Content Analysis
Goal: Describe the content of a message or collection of communicative messages.
"A method of describing communication in a systematic, objective, and quantitative manner."
Content analysis is systematic in the sense that a clear and explicit set of criteria or rules are used for describing the content of a message. These rules or criteria must be applied consistently.
Good for describing different types and amounts of things.
Overall, simply describing the content of a message by placing what you see into different categories.
How does content analysis work?
I Steps Involved
A Researcher decides what messages or artifacts they are going to investigate.
Must decide what you are going to look at?
Are you going to investigate romance novels, books, movies, newspaper articles.
Must simply decide what it is you are going to focus on.
1. Issues of sampling are important.
Sampling issues apply here.
If you select your samples using a representative method (random) sampling techniques then you can generalize your findings back to the entire population.
However, if you use non-representative sampling techniques, then you can only make claims about what you actually observed.
So, for example, if a researcher decides to watch three movies (convenience sample) and talk about how violent they are then, they should only limit their findings to those movies and not all movies in general.
However, if a researcher takes a random sample of movies from a larger population of movies, then the researcher can make claims about even the movies they did not watch.
B Researcher also decides what they are going to look for.
Must not only decide what you are going to look at, but what you are going to look for.
Content of messages or artifacts can be described many different ways. That is, you can cut the world up in many different ways, but you must decide how you are going to do it.
In other words, what variables are you going to use to describe the content of messages?
1 Must chose variables that will be used to describe content
For example, when studying TV shows, you can describe the content of TV shows in a variety of different ways, by focusing on different variables.
In other words, you can describe the content of TV shows in terms of their violence, or their sexually explicit content, or in terms of their family values, etc.
These variables are typically measured using an nominal or ordinal scale.
Good example of content analysis is:
Signorielli, McLeod, and Healy did a content analysis of commercials on MTV
Several variables were used to describe the
content of these MTV commercials:
Sex of person in ad (nominal measure) -
female, male
Body type of person in ad (ordinal
measure) - out of shape, average, very fit
Attractiveness of person in ad (ordinal
measure) - unattractive, neither, attractive, very attractive
Type of Clothing (ordinal measure) -
neutral, somewhat sexy, very sexy
C Must develop a coding system
Must come up with clear up and explicit rules about how the content of the message will be coded.
Not only have to decide what variables to look for, but exactly what counts when describing the content of a message.
In other words, must decide, exactly what counts as very sexy clothing, and what counts as just somewhat sexy clothing.
Involves operationalization.
Two ways of doing this.
1. Manifest coding
Surface level coding of messages.
Coming up with very objective rules or criteria for what counts.
a. compile a list or set of features that describe each attribute
For example, describe all of the features that make clothing sexy.
In this case, you simply come up with a list of what sexy clothing consists of.
b. go through all of the messages or artifacts and look for these features. When features match with description then it is coded as belonging to that category.
c. advantage is that it is very reliable method.
Everyone using this method should get same answers.
d. May not be the most valid method.
2. Latent coding
Involves coding in terms of the underlying meaning, not simply surface features.
More subjective form of coding.
a. Simply have people look at each person and decide if they are wearing sexy clothing or not.
b. Reliability is not as good.
c. Validity might be better.
3. Coding scheme must be exclusive and exhaustive.
Must be able to place everything observed somewhere.
Must be able to place everything observed into only one category.
D Must pre-test the coding scheme.
Go out and test the coding scheme you developed. Have different people evaluate the same messages and see if they get the same answers. If they don't you'll have to change coding scheme.
Must be reliable before it is used.
E Actually go out and code messages obtained
Go through all of the messages obtained and simply code them in terms of the variables that were chosen.
F Describe your findings
Typically trying to determine there is a pattern to the data. Later on we will briefly go over the statistics used to determine if there is a pattern to the data or not.
In general, we use statistics to determine if the pattern is due to chance or is it due to something else.
II Advantages of content analysis
1. Can compare trends over time
2. Artificiality not an issue -- Real world messages typically examined.
3. Reactivity is not an issue -- Messages or artifacts don't change when you observe them.
4. Can be very convenient
5. Can be very economical
III Disadvantages
1. Can't determine causality – Content analysis doesn't tell why pattern exists or what impact it has on people.
2. Limited to materials on hand -- Can't study phenomenon that isn't recorded somehow. Can only study messages that have been recorded.
3. Need proper sampling -- in order to generalize beyond the objects actually studied.
4. Can be very time consuming, and expensive in some circumstances.