10. Content Analysis
Dr. Chantal Faucher and Dr. Rochelle Stevenson
🎯 Learning Objectives
- Define and describe content analysis.
- Describe the ethical considerations that are relevant when conducting content analysis.
- Outline the data sources and research questions that are suitable for content analysis.
- Distinguish between quantitative and qualitative content analysis.
- Discuss the strengths and weaknesses of content analysis.
Content analysis is a research method used to study communications, such as texts or images, and examine their meaning in a particular social context. It gained popularity at the beginning of the 20th century with the mass production of newsprint and was initially a largely quantitative endeavour, then referred to as quantitative newspaper analysis (Krippendorff, 2004). Now, though, content analysis can certainly adopt a more inductive, qualitative approach when the goal is to thematically analyze hidden meanings. It is also possible to use a mixed-method, triangulated approach by combining quantitative and qualitative approaches when analyzing content.
This chapter begins by defining and describing the nature of content analysis and providing examples of studies that have employed this method, both quantitatively and qualitatively. The ethics of content analysis, and issues of bias and representation, are included in this discussion. Next, the chapter reviews the types of data sources and research questions that can be examined with this method. Specific attention is paid to the sampling decisions that are central to the work of content analysts. The chapter then delves into the distinction between quantitative and qualitative content analysis and the coding of both forms of data. The chapter concludes with a discussion of the strengths and weaknesses of this research method.
You may be wondering why this chapter is sandwiched between the qualitative chapters (i.e., interviews/focus groups & field research) and the quantitative chapters (i.e., surveys and experiments). This placement was purposeful; it speaks to the fact that content analysis really can and often should employ both approaches depending, as always, on the research purpose/goal, the research question, and the data available to answer it.
Defining Content Analysis
As mentioned at the outset of this chapter, content analysis involves the study of communications and their meaning. The “texts” that contemporary content analysts investigate include such things as actual written content (e.g., news or magazine articles, legislation, e-mails, letters, blog posts), content we might hear (e.g., speeches, podcasts, lectures), and even visual representations of human communication (e.g., photographs, online videos, television advertisements, movies). Content analysis methods can be applied to almost any form of communication so long as the study of messages is systematic (Maxfield & Babbie, 2011).
Below is a table of the types of communications that could be used for content analysis, mainly focusing on a criminological context. Note that the same source, e.g., online news articles, could have a slightly different focus for the content analysis – on the written text or the images, or both. The same is true for the audio or video sources; the focus could be on what is said, as in the words and narrative, or how it is said, such as energetic and loud or quiet and calm, or both. As you can see, content analysis is very versatile!
| Textual
(written) |
Visual
(images and video; what is presented and/or the accompanying narrative) |
Audio
(what is said and/or how is it said) |
|---|---|---|
|
|
|
Unlike many of the other methods reviewed in this textbook that involve human participants (e.g., focus groups, interviews, field research, surveys and experiments), content analysis is classified as an unobtrusive method (also referred to as a nonreactive method) of research, meaning that the researchers generally do not intrude into the lives of those they are studying in order to collect their data. Researchers focus on data that already exist, that are often publicly available, and that were often not produced for the purpose of being researched. For example, the National Archives in Ottawa contain all the death penalty case files from the Department of Justice from 1867 up until the death penalty was officially abolished in Canada in 1976. These case files include police reports, court transcripts, internal justice officials’ correspondence, media clippings, medical reports, petitions, and letters of appeal to the Minister of Justice and Governor General for a commutation of the death penalty to life imprisonment, among others. None of these documents were produced or intended as part of a research study on the death penalty in Canada; however, they are texts that exist and can be used as data for a content analysis study.
The Ethics of Analyzing Content
As with all research, there are ethical considerations with content analysis, even though it is unobtrusive and there are no human participants. Researchers should consider whether the use of the texts and images they wish to study will breach the privacy of the person who produced the text, whom the text is about, or who is depicted in the image and consider what measures could be employed to protect the privacy of those concerned. Researchers must investigate whether it is possible to gain consent from those whose privacy may be breached directly or indirectly in the course of the research.
Considering issues of representation and bias is also an ethical obligation of content analysts. After all, it is important to remember that texts and images do not exist in a vacuum. Palys and Atchison (2014) and Webb et al. (1981) invite us to think about the processes that lead to certain texts and images being left behind. Are certain groups in society more likely to have their texts and images preserved for posterity? How can we know what is missing from the historical traces that have been left behind? Do the documents that exist provide us with a full, or at least representative, picture of what happened?
A study by Collins (2016) illustrates the importance of considering these issues of representation and bias. In this study, Collins conducted a content analysis of Canadian newspaper articles about female offenders and victims of crime, sampling articles from four major newspapers over a period of 30 years. Collins (2016) employed both a quantitative and qualitative content analysis (to be distinguished and described in more detail later in this chapter), with the analysis showing significant differences in how offenders and victims were portrayed based on gender as well as over time. In the article in which they share their findings, Collins (2016) included a table of descriptive statistics of both offenders and victims, which in itself is illustrative of whose voices and experiences are included, and whose experiences are marginalized. Out of the total sample of 1190 articles, the race of the victim was mentioned in only 338 articles. Digging into this subset of articles revealed that white victims were noted in 209 (62%) articles, while Indigenous victims were identified in 42 (12%) articles. However, according to official crime statistics in Canada, Indigenous peoples are six times more likely to be victims of homicide than other groups (Perreault, 2022), and Indigenous women are more likely to experience physical or sexual assault than non-Indigenous women (Heidinger, 2022). So, despite the difference in prevalence of violence against Indigenous and non-Indigenous women, mainstream news media reports continue to underreport the victimization of Indigenous women, marginalizing their experience. Without critically considering whose experiences are being left out or minimized, the descriptive statistics Collins (2016) presents could be interpreted to suggest that White women are more likely to be victims of crime, thereby overlooking the disproportionate victimization of Indigenous peoples as a whole and Indigenous women in particular. As illustrated here, if we simply take for granted that what is being presented in the text is the whole story (i.e., newspaper article, prisoner record, television commercial), then there is a real risk of reifying existing biases that may be present. Instead, content analysts can and should more deeply investigate the reasons why the discrepancy in reporting exists and question whose voices are included, whose are left out, and why.
Let us turn to another study to illustrate the importance of considering bias and voice. In a content analysis exploring how animal abuse was presented in two mainstream newspapers over the period of two years, Stevenson (2008) identified who was quoted in each of the 48 news articles and who was relied upon for information about the criminal event. Interestingly, Stevenson found regional differences between the papers, with the Toronto Star relying more heavily on police and court sources for details in 65% of articles, while the Vancouver Sun utilized animal welfare groups, namely the BC SPCA, as their main source of information in 69% of articles included in the sample. This finding may have reflected the relatively close relationship between the police and reporters in Toronto and a close relationship between the BC SPCA and reporters in Vancouver. The fact that law enforcement and animal welfare were the most frequent voices means that their perspective was the most commonly shared. Notably missing voices were the animal owners or accused in the cases as well as veterinarians who may have cared for the animals, both of whom could offer a different point of view.
Choosing what Texts to Analyze and What Research Questions to Ask
Now that we have defined what content analysis is and described it as an unobtrusive method that still has ethical implications, let us now turn our attention to the importance of sampling when employing this method and the types of primary and secondary sources and research questions content analysts typically deal with.
As discussed in the sampling chapters in this open education resource (see chapter 7a and chapter 7b), generally, researchers are not able to study all elements of a population (i.e., every news article or podcast cannot be analyzed). This is particularly true in the case of content analysis as the target population is often enormous. For example, in quantitative content analysis, the samples might be composed of hundreds or thousands of items to be coded, while a sample for a qualitative content analysis is often more restricted. Researchers should be deliberate in determining what to include in their sample due to the large amount of work required for each element included in the sample.
Multiple methods can be used for composing a sample for content analysis. For illustrative purposes, let’s return to the study by Collins (2016), wherein they used a date-based strategy to gather a sample for their study. Sampling every Canadian news article about female offenders and victims over a 30-year time span would yield a dataset much too large for a content analysis, so Collins (2016) selected four newspapers from large cities across the country: the Vancouver Sun, the Saskatoon Star Phoenix, the Winnipeg Free Press, and the Toronto Star. Even within these four newspapers, the data from 30 years would be overwhelming. To narrow the sample further, Collins (2016) randomly selected two years from each decade and then randomly selected twenty dates from each year. Each article about crime with a female offender and/or victim from these 120 days (6 years x 20 days) was collected, resulting in a total sample of 1190 articles for analysis.
In another study by Gray et al. (2019), a categorical strategy was used to compose a sample for their exploration of how the connection between intimate partner violence and animal abuse, along with any services for survivors, was communicated to the public via public websites. Rather than taking a smaller sample of the websites for their content analysis, Gray et al. (2019) instead narrowed their sample by including only first-stage emergency domestic violence shelters, which tend to serve survivors in immediate crisis or danger, rather than second-stage shelters, which offer longer-term stays, or domestic violence agencies who did not offer shelter services. They then looked at every website within the category of first-stage shelters, resulting in a sample of 337 websites.
Parameters of the sample are based on the research question. Defining the parameters of a sample may include considering a particular time span, location, and type of source that will be most relevant to the topic to be studied. Collins (2016) wanted to explore change over time, so a time span made sense for their study, whereas Gray et al. (2019) wanted to look at a particular type of source (websites) to address their research questions. Like the examples offered above, defining the parameters of a sample also includes specifying what is and is not part of the sample. For instance, when studying online news, the data available may include text, images, audio, video, and engagement commentary. All of these data may be useful but examining them all can be overwhelming. It is important to set out what will and will not be studied and why.
Sampling decisions may also involve considering if the researcher can reasonably expect to have access to the data source they wish to study. A student in one of the author’s crime and media courses in the Fall semester of 2023 wrote a research proposal at the beginning of the semester about studying video shorts on the war in Ukraine, wanting to examine whether war crimes committed by the Russian army were being portrayed in the same way as the same actions committed by the Ukrainian army. By the time he went to collect his data sample, Hamas had attacked Israel and that war ensued, and the video shorts he had intended to use were no longer featured on the sites he was using. They had been replaced by videos of the Hamas-Israel conflict. With research increasingly being conducted online, researchers must give thought to access in terms of the stability of the data access and how to save data for future analysis.
When researchers examine original sources, their texts represent primary sources. For example, if you wanted to study letters sent between people in prison and their family members and you obtained copies of these kinds of letters, then you are working with primary sources. It is important to note here that even though you have copies of the original sources, they are still considered primary sources because you are analyzing the original documents. In contrast, secondary sources are those that have already been compiled and analyzed by someone else. In your study of letters between inmates and their families, you may not be able to obtain copies of the original letters. However, you might be able to find books or other publications that have summarized and/or analyzed such letters. Other examples of secondary sources include documentaries about people in prison and research reports about communications from people in prison. One way to distinguish between primary and secondary sources is to consider that secondary sources often quote or include information from primary sources.
Content analysis is a versatile method for social research and blends well with a wide range of research questions and methodologies. Say you wanted to investigate how attitudes towards crime change over 10 years. Conducting a longitudinal study with participants via interviews or surveys over 10 years can be challenging and expensive, with the potential for participant attrition and needing to fund the study for a decade or more. However, content analysis offers a way to address the research question without a lengthy process through the use of various texts, such as parliamentary debates about the passage of new crime legislation or online comments about news stories about crime. Alternatively, you could take more of a historical approach, exploring how crime has been portrayed in film over the past century as an indication of popular attitudes.
The table below offers a range of research questions that could be approached with content analysis, along with potential primary and secondary sources. As you can see, there is a wealth of potential texts, and a researcher must make many decisions about the texts they can use!
| Research Question | Potential Primary Sources | Potential Secondary Sources |
|---|---|---|
| How have victim impact statements changed over time? |
|
|
| How does the Canadian criminal court system conceptualize the offence of mischief? |
|
|
| What are the similarities and differences among community-based crime prevention measures? |
|
|
| What is the perception of offenders of probation as a sentence? |
|
|
| How has the use of administrative segregation (solitary confinement) changed over the past 25 years? |
|
|
| How do crime documentaries frame offenders and victims? |
|
|
Qualitative, Quantitative, or Both?
As mentioned earlier in this chapter, content analysis can be qualitative or quantitative, and researchers will often use both strategies to strengthen their research. In qualitative content analysis, the researcher aims to identify themes and examine the underlying meaning of those themes. Quantitative content analysis, on the other hand, involves assigning numerical values to information in the texts under study so they can be analyzed using various statistical procedures. For example, imagine a researcher wanted to conduct a content analysis looking at representations of women police officers in film. A quantitative content analysis might examine how many of all the officers portrayed are women, how many of the main characters are women, how many of the women are ranked officers, etc. The research would code or count the images in the films portraying women officers in order to attempt to establish statistically significant differences between the portrayals of female and male officers based on the manifest content in the films or compared to their representation in policing in real life.
A qualitative content analysis might delve into how those female officers are portrayed, extending beyond their numbers to also include latent content and how it might generate meaning. For example, the researcher could look at stereotypes of women officers in real life (e.g., lacking in physical strength or mental toughness, more focused on their relationships) to see if or how those stereotypes are represented in film. Or, the researcher could investigate issues policewomen experience in real life (e.g., sexual harassment, glass ceiling, tokenism, work-life balance) to see if or how those experiences are portrayed in film. This distinction is explored further below.
Quantitative Content Analysis
As noted above, quantitative content analysis revolves around identifying the manifest content and quantifying data for the purpose of conducting statistical analyses that can range from simple descriptive statistics to more complex inferential statistics (described in more detail in chapter 14). Researchers examine the data for the presence or absence of certain variables they have defined. They may want to know how often particular incidents are reported on in the news or how much time is spent in podcast series discussing specific aspects of a case. They record information that can be used in establishing correlations with those variables, such as the political orientation of the news outlet or the number of people who listen to a particular podcast.
Because of the statistical analyses they will want to perform with their data, researchers taking a quantitative approach to content analysis aim to have a representative sample to ensure their statistics are meaningful. Once the researcher has established the research question and collected a sample, they will want to devise a systematic way to record information about that sample. Researchers will typically establish a set of codes, or rules, that they will use to collect data as they go through each text in the sample (the process of coding will be described in more detail in the analysis chapters of this text: chapter 13 & chapter 14). Codes may be grounded in what are identified in the research literature as significant themes or variables in relation to the research topic. The codes may also be generated from a preliminary qualitative examination of the data, which allows themes to emerge from the data. Researchers will define their codes in a coding manual that clearly lays out what they include and exclude from each code (see an example of a coding manual in Appendix C). They will count every time each code is present in the sample.
Let us return to the example of the representation of women police officers in film to illustrate these concepts. One code might be ‘position of authority.’ The researcher may count the number of times the women officers take the lead in a situation affirming a position of authority (e.g., questioning a suspect or witness) or have their authority undermined in having to correct another character about their position (e.g., the character says something like “it’s actually Detective, not Officer”). In the coding manual, the researcher would clearly state what would be included as part of the theme, or what would be counted. The idea is that another researcher could take that manual, watch the same films, and arrive at the same conclusion as to the presence or absence or extent of the theme of “positions of authority.” Researchers will sometimes use a coding sheet, a sort of checklist to be used while coding to ensure they capture all their codes (see an example of a coding sheet in Appendix D). Both the coding manual and coding sheet will likely be revised as the research project unfolds and more texts are examined. It is a good idea to pilot them with a small sample of texts to see what works well, what does not, and what is not captured. The coding tools can then be revised before being deployed on the larger sample. Establishing a clear coding manual and pilot testing the manual helps to ensure the reliability and validity of the research (recall our discussion of reliability and validity in chapter 6c).
After all the data have been coded, the researcher will then proceed to analyze the information they have collected. In quantitative content analysis, this step will involve calculating statistics to provide a representation of the overall sample as well as relationships between different variables. For example, if a researcher were examining the representation of characters from different ethnic groups in television crime series, their coding might have included items such as whether the characters had leading roles or supporting roles in the series, how many minutes they appeared on the screen, and their roles relative to the police drama (police, lawyer, offender, victim, other). In the analysis, the researcher could calculate how many times or the average number of minutes characters from different ethnic groups appear on screen. They might also use statistical analysis to determine whether there is a statistically significant correlation between the frequency of characters appearing on screen and their ethnicity, their roles, or other relevant variables.
Because quantitative content analysis often involves large data sets, it is not typical for this type of analysis to be done by hand. A number of software packages have been designed for conducting quantitative analysis. Notably, many researchers use SPSS (IBM Statistical Package for Social Sciences) to enter and analyze their data quickly and efficiently (see chapter 14 for more on SPSS). Excel may be sufficient for smaller research projects or those employing more basic statistics.
Let us take a look at a study conducted by Faucher et al. (2015) to illustrate further the features of quantitative content analysis. Their study examined the topic of policies relevant to the handling of cyberbullying at Canadian universities. This study was developed at a time when relatively little attention had been paid to cyberbullying other than what was occurring in the elementary and secondary school (K–12) sectors. Therefore, the researchers were not guided by the existing literature on the specific topic of cyberbullying in universities. They developed a coding manual based on some of their prior research in the K–12 sector as well as their knowledge of the types of university policies that might be relevant to the handling of cyberbullying matters. In all, the study scanned a total of 465 cyberbullying policies at 74 Canadian universities. First, the types of policies were noted (student conduct, respectful workplace, discrimination and harassment, etc.). Then, the policies were coded yes/no for the following: whether the policy actually defined cyberbullying, whether it listed examples of cyberbullying behaviours that would fall under the policy, whether cyberbullying was referenced in relation to hazing, whether cyberbullying was mentioned in relation to discrimination, whether sanctions were mentioned, whether complaint procedures were mentioned, whether the policy addressed prevention, and whether the policy was “cyber”-bullying specific. The descriptive statistics in the analysis paint a picture of the types of policies that existed in Canadian universities at that time to manage potential instances of cyberbullying on campus. It was found that cyberbullying was, for the most part, subsumed within existing general policies on student conduct, electronic communication, or harassment and discrimination. Only about a third of the policies specifically addressed online misconduct, suggesting that the policies were not keeping pace with the evolving technological changes involved in interactions between campus constituents.
Qualitative Content Analysis
As mentioned earlier in this chapter, qualitative content analysis tends to focus on the latent content, looking for the presence of broad themes in the data and ways of discussing or framing a particular issue. A component of this process is identifying themes in a data set, exploring the meaning of those themes, and determining how the information might be interpreted by different groups in society. Researchers might examine comments on social media posts to find the different ways in which people talk about the issue and what stands out to them, but they should also be attuned to what is left out or unsaid. They might delve into television crime shows to examine ways in which particular actions are portrayed. This is a much deeper and more detailed analysis than what is accomplished through quantitative content analysis and is typically based on much smaller samples of data.
As with quantitative content analysis, in qualitative content analysis, researchers will need to establish a method for coding the data. The intent with qualitative content analysis is often to generate new knowledge. Themes found in the research literature may provide initial ideas about codes, but researchers should remain open to seeing new and unanticipated themes in the data. Qualitative coding (discussed in chapter 13) often employs an iterative and inductive process called inductive coding (Strauss, 1987). This means that the researcher will not begin the analytical process with a preset group of themes. They will let the themes emerge from their interactions with the data. They will think about themes as they are collecting the data and keep notes as they go through data collection and the initial reading or examination of the data set. They may start to flesh out a coding scheme or frame as they interact with their data. They may then pilot their preliminary coding scheme and revise it as necessary. A few attempts may be needed to come up with a suitable coding scheme (see a sample coding scheme in Appendix E). In applying their coding scheme, they may begin to see relationships between different themes within their data.
Although qualitative content analysis may have a smaller number of items in the sample, the volume of text may still be large. The analysis of these data is not always done by hand. There are a range of software packages that can assist with qualitative content analysis such as NVivo, HyperRESEARCH, QSR, QDA Miner, Free QDA, ATLAS.ti (see screenshot below), The Ethnograph, dedoose, and many more. Researchers can use these tools to aid in their analysis, as they help keep the data, codes, and themes organized, but the researchers must still establish the parameters of the coding and conduct the sampling, coding, and analysis themselves.

| Quantitative Content Analysis | Qualitative Content Analysis |
|---|---|
|
|
Let us now turn our attention to a few excellent examples of studies employing a method of content analysis – either quantitative, qualitative, or both – to further illustrate the features of this versatile method. We will begin with a study conducted by Goossens et al. (2021), who examined the use of hashtags in Twitter posts regarding a murder case wherein the accused’s defence team argued that the defendant was Not Criminally Responsible on account of Mental Disorder (NCRMD). In this case, de Grood, the accused, murdered five people at a house party in Calgary, Alberta in April of 2014. The high-profile case received a large amount of attention on social media, sparking a few widely used hashtags related to the case. Looking to explore how the general public understood the NCRMD defence in general, and how NCRMD was perceived in the de Grood case specifically, Goossens et al. (2021) collected 4,991 tweets with the three most popular hashtags. After sifting through the tweets to remove duplicate postings, irrelevant content or spam, and journalists or media, they were left with 365 unique tweets for their qualitative content analysis. To start their analysis process and ensure trustworthiness (review chapter 6c for more on this concept of trustworthiness as it relates to qualitative research), two of the researchers independently examined a subset of the tweets to establish preliminary codes, then assessed and reconciled any differences to arrive at a final codebook used to collaboratively and iteratively code the complete sample. Goossens et al. (2016) then identified three main themes: mistrust in the justice system, NCRMD as a loophole to avoid consequences, and a need for change and desire for justice. Overall, Goossens et al.’s (2016) qualitative content analysis revealed a significant lack of public understanding of NCRMD as a criminal defence and the perception that it is a legal loophole able to be exploited to circumvent justice.
There are also noteworthy examples of research conducted by Indigenous scholars employing both a quantitative and qualitative approach. For example, Cripps (2011) conducted a study that includes the qualitative and quantitative analysis of news reports regarding the 2011 murders of two Indigenous women in two different countries: Cindy Gladue in Canada and Lynette Daley in Australia. At the time of the murders, the news media paid little attention, but media attention substantially increased around the trials of the accused several years later. Using 261 news reports about the two cases between 2011 and 2018 and guided by Indigenous research methodologies, Cripps (2021) investigated how Indigenous women were portrayed in the media and the relative newsworthiness of the deaths of Cindy and Lynette.
The quantitative analysis included several metrics. For example, Cripps (2021) observed that only two articles were published about Cindy and only six were published about Lynette in the months following their deaths. These articles were also quite short, with the average word counts in these first articles being 221 and 229, respectively, implying that these murders were not worthy of media attention. In contrast, when the trials for the accused began, there were over 130 articles published about Lynette’s case and over 95 about Cindy’s case. The average word count of these articles was much greater than the few early articles, and female reporters had higher average word counts (640 and 763 for Lynette and Cindy respectively) than male reporters (374 and 592 for Lynette and Cindy respectively). This finding points to a gendered element in the reporting of sexualized violence and also in the representation of the victims, as female reporters tended to describe both Lynette and Cindy in more positive terms.
Findings from Cripps’ (2021) qualitative analysis of the articles revealed a pervasive portrayal of both Cindy and Lynette as being responsible for the violence perpetrated against them, with references to alcohol and drug use, previous sexual histories, their children being in the care of others, and sex work. Cripps (2021) notes that the early articles contained no photographs and were “devoid of information that would invoke sympathy or connection to the victim,” inhibiting the reader’s ability to “make a meaningful connection with her as a person” (p. 307). Combining the qualitative and quantitative analysis, Cripps (2021) clearly illustrates the picture the media paints of Indigenous women who are victims of crime – that they are deserving of the violence perpetrated against them. Cripps (2021) observes that allowing the blame to be shifted from the aggressor onto the victim compromises the process of seeking justice. Further, Cripps (2021) asserts that such media framing of Indigenous victims “perpetuate[s] a colonial discourse that has denied Indigenous women their stories” (p. 316) and calls for a critical and intentional change in how victims of sexualized violence, especially Indigenous women, are portrayed in the media to avoid continued reification of a harmful discourse around gendered violence.
These examples illustrate well the versatility and flexibility of content analysis as a method and the utility of this method to bring to light the messages we consume daily in our lives and to force us to ask ourselves how these messages might impact our perceptions of marginalized groups. See Figure 10.1 below for a visual review of the process of content analysis.

Strengths and Weaknesses of Content Analysis
As with any method for social research, content analysis has advantages and disadvantages, some of which have been mentioned already in this chapter. Among the advantages of content analysis is its versatility, with both the types of research questions that can be investigated and the data sources that can be analyzed. It can be used with any kind of communication, from written to audio to visual data, affording a wide range of choices for data sources, as shown in Table 10.1. It is also unobtrusive, mainly utilizing data that already exist. The use of existing, often freely available texts means that content analysis as a method tends to be inexpensive when compared with other methods. Sometimes, there are online databases that freely and publicly provide access to the types of data criminologists may be interested in studying, such as Canlii, which maintains public court decisions and reasons for judgment in any variety of matters heard by Canadian courts.
An additional benefit of using texts that have already been produced is that there is less reactivity because there is no direct contact between the researcher and the person who produced the text or image being studied. Other types of methodologies raise concerns about the potential for the researcher to influence the research participant (for example, in a survey, the way the questions are phrased might encourage participants to respond in a particular manner). With unobtrusive methods, including content analysis, because the text or image was produced before the researcher began conducting research, there is no way for them to influence the content of the data. However, it is also important to consider the social and political context in which a text or image was produced.
Another advantage of content analysis is the possibility of longitudinal analysis without having to spend years or decades collecting data. The use of existing texts allows the analysis of change over time, a comparison of specific time periods, or a focus on a time in the distant past, again showing the flexibility of content analysis.
Depending on your research question, though, access to data could be a potential disadvantage as some content is not publicly available or free. For example, original court transcripts are often expensive to purchase and are often limited to persons involved in the case. Further, some sources may have been lost over time, are difficult to find, or may not have existed at all. While the sheer diversity of potential texts available for study is an advantage, the quality of the texts must also be considered. Texts must be assessed for credibility, authenticity, and representativeness. Researchers should challenge themselves to identify whose voices are present and absent in any and all content analysis studies.
The development of clear coding schemes and stable and producible data allow for a transparency of process in content analysis that is not found in other methods. Replicability, meaning that other researchers can use the same (or a similar) sample with the coding scheme and get the same or similar results, can lend depth of reliability and validity to content analysis studies. That said, all research requires interpretation, and no research can be completely objective. Content analysis is no exception, and it often necessitates a degree of contextual cultural and social awareness to avoid erroneous inferences or interpretations.
| Strengths | Weaknesses |
|---|---|
|
|
🧠 Stop and Take a Break!
Conclusion
In this chapter, we introduced you to an unobtrusive and versatile method called content analysis. We reviewed the many types of data sources and research questions that can be examined using this method. While historically, content analysis was initially purely quantitative, this chapter provides many examples of studies that employ a qualitative approach or that combine both quantitative and qualitative techniques in the same study.
We also explored the important sampling decisions that are central to the work of content analysts who are uniquely tasked with the responsibility of narrowing down an often enormous and unmanageable number of potential units – sometimes in the thousands – into a feasible sample that can be systematically examined. The ethical responsibilities of content analysts are also discussed, including considering privacy, representation and bias. The strengths and weaknesses of this powerful method were outlined, with particular attention to issues of access.
Ultimately, the beauty of this method is that it provides a way to critically examine the messages we are bombarded with on a daily basis, which may profoundly impact our perceptions about the world around us. In an era where fake news and misinformation are rampant, content analysis serves as a powerful tool in the examination of such messages, the critical consumption of which is perhaps more important now than ever before.
✅ Summary
- Content analysis is an unobtrusive method used to study various forms of communication.
- Ethical considerations relevant for content analysts include issues around privacy, bias and representation, particularly when the content conveys messages about marginalized groups.
- Various research questions can be examined using content analysis.
- There are many decisions that content analysts must make relating to sampling due to the sheer volume of units that can be included in the analysis.
- Data sources included in a content analysis can be primary or secondary.
- Content analysis can involve a quantitative or qualitative approach or a combination of both.
- There are many strengths and weaknesses of content analysis, many of which revolve around access.
🖊️ Key Terms
content analysis: A research method used to study communications such as texts or images and examine their meaning in a particular social context.
descriptive statistics: statistics that focus on our sample data alone, such as frequency distributions, measures of central tendency and measure of dispersion for one or more variables in our study.
inferential statistics: statistics where we use our sample data to make inferences to a larger population from which our sample is drawn. This is based in probability theory.
latent content: The deeper, underlying meaning of a text or image that is not immediately apparent to those who examine it.
manifest content: The surface-level reading of a text or image or what is easily observable to those who examine it.
primary sources: Research data in the form of text or images in their original state, without having been modified, compiled, or analyzed by someone else.
qualitative content analysis: A research method that relies on identifying themes and examining the underlying meaning of those themes in textual data so they can be contextualized.
quantitative content analysis: A research method that relies on assigning numerical values to information in the texts under study so they can be analyzed using various statistical procedures.
reactivity: The effect derived from contact/communication between the researcher and research participant. Research that involves close or prolonged interaction between the researcher and participant is expected to increase reactivity, whereas research that does not involve such contact is expected to minimize reactivity.
secondary sources: Research data in the form of text or images that have already been compiled, summarized, or analyzed by someone else.
unobtrusive method: Also referred to as nonreactive method – a research approach characterized by the lack of contact between the researcher and research participant such that the researcher does not intrude into the lives of those they are studying, and the participant does not react in a particular way due to the presence or influence of the researcher.
🧠 Chapter Review
Crossword
Fill in the term in the right-hand column and it will display in the crossword puzzle. Be sure to include spaces where appropriate.
Discussion Questions
- Review a recent news article. Identify the voices who are included and the voices of those who are not included. Discuss why the particular voices may be included/excluded.
- Write a research question that can be answered using the method of content analysis. Explain why this method is the most appropriate for answering this question and what data source(s) you would try to access.
- Can content analysis tell us what the consumers of these messages think about them? Why or why not?
- What is one potential benefit and one limitation of using content analysis to study legal documents, policy statements, or court transcripts related to Indigenous justice issues?
- In what ways can a researcher ensure that content analysis of Indigenous-related materials respects the cultural context and avoids reinforcing stereotypes?
Further Reading
- Administrative burden and the reproduction of settler colonialism: A case study of the Indian Child Welfare Act
- The racialization of Latino immigrants in new destinations: Criminality, ascription, and countermobilization
- Statutory inequality: The logics of monetary sanctions in state law
- A content analysis of persuasion techniques used on white supremacist websites
References
Collins, R. E. (2016). ‘Beauty and bullets’: A content analysis of female offenders and victims in four Canadian newspapers. Journal of Sociology, 52(2), 296-310. https://doi.org/10.1177/1440783314528594
Cripps, K. (2021). Media constructions of Indigenous women in sexual assault cases: Reflections from Australia and Canada. Current Issues in Criminal Justice, 33(3), 300-321. https://doi.org/10.1080/10345329.2020.1867039
Faucher, C., Jackson, M., & Cassidy, W. (2015). When online exchanges byte: An examination of the policy environment governing cyberbullying at the university level. Canadian Journal of Higher Education, 45(1), 102-121. https://doi.org/10.47678/cjhe.v45i1.184215
Goossens, I., Jordan, M., & Nicholls, T. (2021). #AbolishNCR: A qualitative analysis of social media narratives around the insanity defense. Canadian Journal of Criminology and Criminal Justice, 63(2), 46-67. https://doi.org/10.3138/CJCCJ.2020-0019
Gray, A., Barrett, B. J., Fitzgerald, A., & Peirone, A. (2019). Fleeing with Fido: An analysis of what Canadian domestic violence shelters are communicating via their websites about leaving an abusive relationship when pets are involved. Journal of Family Violence, 34, 287-298. https://doi.org/10.1007/s10896-018-0023-z
Heidinger, L. (2022, April 26). Violent victimization and perceptions of safety: Experiences of First Nations, Métis and Inuit women in Canada. (Juristat, Cat . no. 85-002-X). Statistics Canada. https://www150.statcan.gc.ca/n1/pub/85-002-x/2022001/article/00004-eng.htm
Krippendorff, K. (2004). Content analysis: An introduction to its methodology (2nd ed.). Sage.
Maxfield, M. G., & Babbie, E. R. (2011). Research methods for criminal justice and criminology (6th ed.). Wadsworth.
Palys, T. S., & Atchison, C. (2014). Research decisions: Quantitative, qualitative, and mixed methods approaches (5th ed.). Nelson.
Perreault, S. (2022, July 19). Victimization of First Nations people, Métis and Inuit in Canada. (Juristat, Cat. no. 85-002-X). Statistics Canada. https://www150.statcan.gc.ca/n1/pub/85-002-x/2022001/article/00012-eng.htm
Stevenson, R. (2008). Fido on the front page: What does animal cruelty in the news media look like? [Unpublished manuscript]. School of Criminology, Simon Fraser University.
Strauss, A. L. (1987). Qualitative analysis for social scientists. Cambridge University Press. https://doi.org/10.1017/CBO9780511557842
Webb, E. T., Campbell, D. T., Schwartz, R. D., Sechrest, L., & Grove, J. B. (1981). Non-reactive measures in the social sciences (2nd ed.). Houghton Mifflin.
A research method used to study communications such as texts or images and examine their meaning in a particular social context.
Also referred to as nonreactive method – a research approach characterized by the lack of contact between the researcher and research participant such that the researcher does not intrude into the lives of those they are studying, and the participant does not react in a particular way due to the presence or influence of the researcher.
Research data in the form of text or images in their original state, without having been modified, compiled, or analyzed by someone else.
Research data in the form of text or images that have already been compiled, summarized, or analyzed by someone else.
A research method that relies on identifying themes and examining the underlying meaning of those themes in textual data so they can be contextualized.
A research method that relies on assigning numerical values to information in the texts under study so they can be analyzed using various statistical procedures.
The surface-level reading of a text or image or what is easily observable to those who examine it.
The deeper, underlying meaning of a text or image that is not immediately apparent to those who examine it.
Statistics that focus on our sample data alone, such as frequency distributions, measures of central tendency and measures of dispersion for one or more variables in a study.
Statistics where we use our sample data to make inferences to a larger population from which our sample is drawn. This is based in probability theory.
The effect derived from contact/communication between the researcher and research participant. Research that involves close or prolonged interaction between the researcher and participant is expected to increase reactivity, whereas research that does not involve such contact is expected to minimize reactivity.