As discussed in chapter 2 and chapter 3, the paradigm under which a researcher operates informs the approach they take when conducting research. Generally speaking, the paradigms of social constructionism, critical inquiry, post-modernism as well as the Indigenous paradigm all typically produce theorists who approach their research inductively, employing qualitative methods to find answers to their research questions (Hesse-Biber, 2017). They are more concerned with better understanding a particular process or the meaning their research participants attribute to a certain social situation or personal lived experience (Hesse-Biber, 2017). As such, they are more concerned with strategically selecting a smaller sample than they are with the generalizability and representativeness a very large sample might provide. It logically follows, then, that these qualitative, inductive researchers typically employ non-probability sampling strategies.

On the other hand, positivism produces theorists who approach their research deductively, employing more quantitative methods. As discussed in chapter 2, positivists believe that truths do exist, and we just need to use systematic, scientific strategies to uncover them. They are more concerned with determining causes and explaining phenomena. It logically follows then that these quantitative, deductive researchers typically employ probability sampling strategies. 

As we’ll discuss in this chapter, probability and non-probability sampling techniques are quite different and serve different purposes. Keep in mind as you review these different sampling strategies that they are not mutually exclusive. Many research studies use more than one sampling strategy, depending on what the research question is, what the population of interest is, and what data are available to answer that question. Sometimes when working within Indigenous or other communities, the sampling technique may need to be modified based on input from the community and/or problems with the practicality of implementing an ideal strategy within a typically small, closed population. When working with Indigenous communities, they may have suggestions that pertain to sampling such as which elders you need to speak to before getting started. Those elders may then have suggestions not only about the project design, but also about how to select your sample and the best wording to use for the recruitment of participants. For instance, if you want to discover if traditional cultural involvement helps youth stay away from deviant activities, they may suggest you sample the youth participating in the upcoming powwow rather than youth in school. Often, the support of key elders for your project will convince Indigenous community members to voluntarily participate. Further, a mixed-method, and therefore a mixed-sampling strategy, is often not only necessary but advantageous to provide the most complete and holistic answers to the complex research questions we ask as criminologists.

Non-Probability Sampling Strategies

Non-probability sampling refers to sampling techniques for which the chance of any person or entity being included in the sample is unknown (Maxfield & Babbie, 2018; Ritchie & Lewis, 2003; Palys & Atchison, 2014). The reason it is unknown is because we don’t have a sampling frame. Let’s say, for example, you want to interview Indigenous residential school survivors. The records of who attended these schools are not always publicly available and in many cases have been destroyed; therefore, you likely cannot obtain an exhaustive list to select a sample from, and, consequently, we cannot statistically calculate the likelihood of selection or know whether a sample represents a larger population. This might sound like a problem, but representing the population is not the goal with non-probability samples. The goal, rather, is to include in our sample people or things that possess the detailed and rich information we are looking for (Ritchie & Lewis, 2003), even if that means only interviewing 25 residential school survivors. Researchers still use sound scientific processes to select their samples. The next sections explain some of these non-probability sampling strategies, but let’s first consider why a researcher might decide to use a non-probability sample.

A researcher might choose a non-probability sampling method when designing a research project. For example, before conducting survey research, a researcher might administer the survey to a few people who seem to resemble the people they’re interested in studying to help work out any issues with the survey such as unclear question wording, a missing response option, or confusing ordering of questions. Researchers might also use a non-probability sample to conduct a pilot study or exploratory research before designing a more comprehensive study. This can be a quick way to gather some initial data and help get some idea of the lay of the land before conducting a more extensive study. These examples show how non-probability samples can be useful when setting up, framing, or beginning a research project.

Researchers also use non-probability samples in larger-scale research projects. These projects are usually qualitative in nature, where the researcher’s goal is gaining an in-depth understanding of a topic or issue. For example, evaluation researchers who aim to describe some very specific small group might use non-probability sampling techniques. Researchers conducting inductive research in which the goal is to contribute to the theoretical understanding of some phenomenon might also collect data from non-probability samples. These researchers may seek out extreme or anomalous cases to help improve existing social theories by expanding, modifying, or poking holes in those theories.

In short, non-probability samples can serve important purposes in social science research. They are particularly useful for developing strong research projects and for improving theories through the use of extreme, anomalous, or other purposefully selected cases.

Types of Non-Probability Samples

Researchers use several types of non-probability sampling techniques, including convenience sampling, purposive sampling, quota sampling and snowball sampling. While convenience and quota sampling may be used by quantitative researchers from time to time, they are more typically employed in qualitative research, and they are both non-probability sampling techniques.

Convenience Sampling

Convenience sampling (also sometimes called opportunity, accidental or haphazard sampling) is perhaps the most basic and simple form of sampling (Ritchie & Lewis, 2003). It involves drawing a sample from the part of the population that is close at hand, readily available, and/or otherwise convenient to access (Maxfield & Babbie, 2018). This method is most useful in exploratory research. Journalists also use this technique when they need quick and easy access to people from their population of interest. If you’ve ever seen brief interviews of people on the street, you’ve probably seen a convenience sample in action. It is very much a first-come, first-serve sampling strategy. This is not to say that the samples are selected “randomly,” however. The term “random selection” has a very specific meaning in social science research, and it is not synonymous with convenience sampling, despite the tendency of social science students to refer to it as such.

Let us look at a more concrete example to better illustrate convenience sampling. Suppose a local news crew was on your campus today and stopped 100 students between classes to ask their opinions on installing metal detectors on campus, and the results of the survey reveal that 90 of them are in favour of this crime prevention strategy. Later, you find out that these 90 students who happen to have just exited a class right where the news crew was set up had just watched a very graphic film on school shootings. The result from this survey would surely represent the feelings of the (rather emotional) convenience sample, but the results might be completely different from students on the campus as a whole if they were instead selected using a strategy that yielded a representative sample of all students.

While convenience samples offer one major benefit—convenience—as this example demonstrates, we should be cautious about generalizing from research that relies on convenience samples. These types of samples exclude a large portion of the population, such as the people who didn’t happen to walk down the exact street on which the researcher was looking for participants or the students who were not present when the local news crew attended the campus. As such, data collected from the sample may reflect the unique characteristics of the area or group in which you’ve chosen to recruit participants rather than representing the larger, more diverse set of people or other entities you are trying to study.

Quota Sampling

Quota sampling involves the researcher segmenting the population of interest into mutually exclusive groups and then choosing a non-random set of observations from each group to meet a predefined quota. In this type of sampling, a researcher finds potential participants by:

1. identifying categories that are important to the study and for which there is likely to be some variation,
2. creating subgroups based on each category,
3. deciding how many people (or documents or whatever element happens to be the focus of the research) to include from each subgroup, and
4. collecting data from that number of entities for each subgroup.

The number of entities to include in each group can be determined in a few different ways. In proportional quota sampling, the researcher tries to match the proportion of respondents in each subgroup to the proportion of that group in the population. For instance, imagine you wanted to use a sample of 100 students at your university about their career goals and to compare these goals by gender. First, you would need to know how many students identified as male and how many identified as female; you could get that information from the registrar’s office. Let’s say 63 percent of students identified as female, 32 percent identified as male, and 5 percent identified as non-binary or transgender. You would then want to ensure that 63 of the students in your sample identified as female, 32 as male, and 5 as non-binary or transgender. Quota sampling resembles convenience sampling in a way because the first 63 people who walked by you who identified as female would suffice. Once you have reached the number 63, you would no longer survey female students.

Non-proportional quota sampling is less restrictive because a researcher tries to meet a minimum number of people in each subgroup rather than meeting a proportional representation of the population. In this case, a researcher may decide to have 25 respondents from each gender rather than a number that matches the proportion of this gender in the population and then stop when they reach the quota for each subgroup. A non-proportional technique can be useful in research with small and/or marginalized groups because it can oversample these groups to provide more data on people whose voices may otherwise be silenced by the voices of people in proportionately larger groups. In our example of career goals by gender, those voices that would be silenced if using proportional quota sampling would be the non-binary or transgender students. We would learn more from them if we included 25 instead of 5 of them in our sample.

In sum, quota sampling techniques offer the strength of helping researchers account for potentially relevant variation across study elements, but they are not representative and do not guarantee to yield findings that can be generalized to an entire population. This strategy closely resembles convenience sampling as it’s the first people or unit of observation we have access to who are included in the sample, with the only added element of a predetermined quota or cap for each subgroup of interest.

Purposive Sampling

Purposive sampling (also called judgmental sampling) is aptly named because the researcher is specifically interested in the attributes of the particular sample that was purposely chosen for its characteristics (Maxfield & Babbie, 2018; Hesse-Biber, 2017). It is also called judgmental sampling because the researcher is using his or her judgment in selecting a sample that is specific to the goal of the research.

Let’s refer to the long-term offender study again, as the qualitative interview portion of the study utilized this sampling strategy. In addition to reviewing the case files and psychological assessments of these dangerous sexual offenders in the quantitative portion of the study, I also wanted to interview parole officers who were specifically tasked with the responsibility of supervising these offenders in the community for a period of up to 10 years once they were released from prison. Since the designation was less than 10 years old at the time of writing my dissertation and many of these offenders were just coming up to the end of their community supervision periods, it was very timely to examine how well that supervision was actually going and whether those in charge of their supervision felt it was safe to release these offenders back into the community with zero correctional oversight. The only people who could realistically answer this question with any degree of accuracy were not the policy makers who came up with this rule but rather the frontline parole officers themselves. And not any parole officer would suffice; I had to specifically talk to those senior parole officers in B.C. who actually had long-term offenders in their caseloads. They are the only ones who possessed unique knowledge and expertise on this topic. As such, purposive sampling was the primary strategy used to select these interviewees.

As this example illustrates, no matter how purposive sampling is utilized, the goal of the individual selecting the sample is to be very strategic in hand-selecting the particular sample needed. There is little interest in selecting a representative sample from a larger population; rather, the interest lies in selecting a specific sample that fulfills the goal of the research.

Snowball Sampling

The last non-probability sampling strategy is snowball sampling. With this technique, a researcher starts by identifying a few respondents that match the criteria for inclusion in the study and then asks them to recommend others they know who also meet the selection criteria (Maxfield & Babbie, 2018; Hesse-Biber, 2017). In this case, a researcher might know of one or two people they’d like to include in their study, and they rely on those initial participants to help identify additional study participants.

Let’s look again at the long-term offender study. I had interviewed a handful of parole officers who were responsible for specifically supervising these offenders in the community. I had a hard time getting more names of parole officers from their supervisors, however. So, I asked the parole officers themselves, at the end of each interview, if they knew others who may have been reluctant to tell their supervisor they wanted to participate but in fact wanted my contact information. Thankfully, my sample quadrupled after asking this simple question at the end of my interviews. In fact, I was also referred by one parole officer to a long-term offender who also wanted to be interviewed! By the end, I had numerous interviews, and the sample grew with each subsequent interview, much like how a snowball builds and becomes larger as it rolls through the snow. This study shows how more than one sampling strategy may not only be necessary but very beneficial.

Snowball sampling is an especially useful strategy when a researcher wishes to study a stigmatized group or behaviour. For example, a researcher who wanted to study how transgender police officers cope with police culture would be unlikely to find many participants by posting a call for interviewees in the police station or announcing the study during a departmental briefing. Instead, the researcher might know of a transgender police officer, interview that person, and then be referred by the first interviewee to another potential participant. Having previous participants vouch for the trustworthiness of the researcher may help new potential participants feel more comfortable about being included in the study. For the same reason, researchers may also use snowball samples when they’re interested in studying hard-to-reach populations such as people who share an unpopular opinion on an issue or people who belong to a group with very few members.

The main challenge or limitation with snowball samples is the fact that the initial interviewee shapes the rest of the sample; they are likely to refer the researcher to other potential participants who are very similar to themselves, thus limiting the diversity of your sample (Maxfield & Babbie, 2018; Hesse-Biber, 2017). For example, let’s say you are interested in learning more about sex workers in the Downtown Eastside of Vancouver and you are introduced to one sex worker who identifies as female, white and is a youth. If you ask her to connect you with other sex workers, then they are likely to refer you to others who share the same characteristics and who probably work in the same part of the neighbourhood and not sex workers who have significantly different characteristics or experiences (e.g., homosexual male sex workers). Whether or not this poses a problem really depends on your research question and goals and ultimately the availability of research participants.

Table 7b.1 provides a summary of the types of non-probability samples. As explained earlier, rather than trying to represent a larger population, the overall goal of these samples is to provide a deeper understanding of the issues, gain insights for designing and conducting larger research projects, or build or improve theories about social phenomena.

Probability Sampling Strategies

A core principle of probability sampling is that all elements in the researcher’s target population have an equal chance of being selected for inclusion in the study. This implies that the researchers must first have an exhaustive list – or sampling frame – of all these elements. In research, this is the principle of random selection (Maxfield & Babbie, 2018; Palys & Atchison, 2014; Rennison & Hart, 2019). Random selection is a mathematical process that must meet two criteria. The first criterion is that chance governs the selection process. The second is that every sampling element has an equal probability of being selected (Palys & Atchison, 2014). We won’t go in-depth into the mathematical process of random selection except to say that researchers who use random selection techniques to draw their samples will be able to use statistical techniques to estimate how closely the sample represents the larger population from which it was drawn and consequently, how generalizable the findings are.

Types of Probability Samples

As mentioned, researchers conducting deductive, quantitative studies are the most likely to use probability sampling techniques. This category of sampling techniques includes simple random sampling, systematic random sampling, stratified random sampling, and multi-stage cluster sampling (Maxfield & Babbie, 2018; Palys & Atchison, 2014). The first criterion that needs to be satisfied to be able to use any of these techniques is the existence of a sampling frame. Without this exhaustive list of all possible units in the population that could be selected, probability sampling methods are not an option.

Simple Random Sampling

Simple random sampling is the most basic type of probability sampling. In this technique, all possible units of the population of interest have an equal and known probability of being selected (Maxfield & Babbie, 2018; Rennison & Hart, 2019; Palys & Atchison, 2014). To draw a simple random sample, a researcher starts with the sampling frame. For instance, if you wanted to survey 25 police departments in your state, you’d first develop a list of every police department in your state.

Once a researcher has created their list, they then number each element sequentially and use a random number table (or a set of randomly assigned numbers) to select the elements from which to collect data. One way to do this would be to enter each element into a spreadsheet and then use a random number function within the spreadsheet program to generate random numbers for each element on the list. In the example of a survey of police departments, you could list each department as a separate row in a spreadsheet and then generate a random number to be associated with each row. Then, you would sort the list based on the assigned random number and choose the first 25 departments to survey. It is very much like drawing numbers from a hat: as long as you know how many numbers are in the hat, you can calculate the likelihood of each number being selected and you can start picking numbers until you reach the 25th unit.

Instead of random number functions within spreadsheet programs, researchers could also use a random number table from a variety of other sources such as textbooks or free online random number generators. For example, the website Stat Trek contains a random number generator that you can use to create a random number table of whatever size you might need. Randomizer also offers a useful random number generator.

Systematic Random Sampling

As you might have guessed, drawing a simple random sample can be quite tedious. Systematic random sampling is less tedious but offers the benefits of a random sample. As with simple random samples, you must be able to produce a list of every one of your population elements. Once you have done that, to draw a systematic sample, you would simply select every kth element on your list (Maxfield & Babbie, 2018; Rennison & Hart, 2019; Palys & Atchison, 2014). But what is “k”, and where on the list of population elements does one begin the selection process? The symbol “k” is your selection interval or the distance between the elements you select for inclusion in your study. To begin the selection process, you would need to first figure out how many elements you wish to include in your sample.

Let us say you want to survey 100 students from the criminology program at your university. You gather data from the registrar’s office and discover there are 300 students currently registered in the program. In this case, your selection interval, or k, is 3. To arrive at 3, simply divide the total number of population elements by your desired sample size. To determine where on your list of population elements to begin selecting the names of the 100 students you will survey, select a random number between 1 and k and begin there. If we randomly select 1 as our starting point, we would begin by selecting the first student on the list and then select every 3rd student from there.

Population (N) = all students registered in criminology (300)

Sample (n) = student respondents (100)

k = N/n = 300/100

k = 3

There is one clear instance in which systematic sampling would not be appropriate. If your sampling frame has any pattern to it, you could inadvertently introduce bias into your sample by using a systematic sampling strategy. This is sometimes referred to as the problem of periodicity. Periodicity refers to the tendency for a pattern to occur at regular intervals (Maxfield & Babbie, 2018; Rennison & Hart, 2019; Palys & Atchison, 2014). For example, suppose you want to observe how students at your university campus use the outdoor courtyard space and you need to complete your observations within 28 days. During this time, you wish to conduct four observations on randomly chosen days. To determine which days you will conduct your observations, you will need to determine a selection interval. As you will recall from the preceding paragraphs, to do so you must divide your population size – in this case, 28 days – by your desired sample size – in this case, 4 days. This formula leads you to a selection interval of 7 (N/n = 28/4 = 7). If you randomly select 2 as your starting point and select every seventh day after that, you will end up with a total of 4 days on which to conduct your observations (days 2, 9, 16, and 23). But what happens is that you are now observing on the second day of the week, Tuesdays. As you have probably figured out, that is not such a good plan if you really wish to understand how these spaces on your campus are used. Weekend use probably significantly differs from weekday use, and that use may even vary during the week. Therefore, if your sampling frame has a pattern to it like the days of a week do, this probability sampling strategy would yield a sample that is not at all representative and therefore not generalizable.

Stratified Random Sampling

In stratified random sampling, a researcher divides the study population into strata of these subgroups (Maxfield & Babbie, 2018; Rennison & Hart, 2019; Palys & Atchison, 2014). This technique can be useful when a subgroup of interest makes up a relatively small proportion of the overall sample and the researcher wants to be sure to include representatives from all subgroups, no matter the size of the group. For example, imagine a researcher wants to examine how people with a range of gender identities perceive their interactions with the police. Transgender people make up a smaller percentage of the population than cisgender men and women, so there’s a chance that neither simple random nor systematic sampling techniques would yield any transgender people in the sample. The same logic applies to other non-dominant gender identities such as non-binary, agender, and gender-fluid. Instead, using stratified sampling techniques can help ensure that the sample contains adequate numbers of people in the gender subgroups in the population. This may sound a lot like the non-probability sampling technique discussed earlier called quota sampling, and it is, except for one clear difference: with stratified sampling, there is a sampling frame, and random selection is used to select units from that sampling frame to include in the sample. With quota sampling, since there is no sampling frame, random selection cannot be employed.

In the previous example of selecting 25 police departments from a list of 100 departments, a researcher could start by categorizing the departments based on the population in the area they serve. The categories might include areas with large (more than 50,000 people), medium (between 10,000 and 50,000 people), and small (less than 10,000 people) populations. The researcher would then use simple random sampling to select eight departments from two subgroups and, depending on which group the researcher is most interested in ensuring representation, 9 from the third subgroup to make up the sample of 25 departments. This sampling strategy would ensure that departments serving small, medium, and large populations would be equally represented in the sample even though they’re likely not equal in the larger population.

Cluster Sampling

Each of the probability sampling techniques we’ve discussed so far assumes that researchers can access a list of population elements to create a sampling frame. This is not always the case. Let’s say, for example, that you wish to conduct a study of the experiences that people with different gender identities have had with the police in your state. In the previous sampling techniques, you’d need to create a list of every person in your state along with their gender identities. Even if you could find a way to generate such a list (which is highly unlikely), attempting to do so might not be the most practical use of time or resources. When this is the case, researchers turn to cluster sampling. With cluster sampling, researchers divide the population into clusters (or small groups) for which a list is readily available. Then randomly sample a few clusters, and then randomly select participants within those chosen clusters (Maxfield & Babbie, 2018; Rennison & Hart, 2019; Palys & Atchison, 2014). It should be noted that a sampling frame does still exist as is required of all probability sampling strategies, but in this instance, it is a sampling frame of the clusters and not of the units within those clusters.

Let’s say you are concerned about the financial exploitation of First Nation elderly in Manitoba but there is no list of all affected Indigenous elderly, particularly since it is an underreported type of crime, but also because you would need permissions from every tribe in the province to conduct this research. Instead, you could create a list of all tribal nations in the province, randomly select two of them, and then after you have obtained permissions and support from the two tribes, you include in your final sample the elderly who are randomly selected from within each randomly chosen tribe (cluster).

Table 7b.2 below provides a summary of the types of probability samples. As explained earlier, the overall goal of these samples is to represent a larger population so that research findings can be more generalizable to that population.

Questions to Ask About Samples

When reading the findings of research studies, it’s easy to focus only on findings rather than procedures. But, as the preceding discussions indicate, evaluating how a researcher selects study participants and who they select is very important for understanding research findings. Now that you have some familiarity with the variety of sampling techniques, you are equipped to ask some very important questions about the findings you read and to be a more responsible consumer of research.

Who were sampled, how were they sampled, and for what purpose?

Social science researchers on college campuses have a luxury that other researchers may not: access to a whole bunch of (presumably) willing and able human guinea pigs – students! But that luxury comes at the cost of sample representativeness. One study of top academic journals in psychology found that over two-thirds (68%) of participants in studies published by those journals were based on samples drawn in the United States, and two-thirds of the work derived from US samples published in the Journal of Personality and Social Psychology was based on samples made up entirely of American undergraduates taking psychology courses (Arnett, 2008).

These findings beg the question of what we learn – and whom we learn about – from social scientific studies. Joseph Henrich and colleagues (2010) pointed out that behavioural scientists very commonly make sweeping claims about human nature based on samples drawn only from WEIRD (Western, educated, industrialized, rich, and democratic) societies. Minorities and other marginalized groups in society, including Indigenous peoples, do not have their experiences represented in many of these social science studies; therefore, the theories that develop as a result of the research cannot apply to them.

Risk assessment tests that develop as a result of this research also cannot apply to these minority and marginalized groups. As I demonstrated in my analysis (see Hassan, 2010) of psychiatric assessments administered on Indigenous versus non-Indigenous long-term offenders (LTOs) in British Columbia in the first 10 years of the use of this designation in Canada, twice as many LTOs categorized in the high psychopathic range were Indigenous compared to their non-Indigenous counterparts (46% vs 23%, respectively). Moreover, while the percentage of LTOs categorized in the intermediate category was approximately the same in these two groups (46% of the Indigenous LTOs vs 43% of LTOs not identified as Indigenous), only one (9%) of the Indigenous LTOs assessed using the PCL-R was categorized in the non-psychopathic range, as compared to 10 (33%) of the LTOs not identified as Indigenous. I also found that amongst those LTOs whose files were included in my analysis, a disproportionately high number of Indigenous LTOs were deemed untreatable compared to their non-Indigenous counterparts. The clear overrepresentation of Indigenous LTOs in the high psychopathic range and in the untreatable category calls into question the objectivity and neutrality of the tests used to assess this population. The reduction of human behaviour to a quantifiable score results in an undeniable loss of personal information and blatantly disregards the impacts of colonization and genocide experienced by Indigenous populations across Canada and beyond (see Introduction to Criminology for a more fulsome critical review of criminological theories and their applicability to the Indigenous experience).

It is clear, then, that many robust findings about the nature of human behaviour when it comes to fairness, cooperation, visual perception, trust, and other behaviours are based on studies that excluded participants from outside the United States and sometimes excluded anyone outside the college classroom (Begley, 2010). These points demonstrate that we must pay attention to the population on which studies are based and the claims being made about whom the findings apply to.

A related, but slightly different, potential concern is sampling bias, which occurs when the elements selected for inclusion in a study do not represent the larger population from which they were drawn. For example, a poll conducted online by a newspaper asking for the public’s opinion about some local issue will certainly not represent the public since those without access to computers or the Internet, those who do not read that paper’s website, and those who do not have the time or interest will not participate in the poll. In addition, just because a sample may be representative in all respects that a researcher thinks are relevant, other aspects that didn’t occur to the researcher may also be relevant.

So how do we know when we can count on results that we read from research studies? There aren’t any magic or universally true rules we can apply, but we can keep in mind a couple of guiding points. First, while sampling methods provide guidelines for drawing scientifically valid samples, the quality of a sample should be evaluated only by the sample actually obtained. A researcher may set out to administer a survey to a representative sample by correctly employing a random selection technique, but if only a handful of the people contacted respond to the survey, the researcher will have to be very careful about the claims they make about the survey findings. Second, researchers may be tempted to talk about the implications of their findings as though they apply to some group other than the population actually sampled. This tendency usually doesn’t come from a place of malice, but we must be attentive to how researchers talk about their findings in relation to the population they have sampled.

At their core, questions about sample quality should address who has been sampled, how they were sampled, and for what purpose they were sampled. Being able to answer those questions will help you better understand, and more responsibly read, research results. 

Conclusion

In this chapter, the various non-probabilistic and probabilistic sampling strategies were reviewed.  The key distinction between the two is the availability of a sampling frame, which is a necessary prerequisite for the probabilistic sampling options as it allows for random selection.  While this may make the probabilistic strategies seem superior, a strategy’s effectiveness will be determined by the goals of the research and the nature of the sample.  Representativeness is not always necessary or desired.  As highlighted in the chapter, the combination of different research methods and consequently sampling strategies may be ideal in certain circumstances, again depending on the research question(s) we seek to answer, the data available to answer it, and the unique features of the population we want to know more about.

Now that we have a grasp of the various sampling strategies available to us, the textbook shifts to an examination of the various specific methods we can choose from to answer the types of complex and interesting questions criminologists ask.

References

Arnett, J. J. (2008). The neglected 95%: Why American psychology needs to become less American. American Psychologist, 63(7), 602–614. https://doi.org/10.1037/0003-066X.63.7.602

Begley, S. (2010, July 23). The trouble with using undergrads for research. Newsweek. https://www.newsweek.com/trouble-using-undergrads-research-74633

Hassan , S. (2010). The long-term offender provisions of the Criminal Code: An evaluation [PDF] [Doctoral dissertation, Simon Fraser University]. SFU Summit Research Repository. https://summit.sfu.ca/_flysystem/fedora/sfu_migrate/11562/etd6444_SHassan.pdf

Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences, 33(2-3), 61–135. https://doi.org/10.1017/S0140525X0999152X

Hesse-Biber, S. N. (2017). The practice of qualitative research: Engaging students in the research process (3rd ed.). Sage.

Maxfield, M. G., & Babbie, E. R. (2018).  Research methods for criminal justice and criminology (8th ed.). Cengage Learning.

Palys, T. S., & Atchison, C. (2014). Research decisions: Quantitative, qualitative, and mixed methods approaches (5th ed.). Nelson Education.

Rennison, C. M., & Hart, T. C. (2019). Research methods in criminal justice and criminology. Sage.

Ritchie, J., & Lewis, J. (2003). Qualitative research practice: A guide for social science students and researchers. Sage.

Adaptation Statement

Chapter adapted from

• Applied Research Methods in Criminal Justice and Criminology by Eric J. Fritsch, Chad R. Trulson, and Ashley G. Blackburn, licensed under Creative Commons Attribution-NonCommercial 4.0 International License, except where otherwise noted
• Research Methods for Business and Marketing by George Self, licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License, except where otherwise noted.
• Research Methods for the Social Sciences by Valerie Shephard, licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License, except where otherwise noted.
• Psychological Theories of Crime by Jennifer Mervyn and Stacy Ashton, licensed under a Creative Commons Attribution 4.0 International License, except where otherwise noted.