Validity

In general, validity refers to the accuracy of a quantitative research project. Validity is a concern in public relation research any time a survey instrument or sample is used. Use of a survey instrument [Page 960]raises questions about internal validity, as the accuracy of a research project is influenced by the planning, design, and execution of the project. A number of factors influence internal validity that are controlled through the experimental design. Few public relations research projects are true experiments, so this entry concentrates on the measurement aspect (i.e., surveys) of internal validity. The validity of a survey centers on whether or not you are actually measuring the concept you intended to measure.

An example will help to clarify the ideas related to validity. As part of a revision of an organization's employee communication system, suppose you decide to assess “communication satisfaction” with the various communication vehicles. A survey can be used to measure a variable such as communication satisfaction. But how do you know you are actually measuring the desired variable, in this case communication satisfaction? This is an issue related to internal validity. One check is called face validity. You, as an expert, carefully examine the survey to see if it captures the variable—that the survey reflects how you have defined the variable. A second check is content validity, where a group of experts on the topic review the survey. You could have experts on employee communication review the survey to determine if they think it captures communication satisfaction. A third check is criterion-related validity, in which your survey is proven to be related as anticipated to other established measures. You compare the results of your survey with the results of previously validated scales to see if they are related as predicted. The idea is to determine if the scores on your survey are consistent with the other scales. For instance, communication satisfaction should be positively related to job satisfaction, a variable that has validated measures. You would assess whether the communication satisfaction scores are positively correlated with job satisfaction scores. When two surveys correlate as predicted and correlate positively, this is known as convergence. You can also select a measure for a variable that you believe should correlate negatively with your survey, and this is called divergence. Although time consuming, testing validity is important when you create a new survey that is crucial to your public relations effort. It is better to invest the time and money in validating the scale than using inaccurate data—collected data that do not truly measure the intended variable.

External validity is the ability to generalize results from your sample to a larger population. Samples are used frequently in public relations research. For instance, you survey a small number of your customers, not all of them, to assess their knowledge of or attitudes toward an organization. External validity requires proper sampling techniques. The key is to use a sampling strategy that helps to ensure a representative sample—that is, the characteristics of your sample are the same as those of the population from which it was drawn. Your sample of customers should be representative of your customer population. If your primary customers are women from 20 to 35 years of age and your sample is mostly men 40 to 50 years of age, the sample is not representative. Refer to the “Sampling” entry for a discussion of proper sampling strategies for external validity.

...