Particularly in surveys that involve examining cause and effect, validity can be separated into two more categories: internal and external.
Internal Validity
There are a few variables to keep in mind when you attempt to establish internal validity. Internal validity refers to how well you have set up your study to rule out extraneous factors. It allows you to pinpoint which factors in your study are resulting in the observed effects. In other words, internal validity measures how well your study was designed. Here are a few things to look out for when checking internal validity:
- Events that might occur between the first and second measurements.
- Will the passage of time impact your study? This can be very long term (i.e., age) or in the immediate future (i.e., hunger or fatigue).
- There may be a change in the observers or scorers that might impact the outcome.
- There may be a separate variable in how you designed your survey that unintentionally influences the outcome.
- A pre-survey might increase interactional scores on a post-survey.
External Validity
External validity is a reflection of how well the findings of your study will apply in other situations. In other words, how can the findings of your study be applied to the real world? This might sound like an overwhelming question, but here are a few things to keep in mind when establishing external validity:
- The sample of people you use should match your population of interest as closely as possible, as the degree to which the sample is representative of the population is key.
- Any results that appear as a result of specific settings would remove the possibility of generalization of findings to other settings.
- Re-issuing the survey can help improve external validity if responses are the same or similar.
- Think about the location of your study, time of day, and other external factors that may influence how participants respond.
Of the two, external validity is more difficult to establish by far. For example, local cultures vary and can strongly influence outcomes in ways that are difficult to predict. Similarly, it can be practically impossible to account for all of the confounding factors that could change your research outcome.
