Skip to main content icon/video/no-internet
icon backReturn to Entries

Generalization

Generalization refers to the extent to which findings of an empirical investigation hold for a variation of populations and settings. The definition of generalization is closely related to the concept of external validity, which concerns whether findings of one particular study can be applied to unexamined subjects and contexts. Some scholars, such as William Shadish and colleagues, argued that construct validity is another type of generalization, which concerns how well the variables operationalized in a study represent the abstract constructs they are supposed to represent. Overall, it is common for researchers to use the term generalization to refer to external validity in a broad sense. Generalization pertains to various aspects of a research design, including participants, settings, measurements, and experimental treatments. This entry focuses on one of the most important dimensions of generalization: participants or subjects. It is worth noting, however, that generalization of participants is often inseparable from other aspects of generalization, because subjects are inherently situated in a certain place and time.

The target of generalization can be a larger population; for example, a media effect researcher may ask if the emotional reactions to a televised message observed among a college sample can be generalized to the U.S. population. In addition, generalization can also target another sample of similar type, such as generalizing from a sample of students at one college to students at another similar college. In communication research, generalization is an important criterion of evaluating study quality and significance, especially when researchers aim to apply the findings (and implications of findings) to individuals who are not included in their studies. For instance, generalization is crucial when considering whether an antismoking message proved to be effective in a sample of subjects could be effective if used in a nationwide antismoking campaign.

Even though any single study has to be conducted with a particular sample of participants, using appropriate strategies to select participants can help justify generalization. The remainder of this entry provides an overview of the strategies and practices that help researchers facilitate generalization, specifically focusing on sampling, making generalizations without random sampling, and generalization in various research contexts.

Sampling and Generalization

Probability sampling procedures are considered effective to increase generalization of a study. Using a sample of participants who are representative of the population is key for making generalization from sample to population. If a targeted population for generalization can be specified, simple random sampling, in which every individual in the population has an equal chance of being chosen into the sample, can yield a sample that sufficiently represents the population in terms of mean and variance. Simple random sampling is supported by statistical logics that justify generalization. Statistics of a randomly selected sample provide unbiased estimates of the population, and thus, allow researchers to infer properties of a larger population from observations of the sampled participants.

Random selection of participants requires researchers to have a clearly specified population from which a sample can be drawn. However, in some cases, it is difficult to delineate such a population. For instance, suppose a researcher wants to investigate the role of technologies in the maintenance of long-distance dating relationships; it is impossible to enumerate all the individuals who are involved in a long-distance dating relationship. In such cases, researchers may use purposive sampling to facilitate generalization. One strategy of purposive sampling is to deliberately include heterogeneous individuals in the sample. In the example of examining technology use in long-distance dating relationships, researchers who aspire to attain greater generalization may recruit participants of different demographic backgrounds who are in long-distance relationships of various lengths and distances of geographical separation. The other strategy of purposive sampling involves sampling of “typical individuals.” If it is known that people in long-distance dating relationships are typically young adults of 18–25 years of age, researchers may recruit a sample of participants from 18 to 25 years of age, so that results can be generalized to a “typical” population. But purposive sampling of typical cases cannot legitimize generalization to a larger population that may contain atypical cases. In this example, it is highly questionable to generalize findings among participants aged 18–25 to “atypical cases,” such as those who maintain a long-distance dating relationship during midlife or older adulthood.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Read next

More like this

Sage Recommends

We found other relevant content for you on other Sage platforms.

close

Have you created a personal profile? Login or create a profile so that you can save clips, playlists and searches

close
close
close Signed In to profile You are signed in as:
close
Logged in Institution Institution