POSITIVISTS : in favour of Sociolgy as a Science
Positivists argue that natural science is a good thing because it offers the best way of obtaining knowledge. In sociology Positivists argue that Sociolgy can and should follow the natural sciences. That is, we should regard society as a real thing, develop theories about the laws of social development, and test those theories with quantitative data (data in the form of numbers eg. official stats).
Positivists argue that quantitative data is best because its most likely to be valid (because its unbiased), reliable (because the research can be repeated) and representiative (because it uses large random samples).
An example of Positivist thing would be The Black Reort: the lower the social class, the higher the death rate.
INTERPRETIVISTS: Against sociolgy as a Science.
Interpretivists argue that the social world is so different from the natural world that sociolgy cannot in any way be scientific. The social world is made up of what people take up to be true. So sociologists should try to generate Qualitative data (Deeper data which has more understanding such as interviews and open answered questionnaires).
Interpretivists argue that their methods produce validn data (because they understand what people are doing), reliable data (because tehy can classify their results after theyve collected the data). Instead of representiative samples, they prefer samples chosen on theoredtical grounds to study key issues.
An example of Interpretivist thinking would be Atkinson's Discovering Suicide. His sole purpose is to understand the ways in which coroners classify deaths.
REALISTS: in favour of Systematic Sociolgy.
Realists accept that sociologists have to understand how people see the world, but tehy argue that sociological research should still be done systematicly. So Realists argue that data should be both qualitative and quanitiative. Methods should be chosen as appropriate, not for theorectical reasons.
Validity is obtained by being able (in the phrase the British sociologist Anthony Giddens pinches from Philosophy) to 'go on'. Reliabilty comes either from replicating or by classifying findings under common concepts. Some studies are best done with representiative samples, others without.
An example of Realist would be Sue Sharpe's, Just Like a Girl. She understood the way girls understood the world (validity). She brought the findings under common concepts - husbands a priority or jobs a priority. And she wasn't representiative, but if working class girls were getting more ambitous, its reasonable to assume that the same was true of m/c girls.
Whether you call this sort of research 'scientific' or not, doesn't really matter, according to interpretivists.