Hi Sarah
Thanks for these interesting examples. The MPM is really an assessment of knowledge of mammography, so the best validation approach is a bit different. The MPM would be working perfectly if women's answers corresponded perfectly with their actual knowledge. Mostly, this type of measure is evaluated based on face and content validity, rather than approaches more common for latent variables. I think the issue of whether deploying the MPM influences mammography rates is different (maybe more important in the end) - the MPM could be working perfectly but still have no link to mammography rates if knowledge/understanding is not really a key to mammography in the population.
Interesting article about cancer disparities! Their primary results are presented as risk ratios, and then they drill down for 4 categories and show absolute numbers. They probably had a hard time putting the absolute numbers for all the different cancer types considered on the same scale, because the absolute rates differ so dramatically. Still, it would have been great to have both absolute and relative estimates for all, because without this framing, it is difficult to evaluate the public health relevance of the socioeconomic disparities.
Nice description of the pros and cons of using census tract average SES.
Maria