The researchers compared two models to understand how people choose between different products. They found that a more complex model called random coefficients logit (RC) gave different results than a simpler model called nested logit (NL). The new model they created, random coefficients nested logit (RCNL), was the best at predicting consumer choices. The RC model showed different patterns of substitution and a wider market definition. However, all models predicted price effects from mergers accurately. This study highlights the importance of considering different factors that influence how people make choices when buying products.