Market projections are often undermined by non-representative samples and the tendency of respondents to overstate preference share. Both of these problems must be addressed before estimates can be generated for market sizing.
To make inferences about the overall market based on a research sample, careful consideration should be given to sample design and recruitment, sample size, possible coverage and nonresponse biases, self-selection biases when opt-in web panels are utilized, and much more.
Sample matching or propensity score adjustments may be needed to improve sample representativeness; and post-stratification weighting schemes are almost always needed for market projections.
Further, a plethora of research have shown that stated preference measures tend to overestimate market share compared to measures of actual purchase behaviors, mainly because there are lots more confounds and restrictions in the real marketplace. Hence, it is often necessary to apply an adjustment factor to estimates of stated preference share that correct for this overstatement, mainly drawing from econometric data on competitive products and analog products.
I can help with all the above aspects of market sizing and more. So contact me to discuss your project needs!