crafting qualitative researcharticles on marriages and families. Journal of Marriageand Family, 67, 799–808.Strunk, W., Jr., White, E. G., & Angell, R. (2000).The elements of style. New York: Allyn & Bacon.