A. A quantitative method for analyzing the nature,
functions, & effects of texts.
1. Originated in 18th century Sweden when a
group of scholars & clergy analyzed a collection
of 90 non-orthodox hymns.
2. Counted religious symbols in hymns to see if
they were heretical.
3. Modern biblical exegesis of texts still uses
content analysis as a method (along with other
4. Also used in mass media research.
a. 1st applied to content of leading
b. In 1930s, applied to radio & public
c. In WWII, invaluable in examining
German radio transmissions.
d. More recently, used to examine
B. Sometimes seen as descriptive, critical work
C. Defining content-analysis is confusing & in dispute.
a. Some define broadly as any technique that
objectively analyzes messages in a “systematic,
rule-governed & rigorous way” (Communication
Research, p. 171).
b. This definition permits more qualitative
c. Others, however, would not include purely
qualitative studies as “true” content analysis.
3. Frey et. al. on “textual analysis”: a continuum of
quantitative & qualitative approaches.
a. More quantitative--content analysis, interaction
analysis & some historical analysis.
b. More qualitative--critical analysis (rhetorical or
media) & other historical analysis.
4. Standard definition of content analysis:
Method of studying & analyzing messages in a
systematic, objective, & quantitative manner for
the purpose of measuring variables.
5. “True” content analysis aims for objective rigor like
any social science.
a. Operationalization of concepts.
b. Classification criteria must be both explicit &
c. Coders essential for more objectivity &
d. Study should be able to be replicated by
another researcher with similar results.
e. Requires random sampling
f. Requires equal treatment of all content (through a uniform
6. Main goal--the accurate & precise description of messages.
a. Usually done on many texts, not just one.
b. Quantification aids in making descriptions.
c. Quantification aids in comparisons among categories &
sets of data.
d. Seek generalizable results from sample to population
(called a universe).
D. Functions & uses of content analysis
1. Description of message characteristics--what,
how, & to whom something is stated or occurs in a
a. To compare different data sets.
b. To compare content to the “real world.”
c. To assess images of particular groups in
d. To describe cause & effect.
2. Explanation of message characteristics:
a. Examining a dependent variable.
b. Measuring an independent variable.
c. Testing hypotheses--relate certain characteristics
of the source of a given message to its message
1) Make inferences about causes--why something
is stated or occurs in a text.
2) Make inferences about effects--what happens
when something is stated or occurs in a text.
II. PROCEDURES FOR CONDUCTING A
A. Setting up the study:
1. Identify the research problem.
2. Develop RQ(s)
3. Sometimes develop RH(s)—if testing
IV/DV relationships will have RH(s).
3. Determine a suitable data base.
a. Select textual artifacts directly pertinent to
the research problems.
b. If can’t do a census of all relevant texts
must take a representative sample.
B. Obtaining a representative sample.
1. Operationalize “population of interest” or
universe of texts
2. After defining population boundaries, specify
the smallest unit or elements to be studied.
3. Construct a sampling frame.
4. Take a simple, systematic, or stratified random
sample of texts.
a. Random sampling--key component.
b. Can use randomization techniques instead.
C. Collect data from sample, along with contextual
1. Relevant data about extrinsic matters, such as
source data, historical background, documented
2. Contextual information can emerge from
comprehensive literature review.
3. Can also emerge through audience surveys, or
interviews with the source(s), looking at historical
4. Helps justify inferences made from the intrinsic
D. Develop a measurement system
1. Unitizing--identifying the units of analysis (or
2. Five common units:
a. Physical--actual kinds of texts, speeches,
b. Syntactical--symbols, metaphors, words, etc.
c. Referential --what the text is about.
d. Propositional--specific positions taken in the
e. Thematic--specific topics or themes.
3. Count number of times the relevant units appear in
4. Place units into appropriate categories:
a. Pre-formulated--categories drawn from
b. Emergent--categories emerge from the data
5. Define category boundaries with maximum detail
(helps with reliability).
6. Criteria for categorizing:
a. Mutually exclusive--every unit fits into only
b. Equivalent--all categories are similar in
c. Exhaustive--every unit fits into one of the
d. Any “other” category shouldn’t contain
more than 10% of the units.
7. Categories also must be valid:
a. Face validity
b. Semantic validity
1) Confirmation that words chosen for both units
& categories mean what you say they mean.
2) Panel of informed judges can provide
c. Criterion-related validity
d. Construct validity
E. Coding the data
1. Place units into categories (called coding).
2. Researchers seldom (or rarely) code units
a. Introduce bias or error into the analysis.
b. Independent coders permit reliability to be
3. Use at least two or more independent coders
(also known as raters, or judges).
4. Can also use a computer to do the coding.
5. Coders must be trained, but ideally are “naïve.”
6. Do pilot study to obtain an interrater (intercoder,
interobserver) reliability coefficient.
a. Assess level of agreement among raters.
b. If closer to 1.00, the greater the degree of
intersubjective agreement (hence reliability).
7. Revise coding scheme.
8. Recode the data.
F. Analyzing & interpreting the data.
1. After units numerically coded, apply statistical
procedures to the data.
a. Usually nominal or ordinal level data .
1) Use a Chi-Square to assess differences or
2) Can also determine means, modes,
3) Can also do qualitative analysis.
b. For interval level data, can also use
1) Use a t-test for comparisons.
2) Use a frequency distribution.
2. From analysis, interpret the data.
a. Discuss relationship of the variables.
b. If descriptive, need to justify
III. Assessing Content-Analysis
1. Unobtrusive technique.
2. Accepts unstructured material.
3. Can handle large amounts of data.
4. Studies communication in context.
5. Can study changes in messages over time.
6. Can be used to support theory.
7. Easy to triangulate with both quantitative &
ASSESSING CONTENT ANALYSIS
1. Not relevant to all research (e.g. small numbers
2. Cannot determine the truth of an assertion .
3. Cannot evaluate the aesthetic qualities of a
4. Establishing cause is tricky.
5. Interpretations of meaning can be biased.
6. Requires trained coders, so is time-consuming.
7. If don’t use trained coders, reliability is suspect.