Discussions - Generalizability

Sampling

* In quantitative studies, using representative sampling measures (e.g., random sampling) increases the chances of generalizability. The goal of representative sampling is to get a sample that is representative of the wider population being studied. If this is achieved, there is a high chance that the results are generalizable.
* Similarly, the larger the sample size, the more likely it is to accurately represent the target population, which also helps to increase generalizability.

In qualitative studies, non-representative sampling methods are used (e.g., snowball & purposive sampling). These samples are often small in size & the goal of data analysis is to understand the subjective experience of participants in the study (so they often do not even seek to generalize / transfer their findings). They often use an idiographic (specific) approach: claiming all phenomena are time & context specific (so not generalisable to other times & contexts or should at 

Representational Generalization

generalizing / applying findings to __populations__ outside the population / sample of the study (so linked to __population__ validity). If findings from similar research studies (how many in your opinion?) corroborate / back up findings, some degree of generalization may be possible, even if the sample is small.

Inferential Generalization

generalizing outside of the study conditions to other __settings__ outside the study (so linked to __ecological__ validity). The findings can be "transferred" to similar settings (inferential generalization is also called transferability). Thick descriptions help a reader consider the extent to which the conditions of data collection are similar to other settings, allowing for inferences to be made, but remaining as a hypothesis until it is supported or disproved by further evidence / studies.

Theoretical Generalization

Theoretical concepts developed in the study can be used to further develop wider social theory. The research findings can be used and generalised in order to further develop & refine theories and explanations of human behaviour

Procedures

* When discussing generalizability or transferability, it is important to consider the procedures used in the study. This is especially true for experimental research. For example, a study might lack ecological validity (a type of generalizability) if it can be clearly explained why we might not expect the same results in a different environment. Most experiments use opportunity sampling & artificial conditions, so the results cannot be generalized outside of the sample (population validity) or setting (ecological validity)


*  A study could be said to lack mundane realism??? if the procedures do not accurately reflect real-life situations. This is another important consideration when assessing generalizability. 

Replication

* If a study has been replicated in a different context, it could increase the possibility of generalizing / transferring the findings.
* Even when a procedure / RM does reflect a real-life experience, e.g., participant observation, it is almost impossible to replicate exactly, reducing the generalizability of the findings it produces to other people. Validity is often achieved at the expense of generalizability.