“…For example, MLMs have been used to analyse data clustered because of nesting of parents in couples (Garcia-Lopez et al, 2016;Hartley & Schultz, 2015;Jones et al, 2014;Langley et al, 2017;Pottie et al, 2009), families in households (Pottie et al, 2009), individuals with ID in community homes (Qian et al, 2015), support staff in organizations (Knotter et al, 2016). GEEs have been used to account for clustering caused by carers nesting within households (Totsika et al, 2017), multiple births nesting within women (Brown et al, 2016), or twins nesting within families (Cheng et al, 2015). A key difference in these two approaches reflected in the studies is that those studies that modelled their data using MLMs wanted to describe how much of the outcome variance could be attributed to the factors causing the clustering, whereas in GEE analyses this was not a main consideration, but a characteristic of the design that had to be controlled for prior to the interpretation of estimated parameters.…”