2. Content Analysis
Multiple Definitions
Background
Applications and Procedure
An Illustration
Consistency within Diversity
Growth
Reliability
Validity
3. General Definitions
“...a research technique that uses a set of procedures
to make valid inferences from text” (Weber, 1990, p. 9)
A summarizing, quantitative analysis of messages relying on
scientific method and not limited to types of variables measured or
context of messages (Neuendorf, 2002)
"Any technique for making inferences by objectively and systematically
identifying specified characteristics of messages” (Holsti, 1969 p. 14)
4. Definitions (continued)
“...technique for objective, systematic, & quantitative description of
manifest content communication” ( Berelson, 1952, p. 18)
“ …systematic, replicable data reduction technique, compressing many
words of text into content categories based on explicit rules of
coding” (Stemler, 2001)
“Close analysis of explicit and implicit messages of a text through
classification and evaluation of key concepts, symbols, and themes to
determine
meaning and explain its effect it on the audience” (Reitz, 2004)
“...systematic reading of texts and symbolic matter not necessarily from an
author’s or user's perspective” (Krippendorf, 2004, p. 3)
6. Goal of Content Analysis
To draw inferences about:
A text
To obtain documentary evidence about:
A sender
A message
A receiver (audience)
9. Analysis of Texts
Researchers have analyzed texts since the start of
writing
Content analysis draws on Aristotelian principles of
textual analysis as described in “Poetics”
Subject of text
Creator
Audience
Purpose
Context
Ethos
Pathos
10. Analysis of Texts
Medieval Scholastics recognized latent meanings of words,
leading to categories
First documented content analysis:
18th century Sweden
Collection of religious hymns blamed for subverting orthodox state
church and inciting dissent
Words and symbols in songs were same as orthodox hymnal, but
occurred in different contexts (Dovring, 1951; Krippendorf, 2009)
11. Analysis of Texts
During early 20th century Weber studied content of newspapers
raising fundamental questions:
How do newspapers affect readers?
How do the contents of newspapers compare quantitatively over a
generation?
What is carried as news
What is carried as editorial
What is omitted
Weber is regarded as a pioneer in the sociology of the press
12. Psychological Analysis
Gordon Allport
Pioneer in personality theory
Analyzed “Letters to Jenny” (Diary)
Aimed to understand her experienced world by identifying
psychological traits as expressed in her letters and diaries
(Krippendorf& Bock, 2009)
14. Content Analysis and Data
The Effect of Semantics on Public Opinion and Propaganda
Program Effectiveness
Readability of Text
Television Violence
Library Transactions
Characteristics of Reference Questions
Personality Traits
Determination of Authorship
15. Questions for Content Analysis
“How are minority ethnic groups treated in short stories in popular
magazines?” (Berelson, 1952)
“How can communications be tested for propaganda elements?”
“What ways do motion pictures reflect popular feelings and desires?”
“What are the dominant images in Shakespeare’s plays?”
“What happens to a good book when it becomes a movie?”
“How do criteria used for program effectiveness reflect program
objectives?”
How do attitudes reflect personality structure?
What domains are currently being studied in library and information
science?
16. A Quantitative Illustration
Research Problem:
What role does popular magazine fiction play in
affecting attitudes toward ethnic minorities?
(Berelson and Salter, 1946)
17. Hypotheses Asserted
Sympathetic characters are from the majority
Unsympathetic characters are from minorities and usually
unsympathetic and secondary
Social interaction between majority and minority groups
seldom occurs.
When interaction occurs, minority group in a subordinate position.
The problems of majority and minority groups differ
The majority has individual (personal) problems
The minority has social and economic problems
The goals of the majority (love, comfort) more approved than
goals of the minority (material gain).
18. Hypothesis and Categories
Hypotheses were translated into categories – a priori assignation
The first hypothesis yielded 3 categories for analysis:
Ethnic identification of character
Role in the story
Level of approval (sympathy for character)
Other hypotheses provide more categories:
Socio-economic status of characters
Kind of interaction
Problems
Goals
After categories for the hypotheses are exhausted, they can be
addressed by potential data
19. Final Step in Analysis
Establish Categories
Determine Indicators (a priori)
First general category was ethnicity
Indicators of ethnicity rarely explicit
(i.e. “He was Italian-American”)
Indirect indicators of ethnicity conceived – indicators involved names (i.e.
Antonio), language, appearance, memberships, etc.
Second category was character’s role in story
Was the role major or minor?
Indicator was the space allotted to specific character
Third category was socio-economic level
Indicator included characteristics as occupation, education, income, possessions,
standard of living, etc.
After examining the stories other categories and indicators were noted. These
emergent categories extended the hypothesis (a posteriori)
21. The Basic Steps
Which data are analyzed? (Unit of Analysis)
How are data defined? (Definition of Terms--these are the variables to
measure. Both internal and external validity come into play here.)
What is the population or unit from which the sampling data are drawn?
(Sampling Frame or Unit of Sampling)
What is the context relative to which the data are analyzed?
What are the boundaries of the analysis?
What is the target of inference?
Code Book - Coding Form
22. Comparison: Berelson’s Study and
Krippendorf’s 6 Questions
Krippendorf Berelson
1. What data are analyzed? 1. Units of analysis are words.
2. How are the data defined? 2. Data defined by categories and
their indicators.
3. What is the population or unit
from which the sampling data are 3. Frame is fictional story in
drawn? popular magazine.
4. What is the context relative to 4. Context involves attitudes
which the data are analyzed? toward ethnicities.
5. What are the boundaries of the 5. Boundary is single story in
analysis? selection of magazines.
6. What is the target of inference? 6. Do indicators support
hypothesis?
24. Content, System, Goals
The objective of content analysis is determining answers
to questions that the text implies – explicitly and implicitly
It is a systematic technique of “reading between the lines”
Can this goal be attained most effectively with quantitative or
qualitative techniques?
Distinguishing feature between 2 approaches is starting
point:
hypothesis or grounded induction
methods of deriving categories
what depth of communication accepted as an indicator
25. Quantitative Procedures
Content analysis starts with the hypothesis
Categories derived from possibilities of hypothesis
Categories established a priori and indicators based on
manifest (surface) meaning
26. Qualitative Procedures
Qualitative approach derives categories after examining
document – emergent coding
Factors in latent meanings of words or messages
Context and proximity suggest the intended
meaning of a message
27. Media: Qualitative Analysis
Narrative Analysis (focus on literary composition and structure)
Rhetorical Analysis examines properties of text; emphasis not on meaning,
but how meaning is conveyed)
Discourse Analysis – examines manifest language in text – language and
word
usage
Semiotic Analysis – focus is on latent meanings and deep structures
Interpretive Analysis – formation of theory from observation
Conversation Analysis – used by social scientists in psychology,
communications and sociology
Critical Analysis – analyzes media messages (how are groups
represented in media of all forms)
28. Consistency within Diversity
The definitions, uses, and techniques vary according to
whether the researcher conducts a quantitative study
or a qualitative study
Content analysis stipulates three core conditions for
qualitative or quantitative character of study
These central conditions include:
Objectivity
System
Generality (Holsti, 1969)
29. Objectivity (Holsti, 1969)
The test of objectivity is whether another researcher,
following identical procedures with the same data, can arrive
at a similar conclusion
Objectivity, in content analysis, means that explicit rules and procedures
guide the research process
For example, what criteria will be applied in formulating a category?
What criteria determines content unit (word, theme, webpage, animation,
imagery, kinds of imagery, narrative, dialogue, space, advertisements,
maps, transactions)?
Objectivity requires researcher explicitly state ground a category is
selected
and what rationale prompts one inference rather than an alternative
30. System
Systematic procedures facilitate objective, unbiased
selection of content and categories (Holsti, 1969)
Assigning firm rules to regulate inclusion & exclusion
of content and categories prevents researcher from
biased, self-serving selection
Rules preclude an analyst from intentional or
unintentional bias in selecting only content or category
supporting hypothesis, proposition, or self-interest.
31. Generality
Select content based on clear definitions for determining
content and categories – fosters objectivity
Establish procedures for inclusion or exclusion of data --
contributes towards consistency and systematic, impartial
selection
Promote principled, legitimate descriptive analysis with
systematic procedures – provides a basis for generalizing
about the data collection
Relate data to an established theory or reasoned proposition
for data to be conclusive or meaningful
33. Arguments Refined
Quantitative and qualitative methods are different approaches to investigating same
problem
Must content analysis be defined by quantitative or qualitative methodologies?
Holsti (1969) holds that other measures besides quantitative frequency counts &
enumeration valid
Alternative method of quantifying data with greater refinement is “contingency analysis”
Contingency analysis codes presence or absence of a trait within document or unit of
sampling. Inferences derived from proximity of two or more units of analysis in sampling unit
The use of quantitative procedures or qualitative is not an either/or issue -- they can
interact
Restriction to quantitative analysis has been criticized on several grounds
Such restriction leads to bias in the problems to consider
It is also stated that qualitative methods can result in more meaningful or descriptive
inferences
34. Manifest and Latent Content
Considerations:
Must content analysis be restricted to manifest content – or
surface meanings; elements “physically present &
countable” (Neuendorf, 2002 p. 23)
Holsti offers compelling example of limits of manifest content.
Limit to manifest content would not yield this notion
Describes patient in mental hospital who states he is in
Switzerland. In this case, the patient’s meaning of Switzerland is
“freedom.” By stating he is in Switzerland, he is stating that he
wants freedom.
Latent content & meaning would provide the context whereby the
receiver of the message would understand the idiosyncratic
meaning.
Latent content need not be dismissed. However, as Holsti
emphasizes, it should be verified by “independent evidence.”
35. A Growing Trend
During early 20th century Max Weber applied techniques of content
analysis to journalism when he introduced the notion of studying
the sociology of the press (Weber, 1924)
Over 24 years -- 1971-1995, Riffe&Freitag (1997) noted six-fold
increase in number of content analyses appearing in journal, J&MC
Quarterly
Software applications and the WWW enable text & data analysis
Neuendorf (2002) reported more than 24 text content analysis
programs are available (THAT WAS 9 YEARS AGO)
Many electronic content analysis applications now available – text,
audio, video
36. Reliability
Reliability measured basically in two ways (Stemler, 2001)
Stability – (intra-rater reliability Does same coder get same results
again and again?
Reproducibility – (inter-rater reliability) Do different coders, rating
the same text, get the same results?
37. Internal Validity
Internal validity is degree a procedure of measurement represents
the defined concept
Does the procedure measure what one intends to measure?
Research question determines what one intends to measure
In previous example, the concern was how individuals in majority
population and minority population were characterized in fictional stories
carried by popular magazines
Study valid if results of inquiry yield data related to this concern
and research question
38. External Validity
External validity or the ability to generalize to the wider
population, depends largely on the soundness of one’s
sampling technique and sample
Need experimental controls or validity becomes serious
concern
Triangulation or independent corroboration is one means of
establishing validity
Another involves conducting same inquiry using another method of
inquiry – formal experimental design employing surveys or
qualitative procedure with focus group/interviews
40. Bibliography
Allport. G. (1965). Letters from Jenny. In K. Krippendorff& M. Bock (Eds.), The content
analysis reader (p. 28-37). London: Sage. Print.
Holsti, O. R. (1969). Content analysis for the social sciences and humanities.
Reading, MA: Addision-Wesley Pub. Co. Retrieved from http://
www.questia.com
Krippendorf, K. (2003). Introduction to Content Analysis. CA: Sage. Print
Krippendorff, K. & Bock, M. A. (2009). The Content Analysis Reader. CA:
Sage. Print.
Neuendorf, K. A. (2002). The content analysis guidebook. London: Sage.
Online resource: http://academic.csuohio.edu/kneuendorf/content
Reitz, J. M. (2004). Dictionary for library and information science. CN: Libraries Unlimited.
(Print).
Riffe, D. &Freitag, A. (1997). A content analysis of content analyses: Twenty-five years of
Journalism Quarterly. J&MC, 74, 4, 873-882.
41. Bibliography (continued)
Stemler, S. (2001). An overview of content analysis. Practical Assessment,
Research and Evaluation, 7 (17). Retrieved from http://PAREonline.net
Weber, M. (1924). Toward a sociology of the press: An early proposal for
content analysis. In K. Krippendorff& M. Bock (Eds.), The content analysis
reader (p. 9-11). London: Sage. Print.
Weber, R. (1990). Basic Content Analysis. London: Sage. Print.
Documentary Research: Refers to research that is based on documentary sources or factual records and reports.
Text, in this sense, refers to all forms of recorded material.As a research technique, it is required that all studies be replicable. In order to meet this requirement, all forms of communication need to be stable and durable. (Holsti, 1969 p. 2). Documents include any text, audio recordings, music, videotapes, drawings, television, or images. In the digital age, content analysis includes the analysis of stable email, posts, blogs, podcasts, webinars, and all forms of digital communication. Lasswell, a communications scholar, used content analysis to study propaganda. .This well-known question combines quantitative procedures or enumeration and qualitative procedures or context and proximity.
Categories were formed that encompassed a concept that many words could convey. The notion of the connotation and ambiguity of words and their meanings were recognized.
The process of determining authorship involves compiling a list of probable authors and comparing their, language, use of language, themes, and all stylistic features that characterize each author. That author whose textual content is most similar in all defined categories to the text in question would be designed the rightful author. One of the most recent, successful examples of this use was in identifying the anonymous author of the 1990’s book,” Primary Colors.” The book dealt with Clinton’s 1992 Presidential campaign, but the authorship had not been disclosed. Using the method of content analysis, the literary techniques of rhetorical analysis and narrative analysis, Donald Foster, a Vassar professor, identified Joe Klein, a Times magazine writer, as the author of the book.
Program effectiveness in education would look at mission statements and program objectives, then evaluate whether the criteria used to measure program effectiveness (measures of achievement—test scores, performance, behavior) reflects the objectives. Or, how do criteria used to assess learning reflect objectives established for a course. Objectives and criteria should be established at the start of a program or course—not during.
Strictly speaking, qualitative studies require a priori categories; qualitative studies are characterized by their emergent and inductive coding of categories.
(Krippendorf, 2003)Stemple (2001) cites three problems associated with compiling documents for content analysis 1. a significant number of documents from the population are missing 2. Records or Documents that do not meet the definition of the document should be discarded and a notes citing the reason kept. 3. Documents meet the definition, but can not be coded because they are missing passages or have ambiguous content.
Paraphrased from Holsti (1969, p. 4).
Objectivity, System, and Generality are not exclusive to content analysis. They are required conditions of all scientific or reasoned inquiry.
(Holsti, 1969 p. 10)By interact, I mean that each methodology can inform the other.
Max Weber formally emphasized the importance of applying quantitative and qualitative approaches in analyzing how the contents of newspapers effect public opinion. Weber referred to this study as the sociology of the press.
This is a seriously rough means of testing reliability. However, it may reveal some of the difficulties involved with determining features and consistently coding them.