Chapter 8 – Sampling (pp. 160-184) Overall teaching objective: To introduce undergraduate criminal justice research methods students to the probability and non-probability sampling procedures. Sampling is a scientific technique that allows a researcher to learn something about a population by studying a few members, or a sample of that population. There are numerous types of sampling. Some sampling techniques allow researchers to predict, with some degree of accuracy, something about a population based on what is learned from a representative sample of that population. Other sampling techniques merely provide researchers insight into a population. Table 8.1 - Summary of Commonly Used Probability and Non-Probability Sampling Techniques (p. 161) Probability samples Simple or Systematic random sampling Cases are randomly selected from a complete list of the entire population. Cluster sampling The sample is collected randomly in a series of stages or from randomly selected natural clusters wherein the cases are more readily accessible. Stratified random sampling Cases are selected from well defined strata within the overall population to further enhance the representativeness of the overall sample. Non-Probability Sampling Convenience sampling The sample consists of individuals that are either convenient or readily available to the researcher. Snowball sampling The sample consists of individuals who are referred to the researcher individually by previous research subjects. Typical or Extreme case sampling Individuals or situations are selected to be studied because in the researcher’s judgment they are either typical or extreme examples. Making Research Real 8.1 – Low Morale at the Jail (p. 160) A county sheriff uses a probability sampling technique (simple random sampling) to determine the cause of low morale among the employees in a jail. Because the jail is a large urban jail it was not possible for the sheriff to interview every employee. Sampling Basics (p. 162) Sampling is based on a concept called the central limit theorem. The central limit theorem gives us confidence that if we collect a large enough sample, the sample will be representative of the larger population. A population is the entire set of individuals or groups that is relevant to a research project. A census collects information from an entire population. A sample collects information from a group within that population. Samples are typically less expensive and time-consuming than censuses. The terms ‘members’, ‘cases’ and ‘elements’ are used interchangeably to describe the individual components of a population or sample. A list of the individual components of a population is referred to as a sampling frame. This list enables researchers to select a subset of the population for their sample. The exact process used to select the sample is called a sampling plan. Sample Bias and Precision (p. 164) Random sampling error is one form of bias. It represents the difference between the results the researcher gets from the sample and what the results might have been had the entire population been polled. Samples from highly diverse populations tend to have more sampling error. Selection bias is another form of bias. It is caused by any process that systematically increases or decreases the chances that a member of a population will be selected into the sample. A sample’s representativeness of a population is referred to as its level of precision. The level of precision is influenced by the size of the population, the amount of variability within the population and the frequency with which relevant social phenomena occurs. Making Research Real 8.2 – How Safe is Your Hamburger?(p. 165) A researcher attempts to develop an appropriate sampling plan for measuring the pathogenic infestation within the nation’s meat supply. This is an illustration (outside ‘traditional’ criminal justice) of a practical application of sampling. Making Research Real 8.3 – It’s Just a Simple Telephone Survey (p. 166) The students in Professor Jackson’s class attempt to collect a sample residents to interview. The professor notices a bias in the sampling plan that causes older residents to be more likely to be included in the sample. Making Research Real 8.4 – Predicting the Outcome of the Public Bond Election (p. 167) Although not criminal justice per se, a practitioner may be interested in using a sample This case illustrates the function of sampling precision Probability Sampling Techniques (p. 168) Probability sampling is a general type of sampling that relies on random selection. Random selection means that each member of a population has an equal and nonzero chance of being selected into the sample. In simple random sampling, a researcher randomly selects cases into a sample directly from a population, similar to drawing names out of a hat. In systematic random sampling, a researcher uses a structured process to randomly select cases into a sample. For example, the researcher might select every tenth case from the population. In cluster sampling, researchers identify natural groupings (i.e., clusters) that exist within the population. Some of these natural groupings are randomly selected in the initial stage of the sampling process. Cases are then randomly selected from the chosen clusters until an appropriate sized sample has been reached. Cluster sampling is a type of multi-stage sampling because it involves multiple stages in the random selection process. Stratified random sampling is a multi-stage probability sampling technique that involves randomly selecting cases from groups created within the population. These groups, called strata, are defined by the researcher. This form of probability sampling helps researchers insure the sample will be representative of the overall population. Making Research Real 8.5 – A Survey of Probationers (p. 170) This story illustrates the use of a cluster sampling technique to overcome the lack of a comprehensive (statewide) list of probationers. The researcher randomly selects individual probation departments and then selects probationers from each of these departments. Making Research Real 8.6 – Measuring Binge Drinking (p. 172) This story illustrates the use of a stratified sampling technique to insure that the sample includes (proportionately) all grade levels (freshmen, sophomores, juniors, seniors and graduate students). The strata are the grade levels and a random sample of cases were selected from each strata. Non-Probability Sampling Techniques (p. 173) Non-probability sampling techniques do not rely on random selection and therefore do not allow a researcher to use the sample to predict what might be happening in the larger population. Even so, non-probability samples can provide in-depth information on a population that might not otherwise be accessible and/or information that can be used to develop theories about various phenomena. Convenience samples, also known as availability samples, are created when a researcher selects a sample from a group of people who are at hand or easily available. Normally, researchers rely upon their own experience and judgment when creating a convenience sample. Snowball sampling is a non-probability sampling technique that relies on the sample members themselves to increase the sample size. Members recruited into the sample identify other members of the population and refer the researcher to these contacts until the sample ‘snowballs’ in size. Typical and extreme case samples consist of a single case study of a population or phenomenon. In a typical case sample, a researcher uses a case study to illustrate a common or typical pattern. In an extreme case sample, a researcher uses a case study to illustrate an uncommon or atypical pattern. Making Research Real 8.7 – What is Life Like for an Undocumented Immigrant? (p. 175) A police chief draws a convenience sample of undocumented immigrants to study how they live and negotiate life without benefit of citizenship. He draws his sample from church membership rolls. Making Research Real 8.8 – The Prostitute Study (p. 176) Because there is no definitive list of prostitutes (i.e. a sampling frame) are researcher chooses to collect a non-probability sample. He is interested in them because of an increase in admissions to hospitals of prostitutes overdosing on a new form of illicit narcotic. He uses a snowball sampling technique to find prostitutes to interview and asks them about a new form of drug use among their peers. Making Research Real 8.9 – What Happened to Sally May? (p. 178) A researcher uses an extreme case study approach to study what happened to a girl who was murdered by her non-custodial parent. This study revealed inconsistencies and lapses in the state’s child welfare system. Getting to the Point (Chapter Summary) (p. 179) o Sampling is a scientific technique that allows a researcher to learn something about a population by studying only a few members of the population. o Sampling is based on a concept called the central limit theorem. The central limit theorem gives us confidence that if we collect a large enough sample, the sample will be representative of the larger population. o A population is the entire set of individuals or groups that is relevant to a research project. A census collects information from an entire population. A sample collects information from a group within that population. Samples are typically less expensive and time-consuming than censuses. o The terms ‘members’, ‘cases’ and ‘elements’ are used interchangeably to describe the individual components of a population or sample. A list of the individual components of a population is referred to as a sampling frame. This list enables researchers to select a subset of the population for their sample. The exact process used to select the sample is called a sampling plan. o Random sampling error is one form of bias. It represents the difference between the results the researcher gets from the sample and what the results might have been had the entire population been polled. Samples from highly diverse populations tend to have more sampling error. o Selection bias is another form of bias. It is caused by any process that systematically increases or decreases the chances that a member of a population will be selected into the sample. o A sample’s representativeness of a population is referred to as its level of precision. The level of precision is influenced by the size of the population, the amount of variability within the population and the frequency with which relevant social phenomena occurs. o Probability sampling is a general type of sampling that relies on random selection. Random selection means that each member of a population has an equal and nonzero chance of being selected into the sample. o In simple random sampling, a researcher randomly selects cases into a sample directly from a population, similar to drawing names out of a hat. In systematic random sampling, a researcher uses a structured process to randomly select cases into a sample. For example, the researcher might select every tenth case from the population. o In cluster sampling, researchers identify natural groupings (i.e., clusters) that exist within the population. Some of these natural groupings are randomly selected in the initial stage of the sampling process. Cases are then randomly selected from the chosen clusters until an appropriate sized sample has been reached. Cluster sampling is a type of multi-stage sampling because it involves multiple stages in the random selection process. Stratified random sampling is a multi-stage probability sampling technique that involves randomly selecting cases from groups created within the population. These groups, called strata, are defined by the researcher. This form of probability sampling helps researchers insure the sample will be representative of the overall population. Non-probability sampling techniques do not rely on random selection and therefore do not allow a researcher to use the sample to predict what might be happening in the larger population. Even so, non-probability samples can provide in-depth information on a population that might not otherwise be accessible and/or information that can be used to develop theories about various phenomena. Convenience samples, also known as availability samples, are created when a researcher selects a sample from a group of people who are at hand or easily available. Normally, researchers rely upon their own experience and judgment when creating a convenience sample. Snowball sampling is a non-probability sampling technique that relies on the sample members themselves to increase the sample size. Members recruited into the sample identify other members of the population and refer the researcher to these contacts until the sample ‘snowballs’ in size. Typical and extreme case samples consist of a single case study of a population or phenomenon. In a typical case sample, a researcher uses a case study to illustrate a common or typical pattern. In an extreme case sample, a researcher uses a case study to illustrate an uncommon or atypical pattern.