Construct Validity Assessment in IS Research: Methods and Case Example of User Satisfaction Scale

Construct Validity Assessment in IS Research: Methods and Case Example of User Satisfaction Scale

Dewi Rooslani Tojib (Monash University, Australia) and Ly-Fie Sugianto (Monash University, Australia)
DOI: 10.4018/978-1-4666-2059-9.ch008
OnDemand PDF Download:


Valid and reliable measures are critical to theory development as they facilitate theory testing in empirical research. Efforts in scale development have been put on ensuring aspects of validity. In this paper, the authors address a specific topic of construct validity assessment in scale development. Using data from the five leading IS journals between 1989-2008, in this paper, the authors determine if and how the field has advanced in construct validity assessment. Findings suggest that the proportion of studies reporting construct validity had increased and Confirmatory Factor Analysis (CFA), Exploratory Factor Analysis (EFA), and Multi-Trait Multi-Method (MTMM) were the three most common methods of construct validity assessment. The authors also apply a popular method from psychology and exemplify how the correlation analysis technique can be used to measure construct validity.
Chapter Preview

Construct Validity In General

Construct validity might seem complicated for those who are not familiar with instrument validity. Cronbach and Meehl (1955) provide an often-cited, easy-to-understand definition. They describe construct validity as a condition whereby items measuring one particular construct are considered together and provide a reasonable operationalization for that particular construct (compared with other latent constructs). Thus, an instrument demonstrates construct validity if, in measuring an intended construct, it measures the concept it purports to measure regardless of any other established instruments of other constructs (Nunnally, 1978; Zaltman, Duncan, & Holbek, 1973). Such assessment is important since, knowing the constructs are properly measured and measure what they are intended to measure, (1) will increase our confidence in the overall empirical results, and (2) will show how well researchers translate their theories into actual measures.

Complete Chapter List

Search this Book: