|Route map: HomePublicationsEvaluation CookbookInformation Page|
|Transcribing Evaluation Data|
The manner of transcription may matter whatever techniques are being used to analyse textual data, but consistent transcription, following a basic set of rules will be even more necessary when using a CAQDAS (computer assisted qualitative data analysis software) program, or even word processing search tools, to assist in the analysis process.
SECTION 1. Aims to list some basic Do's and Don'ts - which will help reduce the necessity for later changes to data or preparation time - whatever software package is being used. It would be helpful to give this list to a transcriber - even if you are not certain that a CAQDAS software package will be used.
SECTION 2. Lists particular points which are software specific, without placing too much emphasis the full extent of instructions which might be applicable e.g. for data preparation using NUD*IST. This aims to be a short list providing pointers only. More detailed information will be available in manuals or from software specific workshop curricula.
1. General Guidelines
Data are often inconsistently transcribed within one file. Much of the final preparation which may have more to do with analysis stage decisions and will have to be left to the researcher. However he/she will be saved a lot of tidying up work if some simple basic minima rules are applied at an early transcription stage. This list is general and NOT 'software-specific'. So these are not all the steps required for the preparation of data for individual software packages.
Some of the guidelines below are even more relevant for data with structure - e.g. surveys containing open ended questions, focus group data, structured interviews, etc.
NOTE: none of this need apply to data being prepared for Code-a-Text, which can use rich text, or word processed files.
DO - always make spelling, spacing etc., of repeating speaker identifiers, question headers, section headers, topic headers, absolutely uniform throughout text, e.g. QU1: OR Q1:, NOT a mixture of both. You may need to depend on this uniformity when performing text searches and saving the results. It is easier to use text search tools which look for exact strings of characters, not approximations.
DO - create clear line space between paragraphs. Not always a requirement but will help with e.g. Atlas/ti.
... BUT! ...
DO NOT - have a line space between speaker identifiers, topic headers etc., and the text associated with them. (This will be more relevant in some CAQDAS software than others).
DO NOT - depend on Word processing display characteristics to indicate things (emphasis etc.) about your text e.g. bold, italic, underline, e.g. do not use emboldening or italics to identify the respondents speech.
DO - use a clear speaker identifier instead, preferably in UPPER CASE. This will allow CASE SENSITIVE searches for the speaker identifiers. E.g. IV: or CAROL:
DO NOT - use bullet points (unless you want them to act as section headers in NUD*IST - and you must understand the significance of 'Section headers' as they are used in NUD*IST before deciding this).
DO - be clear about the amount of 'preparation stage' thought necessary for your data; this may vary enormously depending on the CAQDAS software being used. Try to find out how this varies, by asking or experimenting or reading the manuals!
DO - with the above in mind, if possible before transcribing too much data, prepare a small pilot project with one or two files, inside the CAQDAS software. Do some coding and retrieval, and text searches, to test that the formatting of the data seems to work or if you see where improvements can be made - they can be applied at an early stage of transcription.
DO - save an original, 'safe' copy of data, with the above thoughts in mind, in Word Processed format. You can always go back to this if subsequent SAVE as TEXT options go wrong.
DO - transcribe in a fixed width font like Courier or Courier New. Make the points size 10 or bigger. Use the usual default margins of 3.17 cm either side or 1.25 inches. This will usually present a line length which sits well in most text windows - if not, some software packages, like Ethnograph have editors, or format processes which convert the transcribed files for you into convenient line length.
2. Peculiarities of transcription requirements for a sample of individual CAQDAS software programs
WINMAX Pro 97 data will be better imported to the software WITHOUT line breaks. This will assist autocoding text selection later. Do not put a hard return after a Speaker identifier - begin text on the same line. Insertions of special syntax in transcript will allow the cutting of multi case files into smaller files within files. See the software manual for further details.
QSR NUD*IST will require you to make a decision about what the minimum codeable chunk/segment will be throughout your data, before data is imported into the software. The 'text unit' is defined by where the hard return goes. So using a 'Save as Text' option which introduces line breaks at the end of very line will result in LINE text units. Or interactively insert a hard return at contextual breaks or sentences, pauses etc., giving codeable segments of variable length. Or keep hard returns in all natural places, i.e. paragraphs, sections, speaker sections etc. Note: Section headers have a specific purpose in NUDIST. See the software manual for further details.
© All rights reserved LTDI and content authors.
Last modified: 25 March 1999.