Research Design Essay ⋆ Political Science Essay Examples ⋆ EssayEmpire

Research design is the process of creating a scientific plan for answering research questions through sampling, measurement, and analysis. It is the formal and creative process of comparing competing theories and making inferences to yield discoveries about the world. The hallmark of social-scientific research design is a rigorous attention to inference, sampling, and measurement.

Political science research aims to make inferences regarding subjects of political interest and does so based on empirical observation. These inferences may be a mixture of the descriptive (for example, what is the rate of voting among all U.S. adults living in poverty?), predictive (for example, will the turnout rate increase if the government institutes an election day holiday?), and explanatory (for example, what are the fundamental determinants of turnout in the United States?). In each case, a correct and effective research design requires careful attention to issues of controls, sampling, and measurement.

In classic experimental research design, a sample is drawn from a population; an attribute is measured for each individual in that sample; a treatment is then applied to a randomly selected subgroup of that sample; and the attribute is remeasured and compared to the attribute level of those not treated (the “control” group). Experimental design provides insurance against numerous threats to inference, such as confounding variables, endogeneity, and self-selection.

Quasi-Experiments And Causal Inference

Political science research designs are rarely true experiments, because political events and political behavior are often impossible to fit into this mold. Although some political science theories can be tested through laboratory or field experiments, most political science research is quasi-experimental or observational. In quasi-experiments (also known as “natural” experiments), the researcher measures the attributes of a sample, some of which have received a “treatment” and some of which have not. The control group is not randomly assigned, and in many cases, there is no pre-measurement of those “treated.” Observational studies examine attributes of a single sample before and after a test without a control group. For example, a natural experiment might analyze the changes in voter turnout in cities after changes in registration requirements have been instituted. Unfortunately, a relationship between treatment and attribute change in such designs is not a strong basis for inferring causality: the relationship could be created by a confounding variable; not included in the model (perhaps a separate attribute, like scandal, determined both turnout and whether a political subdivision adopted the law); by selection biases (those cities that chose to adopt reform could be systematically different with respect to peoples’ voting behavior from those that did not adopt reform); or could even have the opposite direction (perhaps those cities in which voters were already mobilizing and were ready to turn out were also driven to change their registration requirements).

Sampling

Sampling is an important aspect of all research designs. Good research design draws from a sampling frame that is a good match with the population of interest, aims to obtain a sample that is sufficiently variable in the levels of treatments that effects could be detected, and avoids selecting cases based on values of the explanatory variable bias (unless the intent is the tracing or disconfirmation of a covering theory).

In social science, it is rarely possible to directly measure the properties in which there is the most interest. Instead, scholars are forced to measure concepts of interest, such as “poverty,” through such indirect measures as per-capita income. An indirect measure is judged by its reliability and validity. Reliability is stability over repeated measures, and validity is how well the measure equates with the underlying concept.

Choosing the size of the sample is a critical part of the research design, and there is often a trade-off between small and large samples. Small sample designs (sometimes known as “case studies”), in which a single case or a small number of cases is examined in great detail, make it difficult to apply formal statistical inference. It is usually impossible to draw broad inferences from individual cases; however, it may be possible to disconfirm explanatory or predictive theories that were hypothesized to have applied through examination of case history and the tracing of the causal processes that run through a case. The thick descriptions that can emerge from case studies also can be useful in generating other hypotheses. Large samples allow the use of strong methods of statistical inference, but such samples can make it more difficult for researchers to construct valid measurements (since these need to be applied across a wider domain of cases) and to maintain causal homogeneity.

The Question Of “Science” In Political Science

A good research design attempts to answer a research question that is both important in the world and offers some purchase for current social scientific methods and theory to yield insight. These are fundamentally gray areas, and “‘interesting” is value-laden term, especially in the social sciences. An interesting question may shed light on an important policy issue, test a widely contested (or widely accepted) theory, or posit an explanation of an anomaly—something that is not explained by current theory or even seems inexplicable. Tractability is determined by a variety of factors, including the availability of data, available research resources, physical law, and the state of statistical methods. Still, there remains a core of questions and methods generally considered interesting and tractable.

In practice, the process of research design in political science is both iterative and creative, as Henry E. Brady and David Collier (2004) note. It is rare that researchers start with a crisply defined theoretical question, develop measures of the variables of interest from first principles, go into the world and collects data in one fell swoop, and emerge with compelling results. Instead, they may start with a vague question, immerse themselves in the details of the cases relevant to this question, and use these details to generate new theories and new questions. They might start with one set of measures and find, when they attempt to apply these measures to the world, that they are unreliable or flawed. They may attempt one method of sampling but, on examination of the sample, find that it is unbalanced or non-random.

Nevertheless, problem selection, and research design in general, occur within a scientific framework. This framework emphasizes testing competing theories based on evidence gathered from the world, is always open to the possibility that new data will require the updating of theories and conclusion, and insists on transparency on how data is collected and analyzed.

Bibliography:

Altman, Micah. Managing Social Science Research Data. London: Chapman & Hall, 2010.
Brady, Henry E. and David Collier (eds.) Rethinking Social Inquiry, Oxford: Rowman & Littlefield, 2004.
Shively,W. Phillips. The Craft of Political Research, 5th Edition. Upper Saddle River, N.J.: Prentice Hall, 2002.

This example Research Design Essay is published for educational and informational purposes only. If you need a custom essay or research paper on this topic please use our writing services. EssayEmpire.com offers reliable custom essay writing services that can help you to receive high grades and impress your professors with the quality of each essay or research paper you hand in.