by Mark Williams
For the second phases of the study, the data collection method I am using is a semi structured interview/focus group. This is what Wilson (2006) calls "free format" conducted like "natural conversations".
Wilson notes that there is an issue in social science of the awareness of being a research subject. Participants might alter their "expressions of beliefs and attitudes...[and] behaviours" skewing results, so a balance has to be found between data collection procedure and naturalistic response or highly and less structured.
In the free format interview questions may be phrased differently depending on what has gone before.
Wilson notes though that a "completely unstructured interview is impossible"; conversations have a structure and the interviewer still has a focus/foci
What is the legitimacy of the interviewer? In my study, the students I am talking to know me as a tutor of technology eg. Social Media/Marketing . Whilst I offered participants cake and tea as a reward for helping, many were happy to contribute to a study that they felt was important, which corroborated statements from the survey where students said "talk to us about tech".
Even if respondents are asked the same questions, there will be a range of conditions that effect the response and interviewers "must probe or prompt to ensure a full understanding of the question"
The chief advantage of the self-administered questionnaire is that they are far quicker to administer and get results than any other semi-structured investigation method. It "allows a large number of respondents to be interviewed within a practicable time span and within feasible limits of cost"
However, response rates are often low unless the subject engages the respondents interest or is of value to them. The researcher also has no control over the conditions in which they are responded to... they could be completed in any order and not by the recipient.
Highly structured methods are also criticised for an "unaceptably high degree of reactivity" and "unnecessarily constraining" the respondent. Even with self-adminstered questionairres there exists a degree of social interaction, the researcher emails out to the students or talsk to them face to face and willingness to respond can affect validity.
Respondents may also have varying degrees of digital practice or computer literacy. The order of the questions needs to be considered. I initially placed demographic data at the top to get it out of the way, but realised that this might be "uninteresting" and also "sensitive" in nature which may cause resentment.
Social science often relies on a degree of naturalism, in particular interviews will follow a conversational pattern where it is expected that participants will take turns, rather than the interviewer dominating. Less structured methods allow for a lesser degree of reactivity and for the participants' beliefs and experiences to take a centre stage. This can be at the expense of reliability as it is often impossible to recreate the conditions to produce the same results.
In addition to procedural bias, there also exists personal bias on behalf of the researcher.. the varying interactions with participants around the research. I infact, met with one partcipant beforehand and had an email conversation with another.. another I had known for 3 years.
Wilson, M. 'Asking Questions' in Sapsford, R, & Jupp, V. (eds.) (2006) Data Collection & Analysis. London: Sage