The process of content analysis entails searching through one or more communications to answer questions that an investigator brings to the search. Content analyses are not limited to written or printed documents but extend as well to audio recordings, still photographs, motion picture films, video recordings, and the like. Comparisons can involve documents from different times, different places, different authors, and more.

Investigators typically analyze communications in order to answer two levels of questions--the descriptive and the interpretive. Descriptive questions focus on what a communication contains. Interpretative questions focus on what those contents are likely to mean. Our concern in the present chapters is limited to techniques of descriptive analysis.

Three key descriptive-analysis questions concern qualities, quantities, and patterning. The qualitative question is: Does this document contain the characteristic for which I am searching? The quantitative question is: In what amounts does the intended characteristic appear? The patterning question is: Which relationships among the characteristics interest me?

The qualitative

Among the diverse aspects of quality that you can study are field-of-focus, reflected attitude, and discursive style.

The field-of-focus of a communication is the facet of life it discusses-political-party policies, social-class structures, sexual harassment, macroeconomic theories, forms of government, comprehensive schools, mental disorders, welfare systems, business ethics, and thousands more. In a comparative study, the analyst's task can involve inspecting two or more documents to determine how they treat the field of interest. The documents may represent different times, different places, or different authors.

The term reflected attitude refers to a communication's general emotional or judgmental tone. An analyst's purpose is to reveal attitude similarities and differences among the documents being compared. Kinds of attitudes on which the researcher's attention centers can be reflected in such contrasting terms as optimistic/pessimistic, critical/noncritical, antagonistic/supportive, prejudicial/ nonjudgmental, positive/negative, and the like.

The phrase discursive style means the pattern of logic by which an author seeks to convey his or her message. The content analyst's task can involve comparing the communication modes of two or more documents.

We do not believe that quantitative methods are in conflict with qualitative approaches. Instead, the quantitative are merely extensions of the qualitative, representing an effort to determine with some precision (1) the amount or frequency of existing characteristics (incidence) or (2) the degree of relationship among characteristics (correlation).

In way of illustration, the amount of a characteristic could be the aim of a study about the treatment by different news magazines of different religious groups. A researcher wishing to derive a picture of a possible link between news media and religious denominations could do so by analyzing the contents of five weekly news magazines that are thought to represent political preferences. The analysis, conducted over a period of months, would be designed to reveal (a) the number of articles in which a given religion is mentioned, (b) the total amount of space devoted to each religion, and (c) the articles' evaluative remarks or implications about each denomination (neutral, favorable, unfavorable). These quantitative data could then be manipulated statistically to provide a numerical appraisal of the relationship between each magazine's political stance and its treatment of news in which people's religious affiliation is identified.

Patterning

Sometimes investigators are interested in discovering patterns of relationships among characteristics found in a communication. A historian may study the minutes of city council's meetings in order to establish the chronology of events in a city government's treatment of sexual harassment of its employees. A sociologist may peruse students' cumulative folders to learn the extent of correlation between students' academic success and such home background features as parents' education, housing, family size, and available reading matter in the home. A political scientist may compare a ministry of the interior's administrative-organization chart with a collection of memoranda from different offices in the ministry in order to trace the flow of ideas about a zoning plan. A social psychologist may inspect biographies published in Who's Who to find the sorts of postsecondary institutions that had been attended by prominent business executives as compared to ones attended by eminent artists, authors, and scientists.

Conducting Content Analyses

One typical process of content analysis consists of five steps: (1) stating the general question that the research project is intended to answer, (2) decomposing the general question into constituent subquestions, (3) finding communications that will likely answer the subquestions, (4) inspecting communications to locate passages pertinent to the subquestions, and (5) recording and organizing the results of the inspection.

The following examples demonstrate two approaches to the content analysis task. The first example illustrates a qualitative analysis in which the researcher studies documents and takes notes by hand. The second describes a combined qualitative-quantitative analysis employing a personal computer equipped with a scanner and optical-character-recognition program.

A qualitative analysis of a science-versus-religion dispute

Consider the steps to be taken by a political science doctoral candidate who is planning a dissertation titled Darwin in the Classroom Revisited: Political Strategies and Consequences--1925 and 1999. The student's stimulus for choosing such a topic was the Kansas state board of education's decision in 1999 to outlaw from school achievement tests any mention of Charles Darwin's theory of evolution. News of the decision was widely debated in the public press, with the 1999 event reminding readers of the 1925 Tennessee court case in which a substitute teacher, John Scopes, was convicted of teaching Darwinism in a public school. The doctoral candidate plans to compare the two events' political ramifications in terms of (a) the strategies attempted by the advocates on each side of the pro-Darwin/anti-Darwin controversy and (b) resulting political consequences.

Step 1: State the general question to be answered by the study. The general question delineates the principal domain of communications that can profitably be analyzed.In the Scopes trial of 1925 and the Kansas state board of education decision of 1999, (a) what strategies did anti-evolution advocates and pro-evolution advocates attempt in support of their causes and (b) what political consequences resulted from those two events?Hence, the relevant communications will be ones mentioning the two events. The communications bearing on the Scopes trial can extend from 1925 through 1999. Communications relating to the Kansas decision begin during the month the decision was issued, August 1999.Sometimes the initial question is sufficiently specific to serve as the complete guide to the precise information sought during the inspection of communications. Frequently, however, the initial question is too broad to suggest the precise information that should be extracted from documents. In such instances, a second step is recommended.

This list will serve as the guide to which contents of communications will be included in the final research report. However, it is also the case that, in the process of an inspecting documents, the documents' contents might suggest additional subquestions that can help answer the researcher's initial general query. Those new questions will then be added to the list.

Step 3: Identify communications that likely contain answers to the subquestions. Guided by the key words Darwin, Darwinism, Scopes, evolution, and creationism, the investigator searches libraries' listings of books, journal articles, magazine articles, and newspaper reports to locate resources pertinent to the research questions. The bibliographies and references at the end of relevant books also provide titles of additional useful resources. The ultimate collection of communications to be inspected includes speeches, articles, books, and conference deliberations focusing on the two events--1925 and 1999--and their apparent aftermath. Newspaper archives from 1925-1926 and from 1999 are included in the search.Here, then, are examples of relevant books about the Scopes trial and its consequences:

Step 4: Analyze the contents of the chosen communications. Analyzing the collected documents involves keeping a list of the subquestions at hand to guide the process of scanning the pages of a book chapter or an article in order to locate passages bearing on any of the subquestions.

Step 5: Record and organize the findings. The researcher prepares five-byeight-inch cards on which to write notes about relevant passages of each document inspected. To facilitate the task of later organizing the cards in a useful sequence, the investigator assigns a code to each question. For example, Dis refers to the question: What were the points of disagreement between the anti-Darwin and pro-Darwin forces? The codes AntiDar and ProDar refer to the question: What kinds of evidence did each side adopt in support of its position, and what were the sources of such evidence? AntiDarArg and ProDarArgu refer to: What line of argument did the proponents on each side employ to support their position? As these examples suggest, codes are typically easier for the researcher to recall if they are cast as concise mnemonic reflections of the essence of the subquestions to which they refer. To illustrate, the connection between such a code number as 4 or 9 and the intended subquestion is more difficult to remember than is the code word CrtCase. (What court cases after 1925 addressed the same issues as those of the Scopes trial?)

At the top of a note card, the analyst writes the code that identifies the type of material treated on that card's notes. The code letters are followed by the bibliographic source of that card's material. Usually the name of the author, the publication year, and the page numbers of the selected passage will be sufficient, since a separate card will be prepared with the complete bibliographic reference that will appear in the list of references at the end of the final research report. Beneath the identifying information, the researcher either (a) summarizes the essence of the relevant passage in her or his own words and/or (b) directly quotes an entire passage or a segment of it. Here is a sample note card:

The bibliography card relating to this note card reads:

Note cards prepared in such a manner can later be organized in a sequence that facilitates the researcher's writing about each of the subquestions. In effect, all cards bearing the same code can be organized in a chronological sequence to reflect the way the Darwinism-versus-creationism controversy advanced over the period 1925-1999.

There are several variations of the above procedure. For instance, some authors prefer to take notes on a lined tablet rather than on cards and to place the code words and each passage's page numbers in the left margin adjacent to the notes and quotations.

A qualitative/quantitative comparison of textbooks

The Darwinism study was concerned entirely with qualitative matters--the strategies adopted by opponents of Darwinian theory to replace the teaching of evolution in schools with a biblical version of creation. However, some content analyses center attention on both qualitative and quantitative aspects. Such is the case in a study of textbooks' characterizations of combatants in World War II.The five steps in content analysis described for the Darwinism study are the same as those adopted for the textbook investigation. But while none of the steps in the Darwinism example involved the use of computers, the procedures adopted for the textbook project make heavy use of computer technology. As noted at the end of this section, numbers of software programs have been designed specifically for content analysis, each with its own special features. However, the following history textbook example illustrates a computer application that requires no more than an up-to-date word processing program, a scanner, and optical-character-recognition software.

Step 1: State the general question to be answered by the study. Which nations' military forces were mentioned in successive editions of popular secondary-school history textbooks published in Australia, Great Britain, India, and the United States in the 1950s and 1980s, how much attention was accorded each nation's forces, and what was the tenor of the attention?

Step 2: State subquestions that identify components of the general question. In each edition of the selected textbooks:

In comparisons across the four nations' 1950s textbooks ( Australia, Great Britain, India, United States) and across the 1980s textbooks, to what extent did the answers to questions (1) through (4) differ from one nation's texts to another's? If they did differ, then in what manner?

Step 3: Identify communications that likely contain answers to the guide questions. The investigator writes to education officials in the four target countries to (a) learn the titles of popular secondary-school world history texts from the 1950s and from the 1980s and (b) learn the sources from which copies of the texts can be purchased. Representative textbooks from each country are then obtained.

Step 4: Analyze the contents of the chosen communications. To implement the process of analysis, the researcher first enters representative chapters of each book into a personal computer by means of a scanner and optical-characterrecognition (OCR) software. The scanner photographs one page at a time as the OCR translates the page's words into the same form that would result if the words had been typed into the computer from the keyboard. Even though this procedure uses a large quantity of computer memory, as personal computers provide increasingly large amounts of storage (6 to 10 gigabytes on a hard disk), the length of a scanned document becomes less an impediment to recording multiple pages. (If a scanner and OCR program were not available, the researcher could still copy the book's contents into the computer by typing from the keyboard, but that would be a laborious and time-consuming task, susceptible to typing errors.) The textbooks' contents can now be read from the computer screen rather than from the books themselves.

Analyzing a textbook chapter involves keeping a list of the subquestions at hand to guide the process of examining the pages to locate passages bearing on subquestions (1) through (4).

Step 5: Record and organize the findings. There are several ways that the textbook contents, as viewed on the computer screen, can be analyzed and recorded. Which method is best depends on such considerations as the amount of the computer's available random-access memory (RAM), the types of research questions to be answered, and the researcher's preferred way of working. One of these possibilities is presented here. It consists of five phases:

Phase 1: For each nation, subquestions (2), (3), and (4) are assigned code identifications. For instance, the codes for Italy are:

The same three types of codes are assigned for the other nations, so there are US-0, US-neg, US-pos, GB-0, GB-neg, GB-pos, and the like for each group.

Phase 2: The book chapter that has been entered into the computer will be analyzed for each nation in turn. First, following the end of the computer's version of a chapter, the researcher places the three codes for the particular nation that is currently the focus of attention. Then the investigator begins perusing the chapter contents. A quick way to locate each mention of the ethnic group consists of using the "find" function from the word-processing program's "edit" menu. When the "find" rectangle is brought onto the screen, the name of the desired ethnic group is typed in, so that each time the "return" key is pressed, the computer will find the next use of that name. The researcher can then read the passage containing that nation designator in order to discover how much space is dedicated to that country and whether the contents are solely descriptive [subquestion (2)] or negative [subquestion (3)] or positive [subquestion (4)].

Phase 3: After evaluating the nature of a passage [in terms of questions (2), (3), and (4)], the researcher employs the "copy" function from the "edit" menu to copy the words, phrases, or sentences that comprise the reference. Once again the "find" function is brought onto the screen, and the code appropriate for the copied passage is entered. For example, when the Japanese passage is complimentary, the code J-pos is entered, and the computer cursor jumps to the J-pos location at the end of the textbook chapter. At that location, the researcher places the copied passage.

By repeating phases 2 and 3 throughout the chapter, the researcher accumulates under each code all of the passages that allude to the particular nation's forces, with the passages organized according to whether they are descriptive, negative, or positive. To make space in the file containing the chapter for the next ethnic group, all of the material under the codes can be copied and transferred to a separate file designated for the group that has just been analyzed. The coded material following the chapter is then erased, leaving room for new codes that refer to the next nation whose passages will be extracted and placed under that new set of codes. This same process is repeated for all nations categories, so each country ends up with its separate file containing all references to it in the textbook chapter.

Phase 4: When phase 3 has been completed for all nations, the researcher has a series of separate files, each containing all of the chapter material referring to that file's particular country. It is now a simple matter to compute the quantity of chapter space dedicated to each nation. For instance, opening the German file, the investigator selects (highlights or blackens) all of the material under the code G-0 and activates the "word count" function, which yields the total number of words describing Germany's armed forces. The same procedure furnishes an instantaneous word count for any other set of material under a given code. By this means, the researcher promptly and accurately answers subquestions (1) through (4). And when phases 1 through 4 have been completed for each of the textbooks under review, the material is available for answering the questions regarding (a) trends in the treatment of a nation's armed forces over time and (b) comparisons of one textbook series with another.Phase 5: When writing the final interpretation of the study's findings, the author selects illustrative passages from the separate nations' files to demonstrate the qualitative differences in the treatment of military forces that appeared at different time periods ( 1950s, 1980s) in a particular nation's history books.

Specialized Content Analysis Programs

As noted earlier, a variety of computer software programs have been developed to facilitate the process of content analysis ( Weitzman & Miles, 1994). Examples of ones that systematically organize text for search and retrieval are askSam, Folio VIEWS, and Orbis. Such programs facilitate searching for and retrieving various combinations of words, phrases, coded segments, and memos. Other programs not only include code-and-retrieve capabilities, but also permit analysts

to make connections between codes (categories of information); to develop higher-order classifications and categories; to formulate propositions or assertions, implying a conceptual structure that fits the data; and/or to test such propositions to determine whether they apply. They're often organized around a system of rules, or are based on formal logic. Examples are AQUAD, ATLAS/ti, HyperRESEARCH, NUDIST, and QCA. ( Miles & Weitzman, 1994, p. 312)

In the following list of content-analysis computer programs, the title of the software is followed by the name and address of the creator and/or distributor of the program.