Sampling Issues
What do I analyze? Where do I find it? What sample size do I need? What is my unit of analysis?
What to Analyze and Where to Find It
Deciding what content to analyze will depend almost entirely on theory and your research question. For example, if you are approaching a question using gender theory, and are interested in portrayals of boys in children’s advertising, you would likely narrow your options to “cultural texts” that showed such advertisements frequently, such as a children’s television programming.
Examples of cultural texts commonly used in content analyses, include: newspaper articles; magazine articles and advertisements; television shows and advertisements; movies and movie scripts; books (fiction, non-fiction, guidebooks, advice books, etc.); court rulings; transcripts of Congressional hearings (visit the Government Printing Office online at www.gpo.gov for archives); Internet blogs and websites.
University library archives are excellent sources of some of these materials, especially if you need many years’ worth of content. Contact your library to see if they have the material you’re looking for in their catalog and whether it exists electronically, in print, and/or on microfilm. You may also be able to purchase material off the Internet (e.g., DVDs of television programs that are no longer aired). For electronic archives of print materials, LexisNexis and Factiva are helpful search engines. Government repositories, including the Library of Congress and the National Archives, have extensive collections, many of which are accessible online. It’s also very helpful to consult the Methods sections of scholarly articles and books that are reporting on a similar topic because they will often describe not only what their sample consists of but where and how they found it. See Quinn (2006) on the Recommend Reading page for a good example of this (Quinn citation listed under Methodological Exemplars section).
Sampling
Although most sampling decisions should be driven by theory and your substantive research questions and goals, logistics and accessibility will play a part. Your research interest may be about changes in masculinity discourses through the 20th century; but you only have access to 75 rather than the full 100 years of your cultural text, either because that’s all that exists in your library or that’s how long a key publication has been around. Accessibility issues are unavoidable and a common limitation of this kind of research. Be clear and forthcoming in your write-up about how you chose your sample (a combination of substantive and logistical considerations) and what implications those choices may have for your findings and conclusions.
Alternatively, you may run into the issue of too much data. Logistically, it would be nearly impossible and inadvisable to analyze, for example, every episode of every sitcom containing a father character from the 1950s to the present. That would be an enormous amount of data. Thinking of that as the population, apply a statistical sampling technique (e.g., random sampling, stratified random sampling) to arrive at a more manageable number of episodes to analyze. This is common practice in content analysis, but it is again imperative that once decided on, your sampling rules should be adhered to and reported in the write-up. See Misra et al. (2003) in the Methodological Exemplars tab for a good example of this.
Units of Analysis
This may not be a methodological issue people immediately associate with content analysis, but it is important to establish your unit of analysis at the beginning of the research process. Identifying what it is you are actually going to track, measure, and compare over time or across materials has important implications for your sampling procedure, analysis, and presentation of results. For example, you may be interested in gendered depictions of celebrity couples in entertainment magazines. One of your first methodological decisions should concern your unit of analysis, for example, the units of analysis might be magazine articles or celebrity couples. Are you tracking how articles in key popular magazines portray celebrity couples' relationships, or how celebrity couples are portrayed over the course of their relationships across all magazines that mention them? In the former, your unit of analysis will be articles; in the latter, it is couples. If you look over time, year might be the unit of analysis. In most cases, there won’t be a right or wrong choice of unit of analysis, though there may be more and less effective choices, depending on your project. The key is to identify your unit of analysis clearly and bear it in mind as you create your sample, code your content, and analyze and present your results.
What to Analyze and Where to Find It
Deciding what content to analyze will depend almost entirely on theory and your research question. For example, if you are approaching a question using gender theory, and are interested in portrayals of boys in children’s advertising, you would likely narrow your options to “cultural texts” that showed such advertisements frequently, such as a children’s television programming.
Examples of cultural texts commonly used in content analyses, include: newspaper articles; magazine articles and advertisements; television shows and advertisements; movies and movie scripts; books (fiction, non-fiction, guidebooks, advice books, etc.); court rulings; transcripts of Congressional hearings (visit the Government Printing Office online at www.gpo.gov for archives); Internet blogs and websites.
University library archives are excellent sources of some of these materials, especially if you need many years’ worth of content. Contact your library to see if they have the material you’re looking for in their catalog and whether it exists electronically, in print, and/or on microfilm. You may also be able to purchase material off the Internet (e.g., DVDs of television programs that are no longer aired). For electronic archives of print materials, LexisNexis and Factiva are helpful search engines. Government repositories, including the Library of Congress and the National Archives, have extensive collections, many of which are accessible online. It’s also very helpful to consult the Methods sections of scholarly articles and books that are reporting on a similar topic because they will often describe not only what their sample consists of but where and how they found it. See Quinn (2006) on the Recommend Reading page for a good example of this (Quinn citation listed under Methodological Exemplars section).
Sampling
Although most sampling decisions should be driven by theory and your substantive research questions and goals, logistics and accessibility will play a part. Your research interest may be about changes in masculinity discourses through the 20th century; but you only have access to 75 rather than the full 100 years of your cultural text, either because that’s all that exists in your library or that’s how long a key publication has been around. Accessibility issues are unavoidable and a common limitation of this kind of research. Be clear and forthcoming in your write-up about how you chose your sample (a combination of substantive and logistical considerations) and what implications those choices may have for your findings and conclusions.
Alternatively, you may run into the issue of too much data. Logistically, it would be nearly impossible and inadvisable to analyze, for example, every episode of every sitcom containing a father character from the 1950s to the present. That would be an enormous amount of data. Thinking of that as the population, apply a statistical sampling technique (e.g., random sampling, stratified random sampling) to arrive at a more manageable number of episodes to analyze. This is common practice in content analysis, but it is again imperative that once decided on, your sampling rules should be adhered to and reported in the write-up. See Misra et al. (2003) in the Methodological Exemplars tab for a good example of this.
Units of Analysis
This may not be a methodological issue people immediately associate with content analysis, but it is important to establish your unit of analysis at the beginning of the research process. Identifying what it is you are actually going to track, measure, and compare over time or across materials has important implications for your sampling procedure, analysis, and presentation of results. For example, you may be interested in gendered depictions of celebrity couples in entertainment magazines. One of your first methodological decisions should concern your unit of analysis, for example, the units of analysis might be magazine articles or celebrity couples. Are you tracking how articles in key popular magazines portray celebrity couples' relationships, or how celebrity couples are portrayed over the course of their relationships across all magazines that mention them? In the former, your unit of analysis will be articles; in the latter, it is couples. If you look over time, year might be the unit of analysis. In most cases, there won’t be a right or wrong choice of unit of analysis, though there may be more and less effective choices, depending on your project. The key is to identify your unit of analysis clearly and bear it in mind as you create your sample, code your content, and analyze and present your results.
