The C-OAR-SE (Construct definition, Object classification, Attribute classification, Rater identification, Scale formation, and Enumeration and reporting) measurement method (Rossiter 2010) emphasizes high content validity as the only requirement of a measure. The C-OAR-SE method must replace the standard psychometric method (see Nunnally 1978 and Churchill 1979) because the psychometric method leads to measures that have low content validity. The low content validity of the measure generates "off-construct" results that are masked by psychometric statistics such as confirmatory factor analysis and coefficient alpha. Results from bad measures of multiple constructs are then typically hidden in impressive-looking but largely meaningless structural equation models (grandiosely referred to as "causal models". The "off" results remain undetected by editors and reviewers and end up in our journals as misleading and often entirely false findings. C-OAR-SE is a much better method for researchers to use when designing measures. The new method was suggested by the author's experience as a market research practitioner, where highly content-valid and single-item efficient measures are essential.
Available at: http://works.bepress.com/jrossiter/39/