Learning Objectives
By the end of this section, you will be able to:
- Understand the difference between primary and secondary data sources
- Identify ways in which primary data can be collected
- Differentiate between cross-section and longitudinal data
A critical component of research design is to consider how and when observations will be obtained, or in other words data collection. Researchers must take into consideration the way data will be collected as well as the timing of data collection. Data collection methods can fall under primary sources or secondary sources. Data from secondary sources refers to existing data collected by someone else. Researchers do not need to collect the data again and will instead compile the variables they need for their studies.
For political scientists, a readily available secondary data source is called the American National Election Studies (ANES). The ANES is a collaboration between Stanford University and the University of Michigan. It provides researchers with information about such topics as voting behavior and electoral participation.
Another source of data is the General Social Survey (GSS). The GSS is collected by the National Research Center and the University of Chicago. The data covers topics that might be of concern to social scientists. For instance, psychological well-being and morality are topics the GSS collects data on.
Secondary data sources can be useful and help save researchers time and money; however, the researcher is constrained by the topics collected by the institutions collecting the data. The data available might not necessarily be helpful in answering your research question, so you might have to collect your own data.
Unlike secondary sources, primary sources refer to original data collected by the researchers. Generally, this entails the creation of a data collection instrument. Although obtaining original data may be more time consuming than utilizing secondary resources, one advantage of original data is that it will ensure that the data you get is what you are looking for.
For instance, you might be interested in elections at the local level but the ANES does not ask questions about local elections. You can collect your own data by creating a survey instrument that is specific to elections at the local level. Data can be obtained through multiple approaches. One way to obtain data is to create and administer a survey. Surveys often contain closed-ended questions, limiting the responses that can be provided. An example of a question that might be on a survey is “Are you a registered voter?” or “Did you vote in the last election.” The answer choices to the questions are predetermined. In these two instances, answers that can be provided might be “yes,” “no,” or “not sure.” Interviews are another way to acquire data. In interviews, questions are often open-ended, allowing the cases the opportunity to provide detailed answers which go beyond the limited responses available on a survey. An example of an interview question might be something like, “Why did you register to vote?” or “Why did you choose to vote in the last election?” Questions such as these allow respondents to provide more detailed answers.
Related to what data is collected is when data will be collected. How many observations will you be taking? Will it be just a one-shot survey, or will you be administering the survey over the next few years? A one-shot survey is deemed a cross-sectional study whereas the latter would be considered a longitudinal survey. In a cross-sectional study, observations are taken at a single point in time. A longitudinal study will have multiple observations over a specified length of time with the same individuals. Longitudinal studies can be either panels or cohort studies. A panel study is often a sample of cases that are likely to be representative of the population. Cases in a cohort study are likely to share characteristics or experiences. Multiple observations are collected from these cases over time. A repeated cross-section is a combination of cross-sectional data and multiple observations; however, observations may not be collected from the same cases. This type of research can help provide insight into established patterns.
In this chapter, we provided an overview of research design. You should be able to recognize research design notation and be able to understand the components of the design as well as differentiate experimental designs from nonexperimental designs. In providing you with this overview, we have given you a foundation to begin building designs of your own. Similar to the use of secondary sources to acquire data, pre-existing designs may not fit the needs of your study. When this occurs, you may have to adapt them to what you are trying to accomplish with your study. If making causal inferences is what you are trying to achieve, your foundation should be the design that will allow you to establish causality--the classic experiment. From this initial design, you can then determine whether you can randomly assign individuals to groups or how many times it would be possible to take observations. And from this starting point, you can also determine if you have enough information to implement an experiment. If you do not, then you might reconsider and instead start with an exploratory study that can help you identify possible causes of an outcome.
