Measuring people's preferences regarding ageism in health: some methodological issues and some fresh evidence