At its simplest, content analysis is what happens if you take a political speech and run it through the ‘search’ function in Microsoft Word to see how many times the presidential candidate used the word ‘loser’. In its more complex form, it allows us to answer multi-part questions about social trends, cultural climates, and historical or current perspectives.
Requirements of Content Analysis
To be valid, content analysis must be objective, and all methods used must be repeatable by an impartial researcher to get the same results. It must be systematic; the rules used should always be consistently applied. We also want it to be quantitative (numerical); this opens it up to statistical analysis and means we can give precise, succinct summaries of our results.
It’s this last property—the quantitative feature of this analysis—that really sets it apart from other ways you might look at text, such as critical reading. This quantitative nature of content analysis is what gives it its power, and allows us to condense huge amounts of communication into data that is concise enough to can be studied and described.
Examples of Content Analysis
This type of analysis has been used to answer all of these:
- Can the literary success of a novel be predicted by looking at content variables, and what are the content characteristics of best selling novels?
- How are women and their roles portrayed in media?
- From a mass periodical fiction standpoint, what are society’s values, and how have they been changing in the past years?
How do I Perform Content Analysis?
Content Analysis is typically a multi-part research project. You’ll start by defining your question and hypothesis, and go on by defining a ‘population of interest’; i.e., your data pool. For instance, if your question was “how has public sentiment toward the police force changed in the past ten years”, you might choose small town newspapers, letters to the editors, and magazine editorials for your population of interest. You could also choose online media such as personal blogs.
Then you’ll want to create categories, and decide what units of analysis—or blocks of meaning—your research project will tackle. Will you be looking at words or phrases, or do you want to concentrate on paragraphs and entire newspaper sections?
How will you classify your data? It is easier to conduct a content analysis that focuses on format, but analyzing for content gives you more opportunity for breadth. Decide how you’ll map, or code, textual material into quantifiable coding your computer can understand.
The categories you decide on must be comprehensive and mutually-exclusive. You also need a category for data that doesn’t give any relevant information at all; a non-instance category.
When you’ve acquired your sample, you’ll need to go through and ‘code’ it all by mapping the information into categories.
Because content analysis typically works with huge amounts of data, a researcher will almost always rely on software programs to do much of the repetitive data crunching. Some popular programs for this are MaxQDa, NVivo and Atlas.ti.
But even when you’re done, you’re not done. The last step is always to test reliability. The best way to do that may be to ask someone to categorize the data, independently, go through the data crunching, and see if they come up with the same conclusion as you just did.
Kassarjian, H. (1977). Content Analysis in Consumer Research, Journal of Consumer ResearchVol. 4, No. 1 (Jun), pp. 8-18. Oxford University Press
Retrieved December 27, 2017 from: http://www.jstor.org/stable/2488631 on December 20, 2017
Rourke, L. & Anderson, T. (2004). Validity in Quantitative Content Analysis, Educational Technology Research and Development, Vol. 52, No. 1, pp. 5-18. Retrieved December 27, 2017 from: http://www.jstor.org/stable/30220371 December 22, 2017
Sommer, B. (n.d.). Doing a Content Analysis. Retrieved December 28, 2017 from http://psc.dss.ucdavis.edu/sommerb/sommerdemo/content/doing.htm on December 28, 2017.