Abstract
Proper measurement of technology knowledge and social change enables managers to advance strategies in technology management. Structural equation modeling is the ideal method in Technological Forecasting and Social Change (TFSC) and other leading journals to assess the measurement quality of the relevant decision variables and understand how they are related. However, a myriad of indicators are now available to judge how suitable these measurements are (i.e., how well they fit). Despite a consensus that fit indicators are highly context-dependent and no “one-fits-all approach” emerges, a more contingent perspective is surprisingly missing. To fill this gap, we advocate for a “tailored-fit model evaluation strategy” that is specific to the situation at hand to exploit the particular strengths of fit indicators. Motivated by a synthesis of structural equation modeling in TFSC, our simulation study finds that three critical distinctions regarding (a) model novelty, (b) focus on measurement or structural models, and (c) sample size are vital. The proposed strategy demonstrates that, in many contexts, only a few indicators are recommended to avoid artificially inflated Type I/II errors. We provide a decision tree to reach more accurate decisions in model evaluation in order to better theorize and forecast technological and social challenges.