At the end of the quiz you can see
how other age groups scored. It's worth poking at. You can also read a
report summary that describes and explains their definition of "millennials", how ages have been grouped in previous surveys, how the surveys are conducted and the rationales.
The web quiz is designed around generalizations that are considered sufficiently accurate for predicting the demographic you belong to. Which is kinda like but not exactly the same thing as guessing your age.
More about creating the web quiz, and
more about the statistical analysis used.
Responses to subjective questions -- in this case, perceptions on interracial relationships, careerism, and the contemporary political spectrum -- can change for reasons other than the respondents' ages, since they can be influenced by news, media trends, personal events, or simply by being required to think about these things for a while. It also helps keep the participant focused on the quiz rather than whatever emotions the quiz has provoked. As long as the right questions are asked, keeping things general and relatively non-topical can make responses more reliable, and a shorter quiz is more likely to be completed than a longer quiz.
Attempting to break down generations in more granular ways (
do you have a smartphone or conventional cellphone or house phone? do you play RPGs, casual games, or none? are you communist or capitalist?) risks responses that have less to do with age than with geography, social status, wealth, education, and other factors.
In other words, the web quiz you see is probably more accurate than the web quiz you might have designed. Even if our posts in this thread doesn't make that sound likely.
In years past I've been responsible for some programming gruntwork on a variety of questionnaire-based studies. It grants me no special powers in those fields, but helping with the setup and talking with the researchers has given me an appreciation for how much of the accuracy of the statistical science can be affected by the craftsmanship of the survey designs.
The
detailed report (PDF) based on the survey is worth a read if this is the sort of thing you're into, and possibly even if it isn't. The appendices include all the survey questions and their responses, and shows that detailed questions were asked about politics, religion, interracial relationships, education, and so on... even how many tattoos participants had, and where. The Pew Research Center has been responsible for a great deal of large-scale surveys of this sort, and all the information is publicly accessible - not just in the literal sense that the data's there, but in the interest of informing the public in plain, non-technical language to the extent possible. They're good.