Designing Experiments

Designing Experiments

Copyright: © 2015 |Pages: 15
DOI: 10.4018/978-1-4666-8116-3.ch006
OnDemand PDF Download:
List Price: $37.50


In this chapter, students will learn the process of designing experiments. The classic experimental design is presented first. Following this, three distinct quasi-experimental designs are presented. The benefits and burdens of the classic and quasi-experimental designs are discussed in depth. By the end of this chapter, students will understand concepts related to random selection, generalizability, treatment and control groups, pre- and post-test measurement of the dependent variable, and internal validity.
Chapter Preview

The Gold Standard Of Experimental Design

Recall that experimentation involves exposing research subjects to external stimuli. Think about the classic drug experiment. How do researchers at pharmaceutical companies know whether a new drug works? They simply administer the drug to one group of people (i.e., the treatment group) and they administer a placebo to an alternate group of people (i.e., the control group). The two groups are compared after a certain period of time. They compare the groups to see if the treatment group has gotten better compared to the control group. If there are differences among the groups – i.e., the treatment group participants feel better compared to the control group participants – then there is evidence that the drug has a positive impact on health outcomes. In short, the drug may work. See Figure 1.

Figure 1.

Classic drug experiment

In the social sciences, the process is the same. Consider this example – Do students that complete all of the assigned readings understand research methods better than those that do not? Based on this research question, the UoA is students and the IV and DV are the assigned readings and understanding of research methods, respectively. Figure 2 illustrates the components of the classic experimental design using this example.

Figure 2.

Classic experimental design

The classic experimental design or the so-called gold standard of experimental design, has three essential components: 1) There must be at least two groups – one treatment and one control group. You can have additional treatment and control groups should the experiment warrant this. 2) There must be random selection of participants into the treatment and control groups. 3) The primary outcome being measured, which is the dependent variable, must be measured before the experimental process and after the completion of the experimental process (i.e., the process of taking x number of individuals and splitting them up into at least two groups). This is usually referred to as the pre-test and post-test. We will discuss each of these three elements in greater detail.

Two Groups (at least)

Experimentation as an empirical method is only appropriate if you have a research question where there is some type of external stimuli. In the case of our example presented in Figure 2, “course readings” is the external stimuli, as some individuals will exposed to the readings and some will not. Let us assume that we have an eight week research methods boot camp, whereby there will be eight total lectures and assigned readings. The treatment group will be forced to complete the readings, while the control group will not complete the readings. This experimental process of splitting up research methods student into two groups – the treatment and control groups – and having one group complete all of the readings over an eight week period measures the IV “course readings.” The primary reason for having two groups is comparison. Comparing the treatment and control group allows one to more accurately assess if the stimuli had an impact on the primary outcome measure – in other words, did the IV impact the DV. FYI – it is perfectly acceptable to have more than two groups. For instance, you could have two treatment groups and one control group, where treatment group number one would complete all of the readings, treatment group number two would complete half of the readings, and the control group would complete none of the readings.

Complete Chapter List

Search this Book: