Handbook Of Training Evaluation And Measurement Methods, Third Edition
Jack J. Phillips
1997, Butterworth-Heinemann
Construct validity.
Construct validity refers to the extent to which an instrument represents the construct it purports to measure. A construct is an abstract variable such as a skill, attitude, or ability. Examples of constructs follow:
- Attitude toward supervisor.
- Ability to read a scale.
- Skill in conducting an effective performance discussion.
As a first step in defending construct validity, all parts of the construct must be defined and a logical case should be made showing that the instrument is an adequate measure of that construct. The definition of the construct should be as detailed as possible so that it is easy to understand. Construct validity can be defended in one or more ways.
- Expert opinion.
- Correlations.
- Logical deductions.
- Criterion group studies.
Expert opinion is a relatively easy approach in which a group of experts agree that the instrument, in their opinion, is an accurate measurement of the construct. Correlations are more complex. In this case, another instrument is used to measure the same or a similar construct, and the results are correlated with the first instrument. Statistically significant positive correlations could show construct validity. Another, more subjective, option is logical deduction. In this case, the instrument designer must logically conclude through a series of deductions that the instrument represents a measure of the construct. Criterion group studies can be more useful. A group of people who possess an abundance or deficiency of the construct in question are administered the instrument. If the results agree with the existing knowledge about the group, and it helps make the case for construct validity.
Construct validity is indeed a complex matter. Perhaps an example can help illustrate the process. An HRD program is conducted to improve company loyalty in a group of employees. Therefore, an instrument must be designed to measure loyalty before and after the program. During the program analysis, it is concluded that employees with high levels of company loyalty.
- Have a desire to work hard for the organization (high productivity).
- Become part of the organization’s goals and values (high job satisfaction, job involvement).
- Have a strong desire to remain with the organization (low turnover, low absenteeism).
The construct in this example is company loyalty. Job satisfaction and involvement are measured with a standard instrument. Data are also collected on productivity, absenteeism, and turnover. Employees who are perceived to have a high degree of company loyalty are administered the instrument. Data collected from the instruments show a positive relationship between company loyalty and job satisfaction, involvement, and productivity, and a negative correlation between company loyalty and absenteeism and turnover. This provides necessary support for the validity of the company loyalty construct.