Generalization and External Validity
Generalizations
The degree to which connections among calculated variables can be exhibited in a wide mixture of individuals and with a wide assortment of controlled measured variables is known as a generalization. There are two primary types of generalizations. The first is generalization across participants. This type of generalization is efficient because it provides more powerful tests of the research hypothesis and minimizes variability within the conditions of the experiment (Stangor, 2013). Generalization across people, in the true sense, is not possible. This is because it may not be possible to generalize the results of only one group and apply it to another group in a study.
The second type of generalization is generalization across study and it refers to how findings, beyond the specific settings, will be generalized. Most research studies are conducted with a few people, and sometimes other variables such as the experimenter’s sex, age, race, ethnicity, and demeanor may affect the research (Stangor, 2013). Therefore, the uniqueness of any one test makes it conceivable that the discoveries are restricted somehow to the particular setting; and by extension, to the experimenters, measured variables, or manipulations (Stangor, 2013).
Exact Replications
When a researcher repeats an experiment where the conditions are as close to the original conditions as possible, this is known as an exact replication (Stangor, 2013). If a researcher hypothesized that playing a G-flat key for three hours straight decreased depression in college students, a researcher who does an exact replication of that study would have use the exact controls that the previous researched use, but with new participants. Exact replications are rare, but they are typically used to see if the findings can be discovered again.
Conceptual Replications
A researcher can use conceptual replication to use different methods to repeat the test of an experimental result. Therefore, I can test a hypothesis using the same conceptual variable but have different operational definitions of the dependent and/or independent variable (Stangor, 2013). A researcher could show evidence that skateboarding increases concentration in teens from ages 12-17. However, another researcher could come along and use a different method to test it that looks like the same research that a previous researcher did earlier, measuring the same thing.
Constructive Replications
Constructive replication avoids the duplication of procedures (Stangor, 2013). Researchers use it to test the validity of the hypothesis independent of the methods that were used. So a researcher may add new conditions to the same experiment to rule out alternative explanations. A researcher could predict that people made unintelligent last minute decisions when needing to go to the bathroom, but another researcher could test that as well as under the condition of people making unintelligent last minute decisions when not needing to go to the bathroom.
Participant Replications
Researchers may also choose to use different participants than the ones in an original study (Stangor, 2013). For instance, a researcher could have done a study to gauge seasonal affective disorder in people in the northern hemisphere above certain latitudes in the United States. I could also do the same study using people in Russia. I’m simply not just changing the participants, but utilizing a different culture as well.
External Validity & Replications
When it comes to external validity, replication is important. You don’t know if a study is true if it cannot be replicated. If a study has low external validity, then it can’t be generalized (Stangor, 2013). Likewise, external validity can be checked with replication.