Chapter 8 Overview of Data Collection and Sampling Three main ways of collecting data are available for criminal justice researchers. Fundamental to each way of collecting data is sampling. Sampling makes it possible to select a few hundred or thousand people for study and discover things that apply to many more people who are not studied. Main Points Asking questions, making observations, and examining written records are three ways to collect criminal justice data. Each source of data is widely used in criminal justice research; each has its own advantages and disadvantages. Sometimes, researchers use multiple methods of data collection in combination. Regardless of which data collection methods are used, researchers must carefully consider measurement reliability and validity and the extent to which subjects are aware of the data collection process. Criminal justice researchers should routinely be careful and creative in data collection. The logic of probability sampling forms the foundation for representing large populations with small subsets of those populations. The chief criterion of a sample's quality is the degree to which it is representative--the extent to which the characteristics of the sample are the same as those of the population from which it was selected. The most carefully selected sample is almost never a perfect representation of the population from which it was selected. Some degree of sampling error always exists. Probability sampling methods provide one excellent way of selecting samples that will be quite representative. They make it possible to estimate the amount of sampling error that should be expected in a given sample. The chief principle of probability sampling is that every member of the total population must have some known nonzero probability of being selected into the sample. Our ability to estimate population parameters with sample statistics is rooted in the sampling distribution and probability theory. If we draw a large number of samples of a given size, sample statistics will cluster around the true population parameter. As sample size increases, the cluster becomes tighter. A variety of sampling designs can be used and combined to suit different populations and research purposes. Each type of sampling has its advantages and disadvantages. Simple random sampling is logically the most fundamental technique in probability sampling although it is seldom used in practice. Systematic sampling involves using a sampling frame to select units that appear at some specified interval--for example, every 8th, or 15th, or 1023rd unit. This method is functionally equivalent to simple random sampling. Stratification improves the representativeness of a sample by reducing the sampling error. Disproportionate stratified sampling is especially useful when we want to select adequate numbers of certain types of subjects who are relatively rare in the population we are studying. Multistage cluster sampling is frequently used when there is no list of all the members of a population. The NCVS and the BCS are national crime surveys based on multistage cluster samples. Sampling methods for each survey illustrate different approaches to representing relatively rare events. Nonprobability sampling methods are less statistically representative and less reliable than probability sampling methods. However, they are often easier and cheaper to use. Purposive sampling is used when researchers wish to select specific elements of a population. This may be because the elements are believed to be representative, extreme cases or because they represent the range of variation expected in a population. In quota sampling, researchers begin with a detailed description of the characteristics of the total population and then select sample members in a way that includes the different composite profiles that exist in the population. In cases in which it's not possible to draw nonprobability samples through other means, researchers often rely on available subjects. Professors sometimes do this-students in their classes are available subjects. Snowball samples accumulate subjects through chains of referrals and are most commonly used in field research.