Partial aggregation for complex structural equation modelling (SEM) and small sample sizes: an illustration using a multi-stakeholder model of cooperative interorganisational relationships (IORs) in product innovation'
Where the research task involves collecting quantitative strategic data from senior management, low response rates and corresponding small sample sizes tend to be the norm. When structural equation modelling (SEM) is proposed, the challenge of undertaking sound research can be further compounded by the need to test a complex research models. A simple statistical technique which can help in this case is data reduction through partial aggregation of measurement models within the structural model.
This paper provides a brief review of literature on aggregation and composites – the foundation for aggregation. Then it illustrates the use of partial aggregation in SEM for a six-factor model of cooperative product innovation, including three distinct factors characterising interorganisational relationships (IORs) in relation to four different stakeholder groups. Sixteen multi-item measurement models were developed and empirically tested using survey data from CEOs of 120 Australian manufacturing companies. Partial aggregation was achieved by replacing each measurement models with a single summated scale or composite. The outcomes of this approach are a set of well-fitting SEM models at an overall multi-stakeholder level and a stakeholder-specific level.
von der Heidt, T & Scott, DR 2007, 'Partial aggregation for complex structural equation modelling (SEM) and small sample sizes: an illustration using a multi-stakeholder model of cooperative interorganisational relationships (IORs) in product innovation' paper presented to the 21st ANZAM 2007 Conference, Sydney, 4-7 December.