Checking for bias in data collection

Data from experiments, survey questionnaires and interviews can be influenced by either the context of the study, the respondents themselves, or the researcher.  The term "bias" is often used in this context, but the term is ambiguous. Technically meaning "leaning" in one direction, it is often used to refer to respondents or researchers having pre-conceived ideas or an ideological disposition.  What we mean here by bias is anything that can "contaminate" the picture you are trying to get of either subjects' behaviour or their attitudes and beliefs.

Here is a checklist for the researcher to check for factors which could be influencing or contaminating the data: (adapted from Plummer: 1983)

Source 1: the Life history informant
Is misinformation (unintended) given?
Has there been any evasion of the question?
Is there any evidence of direct lying or deception?
Is the informant trying to present a false front or impression?
What may the informant take for granted and thus not reveal?
How far is the informant seeking to please the interviewer?
How much has been forgotten or overlooked?
Source 2: the Social Science researcher
Could any of the following be shaping the outcome?
  • attitudes of researcher; age, gender, class, race, and so on
  • presentation of researcher; dress, speech, body language
  • personality of researcher: anxiety, need for approval, hostility, warmth, and so on
  • attitudes of researcher: religion, politics, tolerance, general assumptions
  • scientific role of researcher: theory held, and so on (researcher
Source 3: the Interaction
The joint act needs to be examined. Is bias coming from:
  • the physical setting - social space?
  • any previous encounters or interaction?
  • reactions to non-verbal communication?
  • reactions to vocal (verbal) behaviour?

It's important to understand that bias is inevitable and normal. The problem is not the presence of biasing factors, but that the writer seems unaware of them, and interprets interview or questionnaire data as a "true account" of reality.  This can lead to exaggerated claims based on the data.

Go to Values and Stance (Academic Writing) for more on this.