DocuMENTS Interviewing and observing are two data collection strategies designed to gather data that specifically address the research question. Documents, however, are usually produced for reasons other than the research at hand and therefore are not subject to the same limitations. The presence of documents does not intrude upon or alter the setting in ways that the presence of the investigator often does. Nor are documents dependent upon the whims of human beings whose cooperation is essential for collecting good data through interviews and observations. Docu- ments are, in fact, a ready-made source of data easily accessible to the imaginative and resourceful investigator. This chapter exam- ines the nature of documents. various types of documents, their use in qualitative research, and their limitations and strengths. The last section of the chapter presents a look at a relatively new type of documents and data-that which is obtained online. A number of terms are used to refer to sources of data in a study other than inten1ews or observations. I have chosen the term document as the umbrella term to refer to a wide range of ""Titten. visual, digital. and physical material relevant to the study at hand. Docum.ents, as the term is used in this chapter. also include what LeCompte and Preissle (1993) define as artifacts- "symbolic materials such as writing and signs and nonsymbolic materials such as tools and furnishings" (p. 216). Artifacts are "things" or objects in the environment differentiated from doc- uments that represent some form of communication (e.g., offi- cial records. newspapers. diaries). Documents include just about 140 Ql.'\1.11 A 11\'V. Rr '-I '\IH 11 <mything in cxistcnn: prior to Lhe research at baud. Comnwn do<:uruenb inclucl< ollicial ncorch, letters, newspaper accounL'i, poem'\, song.,, cot potaLc records, gmcrnment documenL<,, hi-.tor- accoun_l" diatics, and 1\0 on. Photographs, hlm. and \1deo can also be used a.., data sources, can phy..,i- cal evidence nt trace-. (Ltc, Webb, Campbell, Sch\,anz. & Sechrest, 2000). AlL hough thh chapter concentrates on wriuen tht> genet at dbcusston applies to all fonm of data not gathered through tnlct\'iews ot observations. TYPES OF Diiferem writer.. catcgmite documents in different wavs. Public records and personal documents are two common t;pes of doc- uments used in qualitative research. \Vhat Bogdan and Biklen (2007) call "popular culture documents" is a third type to be dis- cussed here, along with a fourth type-visual documems-which include films, videol>, and photography. VisuaJ documents int<:r- sect with popular culture, and even public records and personal doctm1ents can be visual in nature. so in reality the same docu- ment can be classit1ed in more than one way. Physical ltlatetial document-. 'iUCh as objt'Ct'l in the environment or changes in the physical setting arc not quite as common!) used as the orher bm nevt'rthcless are a potential source of data for the quai- Ltatl\e researcher. Moreover, documents can be generated b) the researcher for the purpose of tlw investig-c1tion. Pl'BLTC RECORDS Public record' are rlw o{fi(ial. ongoing records of a soriet\ \ activities. As Guba and Lincoln (1981) note, "The first and most important to anyone looking for official record-; is to presume that if an e\cnl happened, some record of il exists" (p. 253). Public document.., include actuarial records of births, deaths, and nmrriages, the U.S. census, police records, court transcripts, agency record'i, association manuals, program docu- menL'i. ma. ... s media, govcrnmtnt documents. and so on. Locating public record!> is limited only br the researcher's imagination and Auster ( 19H5), for <xample, demonstrates how \h,.,_INt. IJ,\1 \ HHIM [)u( l MI.N'IS 14} to conduct a study of changing -;ocial e'pcctation ... for lamily. career, gtnder roles. and -.cxual behavior through the data source of Girl SeoUL handbook-.. Youth org-c1nitation handbooks, o;he points out, "represent the intcr-.cction ol biography and {p. 359}, prO\iding an excellent data '>ource for studvmg changing social mores. for those interested in educational qut.'stions, there are numerous sources of public clocumcnts-di-.cussiom of educa- tional is.sues and bilb in the CongTI'HW1WI Rrrord; fedt'ral. state, and pnvate agencv report<;; individual program record'>; and the statistical database of the Cenrer for EducationaJ Statistics. Since many case sntdies arc at [he program level. it is particular!} important to seek out the paper tratl for what it can reveal about the program-"things that cannot be obs<.necl," things "that have taken place before the evaluation began. They may include private interchanges to which the educator would not otherwise be privy. They can reveal goals 01 decisions that might be oth- erwise unknmvn to Lhe evaluator'' (Patton, 2002. p. 293). ldeally this paper trail includes "all routine records on client.<,, all cor- respondence from and to program st.t1'. financial and budget records, organizmional rule.,, regulations, memoranda, charts. and any other official or unofficial dorumcnt!> generated by or lor the program" (p. 293) Such documents are \'aluable "not onh because ofwhat can be lc<lrnerl direct!) !rom Lhcm but also as stimulus for paths ot inquiry that can be pursued onl} through direct observation and intervicwng" (p. 294). Lf you ,.,.ere interested in studving the role of parent invohe- ment in a neighborhood school, for <:.'"ample, vou could look for public record documents 111 the form of the following: notices sent home to parent!>; memos between anrl among teacht'rs. staff, and the parents' association; formal policy stattment., regarding parent involvement; sehoul bulletin boards OL other displays turing aspects of parent involvement; ncwspapet and other medta coverage of activities featuring parent involvement; and any offi- cial records of parent attendance or pnscncc in the school. Olher sources of public infom1ation that are easily accessible but often overlooked include previou-. studies and data "'banks'' of information. However, in using thlse resources the re'iearcher has to rely on someone else's cle!Kription ann interpretation ul' 142 QtiAIHA II VI. Rt.W.AR! ' II dat.a rather thau Ll'lt th( raw data a,., a basis for anah,..,is. These as the) atl' C411ed. arc more common quantita- lJH! althougl! there has bt>cn "orne recent thinking as to how ttm strategy mtght apply ro qualitative studies. For large- scale ot research, relying on previous studies may be the only realtsuc way to conduct the imestigation. . An example of a data bank that is potentially U!)eful in quali- tatl\e .research. esp<>ciall} ethnographic (see Chapter Two), IS the Human Relations Area File (Murdock. 1983). This file b a compilation of ethnographic .,tudies of more than 350 societies; data arc classified and coded by cultural group and aho by m.ore than topics. Education is one broad topic wh1ch subtoptcs c;uch as t:lementary education, educa- twnal theory and method", students, and vocational education can be found. The index is organi7ed so that a resca1cher can reLJicve documems related to the educational practices of one pankular cultural group, or documents can be retrieved about a specific practice -.uch a" 'student uprisings" across many cul- Tvpes of found in this file include ethnog- rapher field notes. dial) entries, reports to \'arious agencies, books, newspaper article-,, works of fiction about the culture, and photographs. PERSONAl In contrast LO public \Omceo; of data, personal document.s " 1 efer to anr fir,t-pel"on nancuin: Lhat descdbes an individual's actions experiences. and beliefs" (Bogdan and Biklen, 2007, p. 133): Such documents include diaries, letters, home videos. children's growth records, scrapbook.-. and photo albwns, calendars. auto- biographies, and travel logs. In some ways documents are like observat!Ons in that documents gi\'e us a snapshot into what author thinks is imponam, that is, their personal perspt:ctive. while observation-. allow us to see O\ert behavior. Such documents can tell the about the innet meaning of everyday event'!, or they may ytelcl descriptions of highly unusual or idio- syncratic human experiences such a'i can be found in Admiral Byrd'-, 1 eport of his e'\.periences alone at the South Pole or Helen Keller\ accoum of overcoming multiplt' phystcal handicap'>. P<:r,onal document." are a reliable source of data conccrmng a attitudes, beliefs. and view of the world. Bm because they ate personal doc umcnts. the matt. rial is highly subjective in that the writer is the only one to s<'lcct what he or she con- sider., impmi.ant to record. Obviously these are not representative or necec;sarily reliable accounts of what actually mav hme occurred. Thev do, however, reflect the participant's perspecti\e, which is what most qualitative research io; seeking. In of and diaries in particular, Bur- ( 1982) notes: Thl' field researcher needs to consider: l'l material trmtwonhy? the materiaJ atypical? Has the material been edircd and refined? Doe'> 1he autobiographical material 011l)' contain highlights of life th<ll .u-e cotu.idcred interesting? Furthcnnure, it could he argued that t.he material is automaticall} onlv certain people produce autobiographies and keep diari(s: tl1ere is self-selenivil}' involved in Lhe sampk of material awilable; they do not prO\ ide a complete historical l'l'Cord. Nevertlwle-,.,, such material doc' prmide a subjective .trcounL of the situation it records: it i" a reconstruction of pan oflife Furthennore, it provides an at.count that i-; ba..'>ecl on the author's exp<.rience. (p. An entire stud) can be based on personal documents. Abratmon 's (1992) ca..,c study of Russian jewish emigration is based -.olelv on his grAndfather'!) diaties written over a twelve-year pedod. A well-known earlier -;rudy of Polish immigrant life relied hea,ih upon personalleuers wntten between immigrants and rela- tives m Europe (Thomas & Znaniecki. I 927). Mruw o( these letters were obtained by placing ads in local newspapers asking for them. PoPULAR Ct LTURF. Doct MNTS In addition to public and personal 1 ecords. society produces materials de11igned to entertain. intorm. and perhap., persuade the public. These are ptlblic in natme and so are c;ometimes catcgoriied under publk records. Popular media forms such as telt'vtsion. film. radio, newspapers, litt.t-arv works, photography, c.Lrtoons. and more rtcently the Internt"t are sources ot "public'' data. communication materiab are especially good sources J44 QtJAII'I:.HfVJ. ltcSJ!.ARCH for dealing with questions about some aspect of society at a given time, tor comparing groups on a certain dimension, or for track- ing cultural change and trends. The changing nature of U.S. presidential political campaigns, for example, could be looked at through the medium of televised debates, with the 2008 cam- paign incor-porating YouTube Internet technology. Studies have been conducted on the roles of minorities in television, the presence of ageism in cartoons, and teenage culwre in movies. Hughes (2003) and Hollenbeck (2005) both made use of popu- lar culture sources for their studies. Hughes asked what mid-life women learned from watching .soap operas. and Hollenbeck studied contemporary Internet-based social protest groups (anti- McDonalcls, anti-Starbucks, and anti-Wai-Mart). Unlike recod-; that are part of a program's history, or per- sonal documents that might augment an interview study, there may be an infinite number of popular cultural documentc; that might be relevant to a particular study. Bogdan and Biklen (2007) offer some advice when using popular culture as a data source: Of alllhe thousands of hours of commercial videos, films, and pop- ular records a.s well as the millions upon millions of printed words and pictures that appear each day in lhe media. how do you ever narrow down the scope to make }'Our task manageable .. . . Think small. Most people who read research do not expect the researcher to cover the universe. Pick a particular program. or a panicular event, and wot-k on it imensely rather lhan spreading yourself too thin. (p. 65) VISUAL DocUMENTS Film, video, and photography are visual documents. Of course these can be found within the categories of document<; just dis- cussed. That is, public records, personal documents. and popu- lar cultural materials can all be in visual formats. However, within the Last decade or so there has been a growing interest in their use as a data source and as a means of presenting the findings of a research -study (Stanczak, 2007). Likewise there has been attention to methods of analyzing visual images (Kress & Van Leeuwen, 2006; Van Leeuwen &Jewitt. 2001; Pink, 2006). l Although there has been renewed interest in visual materi-als, film and photography have had a long history in anthropology dating back to the turn of Lhe twentieth century (Pink. 2006). Most famous perhaps were the early 1940s film and photography of Balinese culture by anthropologists Bateson and Mead. How- ever, despite their ''innovative and landmark text in anthropol- ogy," ir "failed to achieve its potential to persuade anthropologists of the time of the value of systematic visual research and analysis" (Pink, 2006, p. 9). According to Pink, visual, sensory. and applied anthropology were marginalized until the late twentieth cen- tury, when in the 1990s they gained popularity: "Between 1999 and 2001 a series of new publications across the social sciences and humanities revealed a thriving interdisciplinary interest in visual research methods" (p. 15). Film has some obvious strengths and limitations. This fom1 of data collection captures activities and events as they happen, including "nonverbal behavk'r and communication such as facial expressions, gestures, and emotions" (Marshall & Rossman. 2006, p. 121). What can be captured on film is only "limited by what the mind can imagine and the camera can record'' (p. 121 ). Film has other, more practical limitations such as cost and the need for the researcher to have some technical expertise, and it can be intrusive (although as "reality" television shows attest, the camera is soon forgotten in many situations). Unlike film, photography is often less expensive and more easily incorporated into a research study. To begin with. one can make use of what Bogdan and Biklen (2007) call ''found photographs" (p. 142). These are photos that already exist, either in public archives such as historical societies and libraries or in personal collections such as a participant's photo album of family events. Photos alone can tell the story of what the photog- rapher thought was important to capture. what cultural values might be conveyed by the particular photos, and so on. Photos have recently been used by rel-learchers in post-colonial, African American, and women's studies 'to understand how oppressed groups were pictured by those subordinating them" (Bogdan & Biklen, 2007, p. 144). Photographs can also be generated by the researcher. Such photographs, often taken in conjLmttion with pankipant
J46 Q.U\1 1 1.\IIH provide a "means of rcnu:omhering and .,tudying detail that might be overlooked if' a photogtaphic image Wt.'J e not a\'ailablc for reflection" (Bogdan & Biklen, 2007. p. 1:11 ). A.nothet of in qualitative research j., caJled "pholo elicitation." in which participants are shown \ari- ous photos of the topic of interest in order to stimulate di!!tw.- sion of the l()pic (Harper. 2002). photos could havt hecn taken by the researcher, found in public or personal and so on. They are basically prompts for verbal data. ln yet another use of photos. tlw participants can lx prO\ided disposable cameras and asked to take pictures of the plu-nom- enon of interest. Participant-generated photos can then lx: ana- lyzed by lhe rcseat-cher or used bv tlw with participant:-. in the photo elicitation method just chscussed. In a studv of' dillcr- ing percep1jons of white and African American Greek member<; of their univeNity environment. researchers provided disposable cameras to participants to take photos exemplif)ring what their university experience meant (Perka, Matherlv. Fishman. & Ridge. 1992). These photos and interviews asking to imerpret the photo') provided the data for Taylor's (2002) '>tudy of teachers' belief., and Daniels' (2003) stud\ of women in a South MriC'an 'it'ttlement are two other examples of wking pictures ol the phenomenon being studied. and then these photm. being references for point\ of discussion in tlte interviews. Harpct (2003. p. 195) rcminds us. howevet. tltat ''In all examples of photo- elicitation research, the photograph loses its claim to objectivity. Indeed. tlw power of lhe photo lies in its ability to unlock the sub- jectivity of those who see the image diflercmh from the researcher." PH1SICAL MAlR1Al ART1FACfS Phvsical material as a form of document. broadly defined, comi"'" of physical found wilhin the .,tud) setting. Anthropolo- gists typicall} refer to these oqjecb as arlifacts, which include 1 he tools, implemems. utensils. and instruments of everyday living .. Hodder (2003) includes artifacts and Wl"itten texts that have physi- cally endured over time as "mute evidence" in the smdy of cul- ture. "Sue h evidence. unlike the spol...cn word. endures physically and can be across space and time from w; .unhot, :\hi' D.\T4\ H(U\1 llm 147 ptodut.ct', or mer" (p. 155). One ofthe monHmwus studies using physital material b the g-.u bagc stud\ conducted mer a number ol ye.u-s h\- researchers at the l niversit) of .-\ri1ona (Ratl1je & 2001). By .. orting through peoplt.'s garbage these researchers have been able LO tc.:ll a lo1 about the lifestyle choices ofvatious socioeconomic groups. For example. lower-income peo- ple tend to buy small conUlincrs of name brand products rather than less expensive. large-sited genetic brand products. As part of my observation ot an exercise class at a senior cen- ter in Korea (see Chapte1 Six, Exhibit 6.1, for the held notes). I noticed a number of framed plaques on the wall. These "arti- fac.L<;." which were tnmslatcd lor me. spoke to the Korean view of older adults and their learning. For example. one plaque srud "Let's transfer seniors gond experiences and wisdom to young people." .\.nother had a lisl of things older Koreans should do: "I Ielp our society; enjoy our life; be healthy; and participate. even if you are old." These plaques olfered additional evidence of the importance of pan ici pat ion and respect for older adults that I witnessed in obscning the class itsell. Phvsical u-ace material i., yet another potemial o;ourcc of infor- mation. Phvsical Uaces consist of changes in the plwsical setting b1 about b) tht" activitie' ol people in lhat <;ctting. The fo1lm\'- examples of phy-...icaJ tviclcnce being used in re')carch stuilies arc provided by Webb. Campbell. Sch\-\l".trtz. and Sechrest (2000): One investigator wanted to learn the level of whisky consump- tion in a town that was otfkially "dry." He did so bv counting empt) bottles in trash cans. The degree of fear induced hy a ghost storytelling -.cssion can be measured b} noung the <>hrinl...ing diameter of a l ircle of <,cated children. Library ,vithdrawals wen: mcd to demonstrate the cticCL nf the Introduction of tcle\ ision into a community. Fiction titles dropped. nonfiction titles were unaffected. A child's interest in Chrbtmas was demonstrated by distortions in 1 he site of Santa Claus drawings. Ra<.ial attitudes in two colleges were compared by noting the degree of clustering of blacks and whitt:" in lecture halls. (pp. Two basic mcaus of studying U<lccs are Lo no1<: t11cir erosion, '' hich io; the: degree of wear. and to note thei1 accretion, which is the degree of atctunulation. The wear and tear on floor tiles in from of a muo;eum exhibit as a sign of public interest is a well-known example of' erosion (Webb ct al.. 2000); the accu- mulation of v.h1sk\ in the preceding ltst is a good exam- ple of accretion . . More: tommonly. the ebb and of phy ... ical uaces are ll'it'd as data to document a phenomenon. and McDonald (2004), for example. tt-.ed school records to re,eal into the reading habits of junior age children. And Pauon (2002. p. 29:l) give' an interesting example of hem phvsical trace ... can be used in evaluation: "In a week-Jong stall' training program for 300 pt'ople, 1 asked the kitchen to SV'item- atically record how much coflee was consumed in the mo,rning, afternoon. and <vening each day. Those sessions that I judged Lo be particularly bo1 ing had a correspondingly higher lt:vel of col: fee consumption. Actiw and involving sessions showed less coirec consumption, oftlw time ofday. (Participants could gt'l up and get coffee whtn<.'vel they wanted.) Because physical traces can usuaJiy be measured, the, arc most oflcn suited for obtaining information on the incidence and frequency of bcha\ior. They are aL;;o a good check on informa- Lion obtained from inteniews or -;uneys. l n quaJitath:e research. most phvsical tr<tce mcasuns a1e used to supplement data gath- ered rhrough intcn1ews and observations. A researcher 1mght. for example, compare tJw weat and tear on computer terminals in a school program that purports to include computer literacv in it-. basic curriculum. Other ad\Cuuage'i of using U<lce measure' are noted ( 1979. pp. 78-79): Tr.tce measun.-; record the results of actual beha,ior. not reported or experimental approximations. Trace mc<tsurcs are usually nonreactive and unobtrusive. Since they are applied after b<'havior has occurred they do not mod- ify the behavior they seek to study. Material rraccs arc ubiquiwus and readily available for stud). Becaust' material are applied to inanimate objects, they usuallv rcqUJre minimal cooperation and inconvenience from human J Because the number ofrneasun.s of traces dep<nc:b upon tJtc recorcle-'s interest rather than informant patience, a \'ariety of imen elated behaviors can often be -;tudied at onc.e. Because of the minimal inconvtnicnce and expense to infor- mant.'>. uace can he w.ed ove1 long time periods as longitudinal monitoring de\ ires. o Dor.uM l. NT<., When documents are included in a ..,tU<h. what is commonh bemg referred ro are public records, personal documents, and physical material alreadv in the re-.earch .,cuing. Because thev have not been produced for the research purpose, they oiten contain much that is irrelevant to the study: by the same token. they can contain clu<.'s, even startling insighLo;. into the phenomenon under study. Most researchers find them well worth the effort to locate and examine. Researcher-generated documents arc documenLs pre- parcel by the researcher or for the rec;earchcr by participants after the study has begun. Th<. sperific purpose for generat- ing documents is to learn more about the si.tuation. person. or event being investigated. The researtheJ 1mght request that someone keep a diary or log of activiticl> during the course of the investigation. 01 a life hiswrv of an individual or histori- cal account of a program might be solicited to illuminate the present siruation. And a.., di'icu-.sed abme, photographs taken hy the researcher or the participant.., can be a \'aluable source of data. Quantjtati\e data produced b\ mvestigator also fall into this category of documenL\. auiwdinal measures. content examinations. statistical data from suneys on any num- ber of topics-aU can be treated <lS dotuments in support of a qualitative investigation. In swumarv. tJ1en, documen'\ include a broad range of mate- rials .wailaule to the researcher who b creative in seeking them out. Literally millions ol public t\lld private documents, as wet! as physical traces of human behavior, can be used as primary or secondary source.., or data. Further, documents can be generated by the researcher once the stud\ has bcgtm. I 10 Qu" ''"'.'"" RJ..,t..,uc.H USING DOCUM\TS QUALITATI VE R ESEARCH documenLary material as data is not much different from uc;ing interviews 01 observations. Gla.o;er and Strauss (1967) cornpan:: fieldw01k \\ith library reo;e<Jrch. "\l\l1en someone -;tancb in the Jjbrary stacks, ht is, metaphoricallv. surroundt>d hv mices begging to be heard. Every book., C\t:f\ magazine article, repre- at least one person who is LO the anthropologisL's tnlonnant or the sociologist's interviewee. ln those publi cations, people converse, a.nnounce positions, argue .,.,ith a range of elo- qmnce, dec;cnbe ennts or o;cencs in ways entirely compara- hle to what ts and heard during fieldwork" (p. 163) \1\'hcther m fieldwork. or libra!) wot k, the data collection is guided bJ queo;tions. _educated huncht's, and emerging findings. Altl.lOugh the search s systematic, both settings also allow (or Lhe uncovering ol valuable data. Tracking down leads. bemg open to new insights, and being sensitive w the data are the whether the researcher ill interviewing. obscning. or anai}'Lmg Since the investig-ator is the prim, 111 instnt- ment for gathenng data, he or she rehe-. on skillc; and intuition to find and interpret data from document<;. materials is the lirst step in the process. As I meutroned, th1s IS generally a systematic procedure that evolves from the topic of inquin itself. A qualitative stu dv of cta.,sroorn instruction would lead to documentc; in the form of inc;tn 1 ctof'> les-- son plans. studenr a'>signmentc;, in the classroom. official grade reports school records, teachet evaluations, and so on. Besides the setting it-;clt, lhe logical places to look are libraries his-- tmical societies. archi\es. and institutional filec;. haw person_al doctunenL.., like letters and dituics bv placing acherlise- mem_s m and new-;Jetten. or on relevant Internet sites. hus researcher must 1--eep an open mind when it comes open to any lead to serendpt ous dtscovenes. Tobacco company expo- ses of the late 1990s were buttressed by the discovery of buried t.nemos in, which the addictive qual it' of nicotine is discussed; the wen literalh .,tumbled upon during rou- tme questtorung of\\ h1tc House staff. \11Ni r;<. H AlA rROM Dm 1 151 Once c.locumt.rH .. ') ha\e been Jocatt:d. their .lllthentictty must be assessed. "The author. the place and the date of writing all need to be e ... tablishcd and vedficd" (McCulloch, 2004, p. 42). In addition. the conditions under which the document was pro- duced is imponant to ascertain, if possible. A news release to the general public serves a quite different purpose than an inter- nal memo on the same issue. 1n evaluating an artifact-that is, an object used ot produced by a particular cultural group-- LeCompte and Preissle (1993) suggest that the researcher ask such questions as, What is tJ1e history of its production and use? How is its use allocated? Is ilS selection biased? How might it be distorted or falsified? Determining the authenticity and accuracy of documents is pan of the re.o;earch process. It is the investigator's responsibil- Ity to detetmine as much as possible about the document. its and reasons for being written, its author, and the con- text in which it was written. Cuba aud Lincoln ( 1981), citing Clark (1967), list the questions a researcher might ask about the authenticity of documents: What is tJ1e history ol' tht: document? How did it come into my hands? What guarantee is there that it is what it pretends to be? 1s the document complete, as otiginally constructed? Has it been tampered \vith or edited? (f the document is Renuine, under what circumstances and for what purposes was it produced? "Who was/ is the author? 'v\'hat was he trying to accomplish? For whom was the dont- ment intended? What were the maker's sources of infom1ation? Docs the docu- ment represent an eye\\itness account. a secondhand account, a reconstruction of an event long prior to the writing, an interpretation? What was orb the maker's bia'i? fo what extent was the wTiter likely to want to tell the truth? Do other documents exist that might shed additional light on he same storY. event. project, program. context? If so, are they ..tvailable, accessible? Who holds them? (pp. 23S-239) \u important distinction for hi,torians that qualitaLi,e researcher., might abo attend to is whether documenLs <1re pri- mal) or sccondaf) sources. Primary sources are those 1n which of the doCLunent is. recounting firsthand nence wath the phenomenon of interest. The be-;t primal) source.., are those recorded in time and place to the phe- nomenon bv a qualified person. Given rhis definition, moc-,t pt'r- sonal documents and eyewitness accouru.s of <;ocial phenomena could be considered primal) resources. Secondary 'otuces arc of <l phenomenon b) dwsc who have not directh cxpe- nenced thl' phenomenon of inter<"'ll; these are often compiled at a later elate. Interestingly. the same documem could bc cla<;- sified a!'! priman or secondan depending upon the purpose of the stud\. Tht' di;u-y of a loved one caring for someone with ter- minal cancer. for example, would be a primary source of data for a on it would be considered a secondary of fo: understanding how patients themselves cope with a terrnanal chscase. Aftea asscsc;ing l.he authenticit) and nalUre of documents or artifact:'. the reo;earcher must adopt ')ome S}Stem for coding and catalogmg them. If at all possible. waittcn documents should he photographed or \identaped. By establi-.hing basac descnpuve categories early on for coding. the researcher wil.l han. easv access to information in the analpis and interprc- tatlon stage. In il case study of a caner enhancement award pro- gram. for example, applications for the award were part of lhe database (Zeph. 1991). The applications were coded according to the applicant's type of emplovment, dollar amount of request. sex, geographic location, and nature ot the project propo-;cd. In qualitative studies, a fonn of content anaJv:,is is used to anahze documents. EssentiaHr. content analysis is a sy,tematic procedure f01 describing the contem of communications. Histo- rians and litemq critics have long used contenl analysis w ana- lpe historical documents anci literary works. Modern contem analysis has most often been applied Lo communications media (newspaper-;, periodicals. tele,i:sion, film) and has had a stroug quantitative focus. A major concern has been measuring the frequetl<:Y and variet'r of message' and confirming hvpothescs. MoM tCllearch designs using content analvsis are <;equentml in naturc-"moving lrmn category construction to data collection, data analY''i' and inlerpretat.inn" 1987, p. 6H). Data collection and coding are carrwd b\ nov- ices using prmocols and tmined to countnntts oJ anahsts. Quantification need not be a component of conl('llt analy- sis. however. Rather. the 1lflllln' of lhe data can also be AJthcide (1987) describes how qualitative content analysts chffers from conventional comenl ''Ethnographic anal-. ysis is used to document and understand lhc ol meaning, as well as w 'erify IL-, dtstmc- tive characteristic is the and htghly tntcr-acuvt nature of the imestigator, concepts. data collection and analysis .. . The investigator is continual!\ central. although pro.tocol-; ma} be ul!ed in later phases of the research .... The atm Ill to be system- alic and analytic, but not Jigid" (p. 68). LIMITATIONS AND STRENGTHS OF DocuMENTS ln judging the value ot a data ;;ource. a re.searchcr can a!;k \\ hether it conlains infom1ation or ansights relevant to lhe researd.l ques- tion and whelher it can be acquired in a reasonabh yet systematic manner. I( thc'e two questions can be an the affirmative, Lhere is no not to use a sou:ce. of data. Documents or artifano; have been underu'it:d tn quahtatwe research, however. Ove1 forty years ago Glaser and Struuss ( 1967) attributed this undemse to the fact that researchers prefer to P.ro- duce their own data, that the use of documents i-. too much historical research. that re:-;earchers want "tO see llw sll- uation and informants in per-.on (p. and that thcii own competem'} in usmg documentary matenals. These barriers o;eem tnte todav as well. Preferences for othet -.ources of data may reflect a research- er'!'! uncertainty about the potential of for vielding knowledge and insight. But the may also reOect some of the limitations inherent m data Sevt:ral limitations slem from the basic difference between thts '><>urce and data gleaned (rom interviews or most documentary data have not been de,elopcd lor research purpo'e". The matcri.th mav thtrcfi,rc be incomplete from a rt.>search perspective. In contrast to field notes, avai lable materi- als rna> not 'afford a continuity of unfolding evencs in the kind of detail that the theorist requires" (Giac;er & Strauss, 19b7, p. 182). \Vhether personal accounts or nffic.ial documents are involved. the source may prm ide um epresen tativc c;amples. "Oftl'll no one on the project keeps ver-y good notts 011 processes. few memo- ?enerated. and, even more often, the 011l} wliting that 1s done I S m response to funders' requests for technical reports or od1er periodic statements about the progress of the pro- gram or prqject. If no documents exic;t, however, or if the docu- ments are sparse and uninformative, this ought w tell the inquirer something about the comext" (C.uba & Lincoln. 1981, pp. 234-235). ?ocuments are not produced for research purposes, the mformat1on they offer rna> not be in a form that is useful (or unc!erst..<mdable) t? the investigator. Furthermore, such chua may be mcongruent With emerging findings based on ob'lervationaJ or interview data. Thic; i-,, of cour-.e, more of a problem when documents are used as s<>condan data sources to veri\ findings based on other data. ll documents ate used as part of tht process of inductively building categvries and theoretical constntcts as in qualitative case then their "Ci t" with preestablished con- cepts ot models is lec;s of a concern. A third major pmblcm wirh documentary materials is deter- mining their authcnticit} and accuraq. hen public records that purport to be objective and accurate contain built-in biases that a researcher may not bt' aware of. F01 example, the incidence: and frequency of crimes reported in police records may be a function of how certain crimcs arc defined and a particular department's procerlures for reporting them. Personal documents are subject to purposeful or nonpurposeful deception. There is likcl} to be. for example, an underestimation of income in a per,onal income tax report versus an overe'itimation of' expenses in a grant pro- posal. Distortion in personal document..-; may be unintentional in that the wliter is unaware of his or her biases or simply does not remember accurately. Sdlliz. Jahoda, Dcul\ch. and Cook ( 1959, p. 325) quote Augustine. who noted tim problem of authentic- ity in his famous per'\onal docwnem. Gimft'uions. "And when they hl'<IJ me tonfc,.,tng of mv.,elf, how do thev know whether J speak lhl t1uth?'' Comt-rn abou1 authemicity LO historical docu- mcnh a.\ well as to <lllOHpnous prqjcct report.<> and sources who wish 10 remain anonymous. such "Deep Throat" of the 1974 Watergate case (Webb et aJ., Oe,pite these limitations. documenLS are a good source of data fo1 numerous reasons. To bcgrn with. thev ma" be the best ... ource of data on a p<lrticular better than observations or imc.rvicws. Mam document!. arc eac;il} accessible. free. and con- tain information that would take an investigator enormous time and cfJort to gather otherwise. Fot example. if one were inter- e'\t(d in a historical case study of an institution 01 program. doc- uments would be the bec;t source of data particuladv if persons associated ''ith the. institution were not a\-ailable for interviews. Other situations whert documenrs .tn likel} to be the best source of data would be studies that rdy on technical expertise such as medical reports, and studies of intimate personal relat ionships that cannot be obser-ved or that people arc reluctanl to discuss. The data founrl in documelll!- lan be used in the same man- ncr as data from nHnviewo; 01 observations. The data can furnish descnptiH information, vedF. eml'rging hypotheses. ad\-.mce new cat('gones and hvpotheses, offer historical understanding. track ch.mge and dewlopment. aJld :-.o on. Glaser and StrauJ-;s (1967) poim to the usefulm"' of documento.; for theorv building-a pro- that ''begs for compar-ative analysis. The libraf} offers a Jan- ltntrc range of comparison group,, II onl} the researcher has the to disCo\CJ them" (p. 179, emphasis in onginal). One of the gnmest ad\'antages in using documental} material is its stability. Unlike interviewing and observation, the presence of' the investigator does not alter what is being studied. Documen- tan data are sources of data compared to other forms. Such data have also been called "unobtrusive." Webb and others' (1966) classic book on unobuushl' measures in ilS revised fonn is titled VonrPflCIItlf> Afpasure.\ m the Sorml Sri en cPs ( 1981) because. I hey wtite. 'we came lo reahlC mer the years that t.hc original title was not the best one since it was the nonreacti,ity of the measures rather than thcit unobtntsiYenes" that wa.." of majm concern'' (p. ix). :-.Jonreactive measure<; tnclude physical trac('s, official re< ore!<.. priYate do<: uments. and 'iim pie ob-.ervalions. Thus, like :lll} utll<'r source of <.lata, documents have thdr limitations and theit advantage<>. Because the)' are produced for rea<>on ... other than research, they may be fragmentary, the> rna} not fit the contcptual framework of the research. and their authenticin mav be difncult to determine. Howe\'( r, they exist indepenclem of a re-.eaJ-ch agenda. thev are nonreac- the. that is. unaffecred b) the research process. The> are a prod- uct of the comext in which the) were produced and therefore grounded in the real world. Finally, manv documents or artifacts cost little 01 noLhing and arc often cas} obtain. ONLINE DATA SOURCES wl.10 nads a newspaper has seen the te1m infomw- lwu Juperlughwfl)l applied to the Internet and heard about the explosive growth it has und<rgone in the last few years. From its humble beginnings as a communication roo! for uni- versity professors and scientists (initially designed w
the results ora war). the Internt.t has become a standard resource for coUege students. businesses, and an)'one else who has access to a computer with a modem. In addition to providing a number of reference sources-albeit of uneven quality-the Internet supports interactions among peo- ple through various fom1s of computer-mediated communication. E-mail, listserYs, new'igroups. chat rooms, wikis, blogs. and other interactive emironments allm' people who have never met to encounter one another and tven e\tablish relationships conducted primarily through online contacts. These interactions, still some- ill-defined ''itl1in our 'iOCiery, are of obvious interest to quali- tatJ\e researchers. In addition to being a focus for stud> in and of themselves, these Internet interactions can be sources of data relat- ing to other studies. \!\'hat factors mtl.':lt be considered when access- ing and analyzing these data sou1 ces? In this section I eAplore some of the issues associated with the use of online data sources. I l ow are these sources similar to more familiar sources. such as documents, interviews, and obser- vations? How are they different? What issues and concerns are raised by the eflccts of the media on the data-gathering process? \Vhat ethical considerations arise in this new research context? These .tre not questions easily answet cd. nor are they the exclusivt' province of qualitative n<>cclrchcrs. Articles in com- puter-related magazines and the prt>-;s regularl.v discuss ,alious effect-. of the Int<.rnet on soc1cl\ at large, rangmg from explorations of the "multiple -.elves'' po.,sible online. to men- tiom of "'online affairs between pt'ople who have never seen one another in person, to organiing '\Ocial protests. to buying and selling consumer goods. and indted. to conducting illegal acti,;ties. Even standard news magazines highlight issues related to cyberspace-the ambiguou" destination to which the infor- mation superhighway leads. Since the changing electronic land- scape outpaces the publication or specific maps or guides, this discussion merely outlines a general ntngc of concerns. For any particular area of study. the specific application of these consid- erations will vary. ONLINE VERSUS OFF- UNE DATA In qualitative research, the basic ways tu collect data have traditionally been through intcniews, observations, and exami- nations of documents and artifacL'I. Mam of the references and data sources available online reflect dl:cuactcristic:.. of these famil- iar data sources. Web pages, paper!. available through file trans- fer protocol. and 'arious form" of "electronic paper" can be considered documents that are accessed online. Illustra- tions and game,-available m static form to be downloaded by the user can be treated .ts artifacts. as can many of the video formats such as YouTubt., and Facebook. E-mail can be used to question indi,iduals as in an interview, and can observe he online inter.Ktiom among individu- als in a \'ariety of formats. To some extent then, online data coll<:ction ollers an electronic extension offamiliar research techniques, widening the scope ot data available to the researcher. Certainly. many of the dtcisions faced in off-line sin.tations emerge in parallel form in online research: whether to join an online community as a complete observer, a com- plete participant, or someLhing in between; how to select a sample group; how to approach potential participants when initiating a study; how LO gain tru'lt and so on. )'i8 Qll.\11 \AII\'1! R II I Iowevcr, on lim data collection has some important clue to the nature of tlw medium through which iL is conducted. These havt a profound influence on tlw study that must noL be tgnorcd 01 uiviali/ed. F01 example, individuals who do not han access to computer'> will be automaticalh excluded from the study. Is tlw. appropriatl' fm the 'itudy. or will demographic dif: ferenccs that correlate wah computer access diswrtthe findmgs? Though the amount of informatjon mcreases to an over- degree, not all cdtical interactions are necessarily a\ ail- able tor o;rudv. Students in an online course mav also communicate through private e-m.til rnes<;agcs that the researcher never sees. Quantitv of information no guarantee of comprehensiveneo;s. In each fi.lt m of computer-mediated communication has a effect on the information it transmit'>. For exam- ple, an e-mail interview may have the same verbal conLerlt as one conducted iu pcrl,on, but it lacks infl ection, body lan- guage, and the man) other nuances that often commun icate than word'i. Frequent userl, of e-mail recognize iL., hmnatJOns; new arc regularly warned that jokes and sar- casm do not travel well online, and the) are taught "emoticons" that attempt to rtplicat(' tht' emotional richness common in speech. At the same time that some communication character- istics are curtailed or modified. others are artificial!) enhanced. The asynchronous nature of e-mail can add renection time to an online in ten iew that would be unavailable in a face-to-face session. lmmediar e reattiom, strong emotional responses, and unguarded exprt'S!)ion" are all lost to the researcher unless. after second thought, the participant chooses to make these transient thought-; available-and i'i capable of articulating them in !hese coul d completely change the interprt- tauon of a re-;pon-.<. Comcr<;elv, a casual response may ha\'C an unexpected and unsctthng pennanency; e-mail exchanges long forgotten can resurfact, in totally different and e\en misleading contt.xL'i. . as. they become familiar with the evolving conven- tions of onhnc cxpressi()n, rcse:archcrs need to remain alert to the variables of electronic communication. Participants in Jist- servs_and uscnct group<. have an entire terminology to describe certam typt'S of exchanges. Funher, most groups expect new 1\ltNJNl U ;\ 1<\ I ROM 1)()1.1\lt.NTS 159 participants to read the Frequently Asked Questions (FAQ) ?f their group, then lurk. "reading posts lO a newsgroup or in order to undeNtand rhe wpics and tone of the e>.changes tn the group before the offers hio; / her input'" (Chen. Hall, & Johns. 2004, p. 164). . . . . In terms of group interaction ... wnung sktlls an? literaC\ stronglv influence how indh1duals are perceJVed onhne. Often someone will seem to have an entirely different character: a funny, charming person can -.eem camtic and the smile accompanying the words disappears. Another mdivtd- ua1 \'those writing is mature and thoughtful mav prove to ha\'e limited social skill'> when depriwd of refl ection time and forced to react spontaneously. . .. This discrepancy between rlal and onhne personahues occurs even when people are trying to be themselves-or at an idealized version of themselves. It is compounded when mdJ- viduals purposefull} create different _onliue is fairly frequent in somt: electronic tiun can vary widely. from scholarly 111 mdt- viduals Jist their real names with their unrvcrsny affihatwns and degrees. to fantasy games in which make up and descriptions that reflect little of their off-lme where role-playing shade into deception? In tbts .. there is unprecedented capacatv for indi\iduals to mtsrepre- ')ent themselves and their pcrsonalit) mtit.'i in lntemet tions, to the extent of adopting ahernathc personae for thetr Internet interactions from their 'reaJ-Iifc' interactions" (Hewson, Yule, Laurent. & vogel, 2003, p. 115). Under these conditions, the assumption that the world is composed of multiple, chang- ing realities-pan of the qualitative paradigm-:--becomes o_nce a triviallv self-e\ident obscn-ation and a magmfied comphcauon. Judging,inciividuals by the way they choose to pre'iem _themselves online is a risky business, and verification or tnangulauon may be far less reliable than in the .. real work!." Ev<:n online documcnto; and artifacL'i take on new qualities. The Web page cited today may he gone tomo_rrow or the c?n- tent changed radically as to be unrccogmzable. Managmg data assumes a new dimension when iL'> -.tability can no longer be taken for granted. Version con trol , once only of concern to 160 (.!t Tl'.\ Jl\"1 Rt .. , \KI II programmer<> and editots. cmergts a" a nitical issue (OJ ' a.nvunt using the Jntcrnet as a rdercnn or a rc.,ource. This is JWW tenitmy, with rules that change as as the} .tre idt>ntified. Mv IJ<:.,t advice tor researcher' is to recognize that the of their reo.;earch arc strongh infJu- enced bv the chat of th( datA revealed, OJ' altereci becaust ot the nature of the medium through which they pre'ientecl. Anahting. de::.cribing. and discussing rl1e poten- ual effects of tlw-.c characteristic-. will be an important aspect of rest>arch conducted from online data. EFFEcTs oF 1 uE oN Ot\T\ Jn addition to the dif1(rcnces between online and off-line data, differences caused by tlw manner in which data are gathered must be considctccl. In quali tative research. the re'iearcher is the primary instrument for data collection and analysis. This factor is mually perceived a'i an advantage. because humans are both responsive and adaptive. At the samt' lime, it canies the rcsponsi- bilit1' of assessing and reporting researcher biases rl1at might have an impact on the stuch. vVhcn collecting data from the Internet, the researcher ts no longer the pdmarv instrument for data collection; a v<uiet' of software tools nl\l!>l he u..,cd to locace, select. and process it;for- mation. Like the n.,catcher, these tools have inherent bJase ... that rna} affect the study. but thetr biases mav be \'en subtle- and often much mort difficult for a to detect and describe. As Norman (1993) obsenes. "different afl'ord different operation.,. That is, the\ make some things easy to do, orhers diffiwh or impos'iible. It "hould come as no sur- prise that those thing.;; that the .tlfordanccs make easy are apt to get done, those things that the afl'ordances make difficult are not apt to get done" (p. 106). These passage., raise critical concern!'! for qualitative research- ers accessing data from th<' fnternet: How are their tools shaping the task? In a study of how people with disabilities engage \-vith and use computet for example, Seymour (2001) found that bv using au interYicw "ite on the Internet, she could n.: lea 5 e -th e intervte\\' from iLs imptisonment in time and place'' (p. 158). She explains: Accessible discussion optn lot an extenclt-d period of time, prmide access to issues and ideas as thev arise rather than a!> thev are recalled in retrospect. Participants-rc'iearchct and mav revisit the data. ask fut elm ifkation, t'xtend a poml or redirect the 1 -esearch. In comrast to th<: c-.scntialh quc-;tion-re..,pome nature of the face-to-face intcrvte\\ and the qttc"tioonairc. technologr facili- tates lhe development of ongoing communi{ational interaction between lhe participants and \\illthtL<; mak<: a contribu- tion to more egalitarian re!>earch. (p. 159) Seymour also noted that "the long duration and open nature of the interview raised a number of significant issues such as sus- taining a commitment to the pt ojcct, interpreting gaps in com- munication, and so on" (p. 15:3). Again thb is a rapidly evolving area; the researcher's responsibility must be to dcKribe tools and methods, as well as their potential ciT eeL-. on the work. ETHICAL IssuEs In any qualitative studv, ethical relating to protection of the participani.S are of concern. In an online these 1 -.sue.; overlap the public debate abom ownero.;htp of mtellectual propertv, cop)righ!, and hee 'ipetch. The abilit)' to copv, archive, and easil} edit huge volumes of matenal wnuen bv faceless masses can lead a to forget that these are the words of indhiduals. E'en the names are changed. some people are easih identiftetl bv the details their_ messages. The highly public nature of some of the elcctromc in which people exchange ideas can lull researcher., mto forget- ting the right to privacy 1hat thc<>c ha\'e, as_ seeming anonymity of electronic lull tndiVld- uals into revealing highly intimate detmls of thct r hves to anyone who happens to be reading their messages. . vVith the increased use of the lutcrnet [or research. more writers are attending to the ethical issues involved in working in this new medium. Hewson. Yule, Laurent, and Vogel (2003) iden- tify four issues in particular that nted lO be thought through in lnwmct research. The fir t i'> obtaining mfonned consent; tracti- tionall}.' !Jarticipams sign a statement indicating their \.,.iiJingness to. paruCJpate and .nc>cd to be eighteen year<; old or ovet lO give th1s consent \v-.tys have to be established for giving con- sent and Lh.at panicipant is an adult. Ensuring and of information is a second isstw; again. can be put m place to enable confidemiall> but in thts mc:dlllm thev are not as effective as in person-to-per-;on data garhedng. A third ethical issue is determining what is public and what is private: 'The cmcial question is whether the rese<ucher is ethically justified in using publ icly availahle information as data for a research study. Or, more specifically, in which context is this ethical!) acceptable or not acceptable?" (p. 53). The fourth ethi- cal isc;ue is how to develop debriefing procedures so th<H partici- pants may make comments or ask que tions, and to ensure that no ham1 has occurred. The term particijmnls is commonly used by qualitative researchers to describe the individuals being studied. Il is a care- fully chosen identifier, with connotatiom of inclusion and will- ing cooperation. This single word captures a number of attitudes research from the qualitative paradigm. It also o;cnes a-. a litmus .test ethics. If this t<'nn cannot be accurately used-1f subjects more appropdately describes the inclusion of unwilling or uniniormed individuals undc1 the researcher's scru- tiny-then the researcher honestly reevaluate the meth- ods and procedures of the study. The growing imponance of online interaction makes it a nat- ural arena for qualitative research. Three critical areas that the qualitative researcher mus1 consider are the effects of the context on the data, the effects of software functionalilies on UH:' data- gathering process. and the effects the medium tends to have on ethical Explicitly considering and dec;cribing the impact of these tactors 1s a new re...,ponsibility of the qualitative rc-.ea 1 cher. SUMMARY a third rnajo1 source of data in qualitative rec;earch, Is broadl) defined to include public r<'cords, personal papers. popular culture ,;sual documents. and phvsicaJ material and atifacts. Although som(! documenL'> might be prcpan:tl at the investigator's requc...,t (such as a re:-pnncicnt keeping a diary or wliting a life history). most are pruduc_ed indt:pcndently of thC' research study. fhty are thus and in the context Lmder '\tudy. Because ate duced (br reasons other than the study at hand. some mgenUJty is needed in locating documents that bear on the problem and then in analyzing their content. Congruenct- between and the research problem depends on the research.er flexibil- ity in construing the problem and the related quesLJ.ons. Such .a stance is particularly fitting in qualitative studies, by verv nature, are emergent in design anti inducuve m analysis. of all can help the uncover mean- ing. develop understanding. and discover insights relevant to the research problem. . Data gathering onlmc is an emerging area tnterest. for qualitative researchers. However. a of bC: con- sidered when using data from an onltnc InteractiOn; I 1 ev1ewed some of these issues in this chapter.