(Research Methods in Politics Cap 9)011

8/12/2019 (Research Methods in Politics Cap 9)011

1/17

ResearchMethodsin Politics

PeterBurnhamKarinGillandWynGrantnZigLayton Henry

p lgr vem ml n


2/17

h pterliteInterviewing

This chapter is concerned with a technique particularly used bypolitical scientists, lite interviewing. It explains what is distinctiveabout this technique and examines the problems associated with itsuse. Researchers using this technique need to decide w ho they aregoing to see, how they are going to access their interview targets, thebest way to conduct the interview and how they should analyse theresults.When it is carried out effectively, this technique can m ak e aconsiderable contribu tion to the understanding of political phenom-ena. /The majori ty of work by political scientists is concerned with thestudy of decision-makers and henee a key research technique forpolitical scientists is what is known as lite interviewing. This maybe defined both in terms of the t rgt group being studied, an liteof some kind, and the rese rch technique used, mos t characteristi-cally wha t is known as semi-structured interviewing. It is often th emost effective way to obtain information about decision-makers anddecision-making processes.More generally, lite interviewing can beused whenever it is appropriate to treat a respondent as an expertabout the topic in hand (Leech,2002a, p .663).Unlike, for example, electoral studies where the balance ofkn ow ledge and expertise is us ually in fav ou r of the interviewer, liteinterviewing is characterized by a s i tuation in which the balance isusually in f avour of the respondent. This is because of their highlevis of knowledge of the subject matter under discussion and theirgeneral intellectual and expressive abilities. Techniques that workvery well with surveys of the electorate where each voter has onevote may not be appropriate. Indeed, one of the def ining character-istics o f lite interviewing is that some respondents m ay count morethan others in terms of their influence on the dec is ion-makingprocess.In survey interviewing, the emphasis is on standardization. Therespondent is presented with a structured questionnaire and inter-viewers are trained so that the questionnaire is administered in astandardized way. Any variation should arise f rom the respondent s

205


3/17

206 Research ethods in olitics

views rather than the research instrument or itsadministration. Inpractice, this may not be easy to achieve, but the techniques andexpertise developed and accumulated through many decades ofsurvey research are intended to maximize standardization. In lite interviewing, standardization may hinder the successful completionof the research. Not only may the approach used differ from oneprojectto another,but interviews withinasingle projectmayhavetobe handled differently. There is no standard set of techniques thatcan be applied, although this chapter will attempt to offer someguidelines.Despite its centrality in the study of politics, there is not a verylarge literatureonlite interviewing.Threeof themost useful stud-ies are Dexter (1970); Moyser and Wagstaffe (1987); and Rubinand Rubin (1995). Burnham (1997) contains a number of studiesin which PhD students reflect on their own experience of liteinterviewing. In part the lack of a literature may be because it isdifficult to generalize from a seriesof how it was for me' studies.The paucityof the literature requiresthat this chapter should givemore specific guidance to intending researchers about the practicaluse of the technique. It is possible to derive some guidelines based Ion thework thathas been undertaken. The broader methodologi-cal issues raisedby the technique are returned to at the end of thechapter.

The key guideline must be not to base any piece of workentirelyon lite interviewing. This is consistent with the principieof triangulation which entails using more than one method orsource of data in the study of socialphenomena' (Bryman, 2001,p.274). 'Thebest research on lites has utilized a combination ofmethodological approaches to deepen the research findingsHer tz and Imber , 1995, p.ix). One should also use other sourcesand techniques such as archives, materials on the Internet andobservation at meetings of, for example, legislative bodies. Insome cases, of course, lite interviewing m ay serve as a supple-mentary technique.For example, a studentmay beundertakinganarchive-based project,but may also interviewkeyindividuiswhoare still alive.Researchers using lite interviewing as a technique must coverfour key points:a) decidewho youwantto see;b) getaccessand arrange the interview;c) conduct the interview;d) analysethe results.


4/17

lite nterviewing 207 ecide who youwantto seeIt is se l f -evident that who you w a n t to see wil l be de termined by the

purposes of your s tudy. Let us suppose t ha t , as is popu la r in con tem-porary polit ical science, you are seeking to s tudy a particular policycommuni ty or policy n e twork . One o f your concerns wil l be toestablish which actors are included in the community or networkand wh at i ts bound ar ies a re . R ead ing secon dary sources and v isi tingweb sites may give you some ide as abou t who is inc luded and who isexcluded.Con ve nt iona l sampl ing techniques wi l l not usua l ly he lp you inlite in terv iew ing un less you are going to use a s tand ard ized que s-t ionnaire and a mass sample. This is not appropria te, however, inmost l i te s tudies as respondents are not of equal weight . A moreusua l t echn ique is w h a t is r e f e r r ed to as s nowba l l or r e f e r raPsampl ing . In pract ice, this m e a n s t ha t you s tar t o ut w ith a few keyi n fo rman t s t h a t you have iden t i f ied f rom your l i t e ra ture an dIn terne t search. Y ou then ask them to ame o the r key i nd iv idu is Iyou should see wh o are re levan t to your s tudy. Your s nowba l l thusgrows la rger and l a rger as you gain access to a n e twork o f i nd iv id -uis in a part icular pol icy arena. There may be nat ional varia t ionsin the e f fec t iveness o f th is technique depending on the dens i ty oflite in terconnec t ions . G e r m n l ites are s t rongly characterised bydense sectoral and terr i toria l networks, so t ha t i f you are able tosecure access a t one point, y ou wil l be mo re than l ikely to b e r ecom-m e n d e d to o thers wi thin the n e t wo r k (Pa terson an d Grix, 2000,p.18) .The p roblem then becomes when to s top. The most f requent ly asked quest ion posed by s tudents who intend to use l i te in terview-ing in their research is How many responden ts should I i n t e rv iew?There is no simple answer to this question, as in large part it shouldbe d e t e rmined by the object ives and purposes of your s tudy. Sometimes i t is possible to interview the whole un iverse of respon-den t swhen the to tal n u m b e r is re lat ively small . For example , as partof a s tudy of the r e la t ionship be tween bankers and fa rmers , i t waspossible to in terview all n ine heads o f agr icul tural depar tme nts in themain En gl ish and Irish banks, plus their equivalents in thre e specia l-ist agr icul tura l l end ing ins t itu t ions (Gran t and M acNam ara , 1996,pp.427-37).What one has to bear in mind is that l i te in terviewing is a veryt ime- in tens ive t echn ique . I f one a dds toge the r the t ime involved in Iset t ing up the i n t e rv iew; t r ave l l ing to and f rom i t ; the in terv iewitself; p r epa r ing th e t ranscript ; an d a n a l y s i n g th e t ranscr ipt , then a


5/17

208 Research ethods in Politicsf igu re of at least twelve hou rs per intervie w is not u nre al ist ic .Ber ry(2002, p .680) est imates that t ranscript ion takes tw o h o u r s fo revery hal f -hour of interview. Ten in te rv iews wil l consume 120hours of research t ime, or three 'standard' work ing weeks . It isimportantfor aresearcher to consider how much time toallowforin te rv iewing in t e rm s of the overa l l t ime budget for the thesis orprojec t .In practice, it is r e l at ive ly unu sua l for a researcher to u n d e r t ak einsuff icient interviews or to neglect to interview key actors. Rather ,the cont rary is the case: researchers tend to under t ake too manyinterviews, thus extending the completion t ime for the pro jec t .Apoint isreached whe re e ach additional interview yields diminishingr e t u rns . One needs to be able to recognize when the ' saturat ion /point' is reached in a series of interviews where each interview isadding re latively l i tt le to the stock of in fo rm at ion or unders tanding .Bear ing in mind all these caveats, 20-30 interviews might be areasonable target for a pro jec t in which l ite interviewing was theprincipal method. For example , the well- regarded s tudy of theTreasury by Deakin and Parry (2000) w as based on 30 interviewswith snior officials. /

et ccessandarrangetheinterviewThe biggest problem in get t ing an access to a m e m b e r of an litegroupisthat such individuis areu sually very busyand they have to beprovided with someconvincing m otivation for seeingare searcher.Goldstein (2002,p .669) comments, 'Frankly, "gett ing the interview"ismore art than science and, with few exceptions, political scientistsare not part icularly well known for [ their] skill at the art of coldcall ing" '. In a study carr ied out in Russia it took 15-20 ca lis toa r rangea single interview (Rive ra, Kozyrevaand Sarovskii ,2002, p ,683). One PhDstudent found that a government department haddecided not to give interviews to research students. A s Stedwardwrites (1997, pp.153-4), ' it 's worth pointing out to in te rv ieweessome benef i t oftheir part icipation (if youcan'tthink ofone, then askyourself w hy anyone wou ld agree to bein te rv iewe d by y o u ) 'Mostusually, this will be because they are interested in the topic you areresearching.There are probably certain advantages inthis respect in being aresearcher in a fore ign count ry , or at least in a stable democracy.Drawing on their experience of l i te interviewing in Ru ss ia , Rivera ,Kozyreva and Saro vskii note (2002, p.684 ), ' responde nts in more


6/17

lite nterviewing 209poli t ical ly unstable environments may be a good deal more suspi-cious abo ut the go als and purpos es of the research p rojec t .Never theless, respondents outside your borne country may considerthat they wil l learn something f rom you about pol icy design andimplem enta t ion in your ow n coun try t ha t m ay b e of use to t h em . A sPaterson and Grix caut ion U K researchers (2000, p .18), m ake surethat you are not only w ell informed of your Germn research topic,but also the UK equivalent, be it a company, policy or atti tude, asinterviewees will often a sk abou t them vSome litesaremore difficult to access than others. Mem b e r s of |legislatures tend to be sa turated by requests for research interviews.It is c haracter is t ic of business lites t ha t they a re quite good at insu-lat ing them selves f rom unwanted disturbance (Thomas, 1995, p .4) .Non-governmenta l o rganiza t ions , on the o ther hand, a re u sua l lyquite keen to have opportunit ies to put their case across. Before research is s tarted, one of the design quest ions t ha t should be 1confronted is whether the target group is l ikely to be accessible .Stedward (1997) offers some good pract ical advice on what shecalisnegotiating access, including a sample letter sent to a potentialrespondent . A s Stedward points ou t (p .153) , par t icular ly in largeorganiza t ions , it is impor tan t to t a rge t appropr ia te ind iv idu is tointerview . In o rde r to do this, one has to gather a s m u ch in f o r m a -tion a b o u t the o rganiza t ion as po ssible. In the case o f pub lic sectorbodies, websi tes o r directories m ay inc lude lists of officials andshort descr ipt ions o f their duties (e.g., the Civil Service List in theUK).Having identif ied your target , you need to contact them by le t ter \, i f you should p rovide a concise and honest account of wh a t th e research is ?about , w h a t they a re seeking to achieve, a nd why they w ish to nter- view the par t icular person approached. I t m ay be necessary to followup the init ial contact with a telephone cali and this m ay r equ i rege t t ing r o u n d th e g a t ekeepe r s th e t a r g e t ed i n d iv id u a l h a ssurrounded herself or himself with. A s Stedward comments (1997,p.154), T he way you handle the phone cali is as impor tan t as theletter. The best advice I can give is based on m y ow n experience, andit is very simple: be p repared, pol i te and persistent.One r isk w h e n deal ing with a large organizat ion is of beingfobbed o f f . In some large organizat ions, you may f ind yourse l fbeing shun ted to someone w i th a t i t le such a s Di rec to r o f La te r a lP lann ing , a no t pa r t i cu la r ly w e l l- rega rded official who can be keptoccupied ta lking to inquisi t ive researchers. This is probably less o fa p r o b l em th an i t was in the pas t , as personnel pol ic ies have


7/17

210 Research ethods inPoliticsbecome more ru th less . One way of dea l ing w i th such a s i tuat ion isto appea l to the van i ty o f the person concerned and ask them to se tup in te rv iews wi th m ore pow erfu l o r r e l evan t mem ber s of the o rga-n iza t ion .A more common problem is to f ind yourself p laced in a pub l icrelations programme designed as a one size fits all option foranyone f rom a foreign dignitary to a researcher. In such situationsthe rese archer should try to negotiate to see the key interview targetseither then or on a subsequent occasion. Again, appealing to theprofess ional skills of the public re lat ions person may work as theym ay wan t to show t ha t they can mee t any reasonable request.Sometimes you will be invited to lunch. This is to be avoided if atall possible ( lunch after the interview is a different mat te r as i t maylead to valuab le revelat ions in a more informal set t ing) . It is difficult

to know whether to concntrate on eating or ta lking ( it w ou l d be discourteous to eat l i t t le). It is difficult to take notes and eat, and thetape recorder m ay pick up a lo t o f background noise .Researchers need to be real is t ic about the n u m b e r o f interviewsthey can under take in one day. I f you manage to fit two into a day,given the scheduling constraints o f most l i te group members, youare do ing w ell. There is a need to allow f lex ibility in the t im etable asthe interview may be rearrang ed once o r twice. This can be infur iat-ing, but the researcher needs to be patient: the interview is moreimpor tant to the researcher than it is to the respondent . In a busycity, sufficient time has to be allow ed to travel f rom one location toanother. Researchers need to budget some buf fe r t ime anyway fora first review of their notes or recordings. lite interviewing is very /demanding for the researcher in t e rms of the level of concentrat ion jrequired and some recovery time is needed. H arrison recalls (2001,p.101) three interviews of forty-five to sixty minutes was mypersonal best; after this, I began to suffer f rom w riter s c r am p andm y b rain w as s truggling to func t ion .A s in all types of interviewing, the initial impression the respon-dent fo rms is im por tan t. Knowing hpw to dress fo r interview s poses -some difficult p roblems. The cardinal rule is that researchers shouldavoid drawing a t tent ion to themselves by their dress; it is the ques-tions to be asked t h a t the researcher requires the respondent to focuson. Very conservative dress is not always appropriate. Thomasrecalls (1995, p.15):

I rout inely don a (co nservat ive) navy sui t; however , in a com putercompany I fbund t ha t I stood out like a sore t h u m b by compar i -son to the pul lover sweater and s lacks that were the no rm . . .


8/17

lite nterviewing 211people chided me for dressing too much like a consul tant - ac o mme n t tha t w as t an tam oun t to aninsult .It is impor t an t to arr ive on t ime for your interview. Arrivinglate creates an unprofess ional impress ion an d m ay cu t dow n th e

t ime ava i l ab le .T he in terv iewee may, o f course, arr ive la te , some-th ing tha t reflects the b a l a n c e o f power in the interview. Thatcannot beavoided , but researchers need to sort out the logisticso fthe in terv iew beforehand . They should make sure tha t they k n o wthe location of the in terv iew s i te beforehand f this is par t icu lar lyimpor t an t in long st reets where they could find themselveshal f amile or more from their dest inat ion). Plan to arr ive in the vicini tyten to fif teen minutes beforehand: one can a lways browse in ashop or h a v e a cup of coffee. R e m e m b e r t h a t in some bui ld ingssecurity checks m ay take some t ime. Al low t ime to get f ro m therecept ion desk to the actual locat ion of the interview. These m ayseem trivial points , but there is nothing worse than arriving for aninterview t h a t has t aken much p l ann ing and ef for t to secure lateand flustered.

ondu t th int rvi wThe cardinal rule for conducting th e interview is: beprepared. It iswaste of time asking questions which can be answered from theInternet or published sources. The respondent will respect theresearchermore ifs/hecan show that s/heisfamiliar with the subjectmatter and can askwell informed and penetrating questions. Insteadof providing generalities or platitudes as answers, the respondentwill then engage in a frui tful dialogue with the researcher.

Most interviews are conducted by a single individual, but some-times research partners will conduct the interview together. In thatcase,it is im portan t to have agreed in advance who will cover whichtopics and what the general approach to the interview will be.Sometimes researchers are faced with more than one respondent.This is not an easy si tuation to handle,as it isdifficult to know whoto look at or respond to, particularly if the different roles of thosepresentare not ma declear.Thiscan always beprobed byasking, 'Soyour mainresp onsibil i tyis . . .?One of the most vexed practical questions inlite interviewingiswhether to use a tap recorder or take notes (if a researcher does usea recorder, they should take back-up notes anyway).Atape recordergivesacom pleterecordof them eeting,but it mayinhibit respondents


9/17

212 Research ethods inPoliticsfrom being asf r ank as they wouldbe in its absence. Cultural differ-ences can be impor tan t here. Experience suggests that NorthAm er i can r e s ponden t s a re u s ua l l y com fo r t ab l e with a t aperecorder , Bri t ish respondents less so ( a l though no t Irish r e spon-dents ) . With the exception of a few interviews at a snior level,Deakin and Par ry (2000,p.12)were able to tape their interviews a tthe British Treasury. Unlike earlier researchers on the Treasury,their experience was that the inhibi t ing effect was minimal and theadvantage of generat ing a text agreed between subject and inter-viewers was very substantial . 'Germn l i tes in general are likelyto agree to use of tape recorders in interviews . . . but the higherthe interviewee up the h ierarchical scale, the less likely you are toget a f rank answer (Paterson and Grix, 2000,p.18). Researchersusinga tape recorder should ask permission to do so at the begin-ning of the meeting. Good miniature models with at tached micro-phones a re avai lable on the market . Ensure beforehand that thebatteries are fully charged and that the machine is working prop-erly It should alsobenoted thatthe transcription t ime for a typi-cal taped interview will be mu c h longer than for an interview atwhichnoteshave been taken.lite s tudies, particularly of legislators, have been successfullycarried out using standardized questionnaires (see, for example,Presthus, 1973 and 1974). Equally, there m ay be occasions in theearly stagesof aresearchproject w hen a fewunstructured interviewswould help the development of the research design. Such an inter-view would take the form of an exploratory conversation around astructured theme. Thus, with this format 'the researcher suggests t hesubject fo r discussion but has few specific suggestions in mind'(Rubin and Rubin, 1995, p.5). This form of interview may be usefulwhen the researcher still has a l imited understanding of a topic.Their very lack of structure, however, means 'that th e interview s willnot be a very consistent source of reliable data across interviews(Leech,2002a, p.665).The semi-structured interview does no t follow one commonformat . M any qua litative interviews have both m ore structured andless structuredpars but vary in the balance between them' (Rubinand R ub in , 1995, p.5). Despite their variability, three gen eral propo-sitions may be made about such interviews.First, th e respondent will have a list of topics o r questions s/hewants to cover. These m ay take the form o f precisely formulatedquestions, as in the example provided by Stedward (1997,pp.164-5). More experienced interviewers may consider that stan-dardization of the form in which most questions is put is of


10/17

lite nterviewing 213

\y importance as the answers are no t usu

They are m ore com for table wo rking from alistof topics. 'In contrastto apredef ined qu es t ionnaire , th e interview guideisusedas acheck-list oftopics to becovered, a l thoug h th e order inwhich they are notdiscussed is not pre-o rda ined ' (Dev ine, 1995, p.138).In anycase, th equestions asked will vary depending on the particular expertise ofthe respondent .Second, there is a need to prioritize th e topics to becovered. T helength of interviews can vary considerably. Charles Raab (1987,p.120) probably f ilas the UK all comers ' record for an interviewlasting fifteen hours. Moretypically, abusy individual may set aside45minutesor onehour for aninterview ,atime which c annot usua llybe extended. Everything has to be covered in the t ime made avail-able. As Stedward says (1997, p.155) there is no point in asking aquestion if one can get the in f o r m a t ion elsewhere . In broad terms,quest ions or topics may be classified as 'essentiaP, 'necessary' and'desirable' . Essential questions are those which, if tley are notanswered , wou ld m e a n t h a t the interview is a fa i lure . Necessaryquestionsare those which areimpor tant to theproject and shouldb ecovered if at all possible. Desirable questions are those tha t can beinserted ift imeperm its .Third,the interview ermu s tnot seek toimposetoo rigid a frame-work on the interview. 'Qual i ta t ive interviewing requires l is teningcarefully enough to hear the m eanings , in terpreta t ions , and und e r -s tandings th a t give shape to the worlds of the interviewes. ' (Rubinand Rubin, 1995, p.7). T he in terviewer must allow the responden ttoopen up new topics that m aylead to reasofinquiry that had notbeen previously considered. The advantage of the informal in ter -view istha t it leavesthe investigation open to new and unexpectedin fo rmat ion ' (Daugbjerg , 1998, p.15). There needs to be awilling-ness to rede f ine the objectives and the scope of the project in termsof the mater ia l obta ined from the interviews. 'Interviewees canconfirm or reject one s initial interpretations and , consequently,m a k e the subsequent analys is more precise ' (Daugbjerg , 1998,p.16) .Within th e interview, a delicate balance has to be mainta inedbetween, on the onehand , coveringtheground theresearcher thinksis impor tant and , on the other hand, al lowing the respondent toopen up new reas wi thou t going off on irrelevant tangents. I t isques t ionable whether one should follow Bryman's (2001, p.312)advice that 'In quali tat ive interviewing, rambl ing or going off attangents isoften encouraged- itgives insight into w hat theintervie-wee sees as relevant and important . ' O ne wants to unders tand the


11/17

214 Research ethods in Politicsrespondent's definition of the situation, but one is also trying toplace that definit ion in the context of a theoretical ly informedresearch project . O ne cannot let the respondent entirely control theinteraction. Striking the right balance is one of the most difficulttasks in lite interviewing , given the b alanc e of authority betw een theinterviewer and the respondent, and the fact that respondents tendto do most of the talking. Harr ison notes (2001, p.102) , 'a carefulbalance needs to bestruck here between th em ethodologicalneed torem ain neu tral and the practical it ies of gett ing a respo nde nt talkor cut t ingthem short ifthey are going off on a tangen t '. It requiresa combinat ion of firmness and tact . Harrison advises (2001, p.102),'if you want to steer the discussion into a certain direction, youm us t never show your approval or disapproval of a voiced opinin,but signal your understanding or interest in a particular topic inmore neutral terms ' .The in terview exper ience may, of course, differ cons iderab ly int e r m s of the effects of the cul tural context wi th in which it isconduc ted . Ref lect ing on h er in tervie w ing exper ience in Japan,Koen notes (1997, p .32) 'I n Western countr ies , people wantprecise ques t ions and will give you as precise an answer as possi-ble. In Japan th is did not seem to w o r k . ' O ne s t ra tegy sugges tedby local experts was to ask very broad and genera l ques t ions .Another was to keeping asking the same question in a slightlydi f feren t f o r m .Proceeding from a basic mode l suggested by R u b n and Rubin(1995 ), the interviewmay beseenasgoing throug h aseriesofstages.These can be seen as a means of developing rapport between theinterviewer and the respondent. 'R a p p o r t means more than justputt ing people at case. It means convincing people that you arelistening,thaty ou areunders tandand areinterested inwhat theyaretalking about, andthat they should con tinutalking ' (Leech,2002b,p.665) .The opening stage begins with some informal conversation struc-tured by the exchange of business cards (having one is essential asthis is a key ritual, especially in East Asian countr ies , as well asgiving you in format ion about how you can contact the respondentagain) . One needs to establish an atmopshere of mutual trust with-out being grovellingor sycophantic . C ultural factors wil l be impor-tant in this stage of the interview. For example, informal directnesscharacterizes the USA , especiallythe west coast.There should then be some discussion of the purposes of theresearch. Respondents vary in terms of how mu c h they want toknow (they will normally have been sent a short description in


12/17

lite Interviewing 2 5a d v a n c e ) . On the whole, i t is better to of fer too much than too little,but wi thout consuming too much in terv iew t ime.Ear ly quest ions should deal with matters t ha t the in tervieweecertainly knows about and feels good about . In this respect unstan-dardized ap proache s are s imi lar to more s tanda rd ized fo r mso f in ter-viewing by s tar t ing with quest ions t ha t do not challenge ther e sponden to o m u c h and bui ld ing rappor t . Opening ques t ions also jprovide an opp or tun i ty to steer the in terview by ind ica t ing wha t theinterviewer knows abou t a l r eady an d what s /he wants to explore.For exam ple, Gran t 's op ening quest ion in an in terv iew with aCalifornian legislative aide was: Tm fairly familiar,Ihope, with thes t ruc tureof C al i forn ian pol it ics , I know in genera l how th ingsw o rk,w h a t Fm n o t sure about is how things work in relat ion to air qual-itym a n a g e m e n t , as far as the Senate iscon ce rned . There seem to bea n u m b e r of commit tees involved . 'A s the in terv iew progresses , one moves in to the more cen t ra l tques t ions . For example , in the in terv iew c i ted above , a key ques -t ion w a s How m u c h a re peop le c once rned in th i s abou t the wholeissue o f g lo b a l c l i m a t e c h a n g e ? ' As the in t e rv iew progres ses an dr appor t deve lops , one can move in to more d i f f icul t or sensi t iveques t ions . In the ci ted in terview , the leg is lat ive aide w as ask edabout a p r o m i n e n t state senator, Is he a bit politically isolated?T he in i t ia l response was , He t e n d s to be v e r y in d e p e n d e n t w i th inthe Democra t ic Par ty ,Iguess that 's agrac ious way to sayit , be foreproceed ing to a ve ry full an alys is o f the s trengths an d l im i ta t ionsof his posi t ion.Probing can play an impor tan t ro le in successful in terv iewing.' Insuff icient probing ind ica tes boredom or ina t ten t ion : too m u c hprobing and the researcher turns in to an inquis i tor ' ( R u b n a n dR u b i n , 1995, p .150) . The Rubins d is t inguishuseful ly be tween threetypes of prob. Continuat ion probes s ignal for more detai l . Theym ay evenbe executed n on-verba l ly ,but typicallywould bequest ionslike Sow h a t happenednext? Clarification probes a re used when itis unc le r wha t is being said . A n example would be 'Could you saya l i t t le more about tha t? ' Comple t ion probes re turn to someth ingtha t w as said earlier such a s 'You said ear l ier that . . .'A t the end of the interview, rspO nd en ts should begiven an oppor-tunity to say whether there an y aspects of the topic being discussed h h a v e n oshould be a qu ick summaryof any documen ts , ames and addressesor o ther in format ion to be exchanged subsequent ly . Casual conver-sa t ion may resume. This may lead to more v a luab le in fo rma t ionbeing m ade ava i lab le .


13/17

216 Research Methods inPoliticsBizarre incidents can sometimes disrupt an interview. Stedwardcomments (1997, p.158) It san obvious point, but no less impor tantfor that, to caution against interviews conducted under the influenceof alcohol. In practice, following this counsel may be difficult if therespondent arrives at the interview already drunk. There is no easyway to deal with incidents such as these, but they emphasize that

successful interviewing re qu ires good interpe rsona l skills. Theseinclude a certain level of self-confidence ba lanced by a w illingness tolisten and l earn f rom others.Respondents may ask to see a transcript of the interview or chap-ter drafts. Such requests canyield additional information or helpfulcomments. They do not usually lead to an a t tempt to censor theresults or exercise undue influence on them. In ethical terms, suchrequestswould seem to be reasonable and help to ensure access forfuture interviewers.Although the interview needs to have sonie s t ructure if the resultsare going to meet research o bjectives, i t is im porta nt to av oid domi-natingthe interview and to le t the respondent s voice come throug h.It is their perception of the situation that researchers can obtain f romthis type of interviewing.The contribution of interviews to the research m ay well change asit proceeds. In the early s tages, there ma y be m any them es that inter-estthe researcher w ho will be def ining, refining and exploring them.In the central partof the research process, the researcher will havebecome more selective and will be focusing on a more limited rangeof themes. In the f inal interviews, the interviewer may wish to checkemerging interpretations with respondents.Follow-up interviews with respondents do not generally yieldgood results. Particular points may be pursued by e-mail or on thetelephone. If one does go back to people on repeated occasions, al relationship m y become persona l reducing theresearch utility of the t ransaction. There is also a danger of becom-ing over-reliant on key informants who, al though they may beinside dpp^sters , a re offering their own par t icular interpretat ion ofevents which may capture the researcher. Indeed, this would be akey distinction between the use of interviewing by journal is ts andresearchers.

Once a standardized questionnaire has been designed and imple-mented (an expensive exercise), the researcher is stuck with theresults i t has yielded. With semi-structured interviewing, theresearcher can redesign the questions as the research proceeds totake account of new themes. This does not mean that the researchershould lose sight of the original goals of the research or its central


14/17

lite nterviewing 217theme,but it is possib le to tak e acco unt of new topics rev ealed by theresearch.Ultimately, lite interviewing is atechniqueinwhich, evenifprac-tice does not make perfect, experience is invaluable. One way oftrying to givestudents some prior experience is thro ug h role-playingexercises. Some people are, however, m ore tem per am en tally suited tobeing good interviewers than others. As Stedward observes (1997,p.152), 'given the nature of the research method, you should alsogive some consideration as to whether it is right for you. Techniquescan belearned and skills developed,but it is an unalterablefact thatsome people will never make good interviewers because they areuncomfor table with the process. ' Harrison warns (2001, p.102),cise face-to-face involvement with your subjects means that thisway of exploring controversial issuesis not for the sensitiveor faint-hearted' . Researchers should confront issues of this kind beforeembarking on their research.

nalysethe r sultsAfter the interview the researcher should read through the notestaken or listen to the tape as soon as possible. If any parts of thenotes are difficult to read, they should be clarified while the inter-viewis stillfresh in the researcher's mind. From the notes or tape, theresearcher should highlight any new points thatneed to be followedupinsubseq uen t interviews. Harrison notes (2001, p.102), thetran-scription process for me proved crucial in that it allowed me tobecome famil iar with m y data by milling over them time andagain' .Researchers should try and build in routines of self-evaluation.After each interview, they should reflect carefully on what is goingright orwrong. Are the interviews covering the topics theresearcherwants to cover? If not, is the researcher allowing the balance of theinterview to swing too m u c h in favour of the respondent?Alternatively, is the researcher adhering to too rigid an interviewstructure or lead ing the respond ent tow ards particular kin ds ofanswers? Is the understanding of the topic too broad or narrow? Isthere anything that could be excluded? Are there new angles thathave to be followed up?The researcher should then codethe data to pickout themes.Thepoint of this task is to m ak e a judg em en t abo ut the data in the lightof the theoretical framework. One can use a very simple codingframe, numbering topics or themes from one to and then putting


15/17

218 Research ethods inPoliticsa note in the m argin of a copy of the interview whenever they occur.There are computer programmes (the best know n are Nudis t andQualidata) which allow the researcher to analyse qualitative data,and these may be part icularly useful if the researcher is engaged insome kind of content analysis. If the research is concerned with theinterpretation of meaning as it is constructed by respondents, suchanapproachmay be too mechanistic.Particular care needs to be taken in writing up data of this kind.It is essential that early on in the analysis, as part of the generaldiscussion of methodological issues, the researcher provides a shortaccunt of the way lite interviewing techniques were used in theresearch. This should include a discussion of their importance rela-tivetoother techniques. Publications should include atable catego-rizing the respon dents interviewed (e.g., Comm ission officials 8, UKgovernment officials 6, Germn government officials 7, EU levellobbyists 6, and so on).Quotation from the lite interviews can do agreat deal to enlivena pubication. However, there is a risk of making excessive use ofquotations.One does not wantsectionsof a book,article or thesisto be nothing more than a series of quotations with linkingsentences. In particular, researchers should not place too muchreliance on any one respondent, however e loquent they may be orhowever much their viewsmaysupport the argument being made. Ajournalist may be happy with selective quotations that happen tosupportthe particular line taken in the story, but researchers have adutyto consider conflicting exp lanations and c ontrad ictory p oints ofview.on lusionslite interviewing m ay lead to good (topical, interesting, incisive)writing about political themes and that is no bad thing given theturgid character of much contemporary political science literature.However, is it good social science? This in part depends on wha t onedefines as good social science. Critics portray qualitative research asbeing unrepresentative and atypical. Its findings are impressionistic,piecemeal , and even idiosyncrat ic (Devine , 1995, p.141).Quantitative data sets can be re-examined by other researchers.Records of qu alitative interviews can also be opened for scrutiny byother researchers, although problems of respondent confidentiali tymay arise. Even if they are available, they may not reveal how thenatu re of the interaction b etween the interviewer and the respo nden t


16/17

lite nterviewing 219has affected thedata. Two researchers might interpret the same data in very d i f fe rent ways.However, if one is interested inactors perceptions of theworld inwhich they live, the way in which they construct their world and theshared assumptions which shape i t , there is much to be said for themodel of the lite interview as an extended conversat ion. The advan-tages of qual i ta t ive research are c lear where the goal of a piece of research is to explore people s experiences, practices, vales and atti-tudes in depth and to establish their meaning for those co n cern ed(Devine, 2002, p.207). The way in which is this exploration isconducted is , however indirect ly, informed by the theoret icalconcerns of social scientists and is u n d e r t a k e n by a researcher seek-ing to ga ther data systematically to pursue defined questions in areflective fashion. While quant i ta t ive research is usual ly rel iable,qualitative research tends to bevalid (Devine, 1995, p.146). It iscertainly not difficult to identify classic political science texts whichhave been based on li te interviewing (e.g., Heclo and Wildavsky,1974).The reali ty of niodern democracy is that many polit ical decisionsare taken by small groups of highly qualified and knowledgeableindividuis. The t e rm pol icy community may have fallen out offashion, but many under-ven t i la ted (Heclo and Wildavsky, 1974,p.xx) pol icy communit ies remain in existence. The shared assump-t ions and meanings which inform these pr ivate worlds still r equi reexploration, and l i te interviewing remains the most appropriatet echnique. It is a technique whose exercise benefits f rom the accu-mulation of experience, but it is also accessible to students startingout on their research careers. li te interviewing brings the wor ld ofthe pract i t ioner and the academic together in a hopefully f rui t fulm ut ua l d ialogue.

This chapter has tried to explore some of the issues and p roblemsthat arise in lite interviewing and to offer some practical guidance. lite in terviewers must be p repared to explore the world of the[ respondent sympathet ical ly w ithout being cap tured by it . They must have a research design that is flexible and open to new ideas and

interpretations, while st i l l being informed by an over-arching theo-i_reticalf r a m e w o r k and a central research question. There is a limit tothe utility of the advice that can be o f fered, as much d epend s on theability of the researcher in m anaging a series of pressures (of time, ofin terpre ta t ion and of judgement) . Never the less , skilfully executedlite in terviewing can add significantly to our stock of poli t ical 1knowledge and understanding.The fur ther methodologica l development of lite interviewing


17/17

220 Research ethods inPoliticsmust be related to the current emphasis on see ing qual i ta t ive andquantitative approaches to research a s com plementary rather thancompetitive.Devine notes(2002,p.207) that To date . . . there havebeen few genuine attempts in polit ical science to bring quantitativeand qualitative data together to address consistencies a s well asinconsistencies. This theme is one that is re turned to in Chapter 12.

(Research Methods in Politics Cap 9)011

Documents