> Could this be a situation where politicians have faulty data and use that to guide how they act?

I'm not aware of any data on this specifically with regard to drug control policy, but there is research from 2013 which reveals that American politicians vastly overestimate how conservative American voters are on multiple key issues.

See e.g.:

Primary source: https://www.vanderbilt.edu/csdi/miller-stokes/08_MillerStoke...

Popular media reporting: https://nymag.com/intelligencer/2018/11/study-congress-overe...

Further analysis: https://theprogressivecynic.com/2014/05/02/is-the-usa-center...