How we change what others think, feel, believe and do
Control vs. authenticity
When designing a social research experiment, the question of control is not an easy one and there is an ongoing debate about which is best or most appropriate in any situation. Often, the question is a trade-off of scientific control against real-world authenticity.
Three general approaches are discussed below. As the table here indicates, you cannot have high control and authenticity together.
Scientific social research experiments take method from the natural sciences (physics, chemistry, etc.) and apply them to the social sphere. The basic principle is to hold all variables steady except for those under investigation.
The scientific method includes:
Positivists promote the scientific method as the only valid method of study and creating true and factual knowledge.
The problem with the scientific method is that, although this method works fine in a chemistry experiment and quickly leads to conclusive and reliable results, this is not so straightforward in social settings. People are thinking beings and if you try to control things around them, they will change how they behave and hence affect your results. This is the subject-object problem. People cannot help but affect one another and any connection between the researcher and the researched can bias results.
The scientific method assumes a closed system, in which unwanted variables can be excluded or held stable. The problem with people is that they do not think the same thoughts or respond in exactly the same way no matter how hard you try to control the situation.
There is also an ethical question in social experiments, for example when you change things without a person knowing and particularly if this may be harmful to them in any way.
A very common way of getting around the classic experimental problem is to ask the target person questions, rather than directly trying to manipulate them. This assumes they can answer fully, honestly and without bias.
Surveys ask respondents to fill in a form by themselves. Traditionally on paper, they are often now done on the internet. They may ask the respondent about attitudes, events, beliefs and so on.
Surveys are often standardized, in that the questions are tested and calibrated beforehand. Psychometric instruments evaluate individual differences, comparing them against a standardized scale.
Surveys, although qualitative in subject-matter, often give quantitative data that may be statistically analyzed.
Structured interviews are little more than researcher-read survey questions. What this does allow for is clarification of what questions mean, branching in the survey and the use of careful probing.
Open interviewing is a looser method and may resemble a 'general chat', although the interviewer has an agenda and will try to steer things in the 'right' direction whilst avoiding a forced situation.
Open interviewing is also known as life history interviewing or unstructured interviewing.
A big problem with questions is that it requires self-reporting and thus is affected by the respondent's views.
Interviewing directly includes the researcher in the process and their bias might creep in. Social desirability and other effects that change how respondents behave can also creep in.
Surveys remove the researcher to a remote position. This can still lead to bias and other effects. It also means
The lowest amount of control is applied in methods where the researcher has the role of 'witness', carefully not intruding and perhaps remaining hidden such that the target does not know they are being observed (and hence giving them no reason to change how they behave).
Participant observation includes methods where the researcher may watch people in natural surroundings, converse with them about 'what it's like', take photographs and even live in the same circumstance to get some sense of the experiences.
Whilst observational method minimize the chance of bias, they are pretty singular experiences and depend on the observational powers of the researcher as well as their ability to be fully objective (which can be difficult when they would like to report something interesting). This makes it difficult to generalize conclusions to wider contexts, a factor that prevents any reliable 'laws' from being identified.