T h e lirscarch 1nter.i.iev- T//is dozible-clzecking is an essential part of tile rigour of your ciiin!l~.ri,r,and needs to he reported as part of your procedure. You don't have to accept pour peer's judgement as the correct one. Instead, you use it as a basis for reviewi?zg-yoz~r oiol/ jirdgen2~nt.r.Typically: you find you accept some of the 'disagreements' but not others. That is perfectly acceptable, tlle essential point being that you have soz~ghtthat kind of \-alidation of your judgements. Asking someone to parallel your categorical analysis is rarely feasible or reasonable; if you ecidence your categories and acknowledge thcir f~~ndarnentalsubjectivity that is ~~suallyenough. Writing up Interview Data The great virtue of a rigorous content analysis is that it immerses you in the detail of your substantive findings. The process of classifying and categorizing, difficult though it is, has a disciplining effect not just on your spreadsheets but on your own intellectual grasp of your material. This organization has the effect of enabling you to see more clearly the significance, particularly the general significance, of what people have said to you. Your mind will, therefore, have been working on two levels: the task of categorization and the task of interpretation. The interesting thing about the latter is that it is a process that goes on without any prompting from you. At the end of Chapter 8 we said that categoriza- tion is characteristic of human intelligence, but so also is interfiretation: it is part of the everyday process of living. And by the same token it is not entirely a self-conscious or even a conscious activity. When you are dealing with a wide range of rather complicated information there is a good deal of unconscious work of this kind. dndyou haze to allow timefor this unconsciousprocess to ope~ate. This is just as well. Having carried out a comprehensive content analysis you will feel the need for a break. A two- week interval will not be wasted, because at the end of that time you will come back to your material ~ i i t ha fresh eye and a better organized ~nilzd for the writing-up process. 73 The Research Interview Getting down to the business of writing An organized mind still needs organized material - and, of course, to a large extent the former emerges from the activity of constructing the latter. Let us review the process of organizing your data. It started a t the level of question development - identifying questions of real importance that are likely to call forth an interested response on the part of the interviewee. If there is substance in the question there will be substance in the answers to it. And if each question deals with a different - topic there will be a discernible difference in substance between them. So when you come to write u p your material (the quota- tions you are going to use) your organization is there in a readily accessible format, at two levels: the main question you posed, which )7ou can treat as a main heading - perhaps a section or even a chapter; and your categories, which you can treat as sub-headings. Since you have coded your quotations on the spreadsheets to the specific transcripts they come from, you can refer back to them if you need more 'context' a t the point ofwriting up. Writing up The essential character of writing u p interview data is to weave a narrative which is interpolated with illustrative quotes. Your task is essentially to allow the interviewees to speak for themsel~~es,with linking material which does little more than ensure continuity and point u p the import of what the interviewees are saying. A good example of this is the study of single lone mothers by Burghes and Brown. previously cited. -4more extended quotation is given in Box 9.1, exemplifying exactly the points that have just been macle. IVriting up Interview Data Box 9.1 Other he@ I Other sources of help mentioned by the single lone mothers included a local family centre, a foster inother and other mothers living in the same hostel. 'I was quite lucky . .. even though I didn't have support from my fainily ... I were in a hostel for young mothers when I first had J . .. we all rallied round together ... they helped me a lot . . . told me about their experiences about bringing up a baby.' Four mothers also referred specifically to help from their boyfriends' mothers, while two cited support received from the boyfriends themselves. Grandparents and friends were also mentioned, sometimes as part of the assistance received fi-om an extended family consortium: '... his sister ... used to come over and talk to me. She used to stay in the house ~ : i t hme till him or his mum came back from work so it weren't so bad ... she either watches him or me mum watches him or S watches him when I go out.' '. . . I had all my friends and they helped me.. .. mum, my auntie and then my cousins would come and take him out and things like that.' 'I was living wit11 a full fainily so there was a lot of help there. So it'd probably be a totally different story if I was on in-y own.' Sometimes, however; support from estended family and friends had not lasted beyond an initial burst of enthusiasm following the birth: '.. . for the first six months of having her it was fine ... people were very willing to babysit for a new baby; but then xvllen it got to the teething stage, the friends disappeared.' '. .. Mrhen I had her I had a lot of help but now she's one .. . they don't want to know. And it seems harder for me all the time.' The Research Interview You don't include all the relevant quotes -just enough to give the range and i7ariety of the answers. And if there are 'discrepant' quotes you add them as a qualifying insight. You can see how Burghes and Brown have done this. Quantitative analyses One way in which you can reflect the generaligi of the kinds of statements quoted is to cite how many of the interviewees nlade that point (or one like it) or how many made different or contradictory points. Burghes and Brown make an obser- vation of this kind in their connecting narrative in Box 9.1. That is usually sufficient but there are occasions, i.e. when the picture is a little more complicated, to set out the different pattern of statements in a tabular form. If it is easier to 'see' the point like that then it isjustified. Numerical relationships are sometimes clumsily expressed in verbal form. The balance of quotation and linking narrative It goes ivithout saying that the quotations you select should be iepreselztatir~eof the total range. Soine people will have made the same point in a more vivid or compelling way than others; and, of course, you should select those. What you have to guard against is selecting quotations that suit your particular preferences or present a neater picture. The best lies are half-truths and caref~~llyselected quotations can totally distort the picture. An honest balance has to be struck there. An equally important balance is that between quotations and the ainount of linking narrative. An approximate prac- tical guide is that quotations should make up not less than a third of the text, but not more than half. LVriting LIPIntervieuj Data Under the category sub-headings you will need an intro- ductory paragraph or tw-o, but then you should let the interviewees take over, with no more than a n~eanii~gful linking between the quotations - and sometimes you should simply cite several in succession. This linking should be like a jamerwork that holds the quotations together - but it is these (what )-our intervieu-ees have said) which should make the point. In terms of emphasis, the material should be 90 per cent from the interviewees. Reviewing your selection of quotations We have already cautioned against the risks of 'selective bias' - selecting to favour a particular emphasis. That is not necessarily a consciously corrupt process, but you have to guard against it all the same. Your selection inay be unbalanced for no malign reasons whatsoever. There can be a 'drift' in the quotations that catch your attention when you are writing up, wllich results in a completely unintended bias. HOWdoes one guard against that? The basic procedure is a simple one. You have your spreadsheets with the category headings and the colun~nsof statements from individual interviews. As you use a quotation in your write-up you should highlight it. That tells you which ones you have used and ~ r l ~ i c hones you 11al.e choseil not to. At various points you should scan the selected and non-selected quotations to check your justification for your choice. IC'hat you are after is a balanced representation. Some of them will make the 'representative' points better than others. However, there may be shades or nuances that are not caught by just one or two quotations. The Research Interview How many quotations? There are two main determinants here. If all or most of your interviewees have made the same kind of point then the commo~zalit_yof this needs to be demonstrated by a range of quotations: one in isolation might convey the incorrect message that it u-as a 'one off, although you should or could indicate that x number of people said essentially the same thing. But the number of quotations is still part of the impact of quality. The second point is that although people may be saying api,,-~ximate[~the same kind of thing there will be shades of opinions, important variations of detail which can only be conr:eyed by a range of quotations. In a sense these state- ments are 'unpacking' what the category heading signifies. When you scan the highlighted items on your spreadsheet you may find that you have been too selective. It has to be borne in mind that you identified these initially as having 'something to say'. If they don't add anything to other quotations then it is perfectly fair to omit them. But you have to justify that choice to yourself and to others. Basic to the kind of research that semi-structured inter- ~ . i e ~ ; sare a part of is the trustt~~o~t/zinessof procedures. This means more than being honest and checking that your data are sound, and acknowledging their limits. It also means that the processes, like data analysis, are open for inspection. This kind of open accounting is part of what E.G. Guba and Y.S. Lincoln (1981j in Efective Evakiation, San Francisco: Jossey- Bass, call the 'audit trail' - a trail that someone else, the 'auditor', can follow to see how you reached your conclusions. Quite apart from the integrity issue, by documenting or preserving the records of your process of analysis you can, if necessary, backback to check on your chain of evidence and the reasoiling derived from that. In the same way that traditionally 'scientific' quantitative researchers might check back on their calculations (because IVriting up Interview Data the results are of a level that raise questions,),so naturalistic researchers may need to review the 'calculations' that led them to draw the conclusions they have. All researchers must expect to be challenged on their findings: your justification is only as good as the means by which you achieved them. As we have said before, data do not just speak for themselves: selection and interpretation are required but these should be kept to the minimum necessary for the implications of the evidence to be apparent.