Bias in qualitative research designs

Quantitative researchers speak of 'bias' and 'generalisability'. Qualitative researchers address the same issues, but seldom use these terms. Like any other researchers, they are concerned with the extent to which their research is valid and representative of the area being investigated, but the way in which these issues are addressed is substantially different from the way in which quantitative researchers go about the task. Here we offer our views on the process of doing good qualitative research. These views underpin our decision-making when papers using qualitative methods are submitted to this Journal. Qualitative research, like any research, starts with a systematic review of the literature to show that the topic being studied is significant and unresolved. Some forms of qualitative research focus on existing data, for example from archives or media reports. Our focus here is on interview studies. Both formats face the same problem, that of selecting and justifying a sample, and then defining the extent to which the results apply to other groups or settings. Sampling in an interview study is a bit like collecting a slice of life and taking it into a laboratory for dissection and analysis. It makes sense to select a slice in which the topic under investigation is present in high concentration. To take a current example, let us say that we are interested in women's use of hormone replacement therapy. We might start by enrolling five women attending a menopause clinic and conducting open-ended interviews about their experience. Analysis starts immediately, sorting the women's accounts into categories that can come from social theory, the literature review or from analysis of the data itself. If necessary , we return to the field to enlarge the sample until we are sure that we understand the experience of these women. When we reach this point of saturation, we cease to enrol further research participants of this kind. There are situations where this first slice of life is all that is needed. A small sample of people can all have very similar responses despite being from very different backgrounds. Sometimes this happens because the experience that they have in common is of overwhelming significance in their lives, such as being caught up in a bushfire or a plane hijacking. More commonly , we find differences of opinion in the sample. In such cases, we need to return to the field to sample for these different experiences. …