A case study is performed by taking one individual and studying him in depth over a period of time (e.g. Freud's "Little Hans"). As a means of gathering data, it has some points in its favour:
In an observational study, the experimenter obtains information by observing how people respond to certain situations. In this case it is important to distinguish between observations and inferences: you can observe someone talking but can only infer that someone is listening.
Naturalistic observation involves observing behaviour that takes place in a "natural" day-to-day situation, without the experimenter being involved at all. For example, an observational study could be conducted to gather data on the behaviour of drivers at traffic lights. Typically this will be covert, or undisclosed i.e. those people being observed do not know that they're being observed.
Participant studies can be made where the experimenter observes behaviour by becoming a member of the group he wants to observe. Festinger (1956) used this technique in his work on cognitive dissonance, when he joined a group of people who were convinced that the end of the world was approaching. In this case, the group assumes that the experimenter is "one of them", and the experimenter can observe the group's behaviour at close quarters. Obviously, this is also covert.
Overt, or disclosed observational studies occur when participants are fully aware that they are being observed. In this case, the hope is that once they become used to the observation, they will behave as if it were not taking place.
The observational method has certain strengths:
While observation might be able to tell you what people are doing, it can't tell you why they're doing it. A questionnaire is a way to obtain this information without having to conduct an interview, and so it may be a useful way to gather information on a large number of subjects without having to spend a lot of time with them.
Questions on a questionnaire may be open or closed. A closed question has only a limited number of possible answers, e.g. "Do you prefer coffee or tea?". An open question invites the respondent to supply information in free form, e.g. "Why do you prefer coffee to tea?". Closed questions typically provide quantitative data (i.e. numbers), while open questions generally give richer, qualitative data.
Because desiging questionnaires can be difficult, it is usual to conduct a pilot study by using an initial set of questions which is used on a small group of people, and then basing a final questionnaire on the results or feedback from the pilot.
Advantages of questionnaires:
A correlational study attempts to demonstrate a relationship between two variables by gathering data about those variables. For example, a study might attempt to show that there is a relationship between the amount of pets someone has and the amount of television he watches. There is no attempt made to establish whether one thing is causing the other.
A positive correlation occurs when both variables either increase or decrease. For example "the more pets you have, the more tv you watch". Or "the fewer windows in a house, the less butter is consumed by the household.". A negative correlation occurs when one variable increases as the other decreases. "The more sweets I eat, the less distance I run".
Correlation is measured to indicate the strength of correlation, with values ranging from -1..0..1, where 0 indicates no correlation, and +/-1 indicates perfect correlation. Positive values are used to indicate positive correlation, and negative to indicate negative correlation.
To establish the degree of correlation, a scattergram is drawn with the x and y axes representing the two variables. A 45 degree line rising from left to right would indicate a correlation of +1. A 45 degree line falling from left to right would indicate -1. A less clear line (i.e. the dots roughly form a line but not perfectly) would indicate lower strengths of correlation, and a mass of dots with no discernable line would indicate 0, or no correlation.
See "A Level" book p168-170 for more info on correlation and scattergrams.
A correlational study, by definition, does not provide information as to cause-and-effect, it just says "x is related to y". But it can be useful to rule out any relationship between two variables. For example if the correlation between "number of pets" and "number of windows on the house" is 0, we know that there's no relationship between the two. On the other hand, if a correlation is shown, then this might prompt further study to establish cause and effect.
Back to class notes index page