How we change what others think, feel, believe and do
When designing social research, one of the biggest questions that will affect your research is what data you will collect. There is generally a return on investment -- the more effort you put into collecting data, the more chance you have of creating useful results. There is also a law of diminishing returns. 'More' does not always mean better and 'enough' to meet your goals is often the best option.
There are two basic ways of calculating with data: counting how many of a particular item you have (eg. 'How many people prefer Sudso soap'), or doing more complex analysis.
With scores, you have more choices, as in the sections below. With frequency counts the choice stops here.
There are four basic types of data you can use: nominal, ordinal, interval and ratio. Depending on the experiment, these can be increasingly difficult to collect, but they give increasing rewards in what may be concluded.
Interval and ratio data may allow for parametric analysis, whilst nominal and ordinal data limit the analysis to non-parametric forms.
Independent variables are those that you control, whilst dependent ones are those which change as a result of this. When considering multiple variables, you are often looking to understand the relationship between these.
Variable dependence depends on what you want to discover or prove. To show cause, you may have one independent variable and one or more dependent variables, with which you may seek correlation (and later infer cause).
Sometimes you can measure two or more independent variables and look for
relationships between these (and perhaps, later, discover a common cause).
In a study, you look at what is there, seeking to discover from simple observation. You can look at people in different contexts and measure them in different ways, though you usually want to avoid change what you are watching and so may take a carefully non-invasive position.
In an experiment, you have people, measures and treatments. If you change any of these or change sequences, you have different type of experiment.
Repeated measures are used in experiments where you apply the same treatment to the same group, for example in a before-and-after test.
Independent measures are used where the same measure is applied to a range of different groups, for example where an ability test is applied to separate groups of men and women.
You can use a combination of repeated and independent measures, for example where you measure men and women for intelligence before and after a brain-stimulating treatment.
Parametric data follows particular rules and mathematical algorithms. As a result detailed conclusions may be drawn about the data. Experiments are thus often designed to use parametric data.
Research that does not create parametric data is non-parametric. Much research data is of this form and useful information can often be gained.
There are very different parametric and non-parametric tests used in analysis, depending on the type of data you chose during the design.
And the big