Marketing Claims
Case study
#Marketing #Pharma
How does your product compare to competing brands – as good, or demonstrably better?
🡒 Marketing claims require proof. While these might be quantified by consumer trials, their veracity ultimately rests on solid statistical analysis. We provide solid statistical analysis.
What we’ve done
A good consumer product study is conducted very much like a pharmaceutical Randomised Control Trial (RCT). There is randomised treatment allocations, double/triple-blinding, appropriate blocking, placebos/controls, power analyses, pre-defined Statistical Analysis Plans (SAPs) and protocols, etc. Subsequent analysis is similarly treated with the rigour of a clinical trial – ultimately providing defensible comparisons of treatments, estimation of effect sizes and due inferential care. Final claims of "treatment A is better than treatment B", "X% of subjects displayed improvements" are on a firm basis – fully defensible and robust under scrutiny.
Technical aspects
The studies here typically consist of repeated measures on subjects over time, often with within-subject treatment comparisons. Modelling is typically via mixed models (perhaps generalised) or similar (Generalised Estimating Equations – GEEs) to account for correlated errors. Problematic responses may call for computer-intensive inference e.g. block-bootstrapping. Studies frequently have a subject perception component, leading to discrete choice modelling.
#Marketing #Pharma
How does your product compare to competing brands – as good, or demonstrably better?
🡒 Marketing claims require proof. While these might be quantified by consumer trials, their veracity ultimately rests on solid statistical analysis. We provide solid statistical analysis.
What we’ve done
A good consumer product study is conducted very much like a pharmaceutical Randomised Control Trial (RCT). There is randomised treatment allocations, double/triple-blinding, appropriate blocking, placebos/controls, power analyses, pre-defined Statistical Analysis Plans (SAPs) and protocols, etc. Subsequent analysis is similarly treated with the rigour of a clinical trial – ultimately providing defensible comparisons of treatments, estimation of effect sizes and due inferential care. Final claims of "treatment A is better than treatment B", "X% of subjects displayed improvements" are on a firm basis – fully defensible and robust under scrutiny.
Technical aspects
The studies here typically consist of repeated measures on subjects over time, often with within-subject treatment comparisons. Modelling is typically via mixed models (perhaps generalised) or similar (Generalised Estimating Equations – GEEs) to account for correlated errors. Problematic responses may call for computer-intensive inference e.g. block-bootstrapping. Studies frequently have a subject perception component, leading to discrete choice modelling.