On top of that, many concepts are only feasible to measure by using human-invented models+metrics, under very tightly controlled conditions. Oftentimes these metrics' equivalence to real-world circumstances is matter of faith or philosophy.
So everything is untrustworthy. The people doing it can be totally incompetent even in making simple observations. They can be wilfully or negligently unaware of competing evidence or errors in analysis. They can be measuring the wrong thing entirely due to historical precedent or because it's good publicity or because it's impossible to measure the thing you really want to test.
Ultimately this probably shouldn't be surprising to anyone that's experienced the real world practice of anything. But it is worth remembering that anecdotal evidence from your own eyes can be more predictive than statistical evidence from strangers.
I wouldn't restrict this to journalists. Wasted a year during undergrad working for a political scientist who knowingly ignored evidence on a historically oriented project that would render their fashionable hypothesis void. Bringing said evidence up in meetings on multiple occasions didn't help my prospects of getting anything positive out of the professional association (besides perhaps that should get out of the racket?).