It is neither obvious, nor is there any data that suggests that is the case for Bitcoin. I suspect people who hold this opinion form it passively while reading the occasional flashy news headlines "Drug lord used Bitcoin!" and don't actually do any research. After all, news reporters write about what drives readership, and "Grocery merchant used Bitcoin" just doesn't quite get the same engagement.
There are very few research papers that conclude a large percentage of Bitcoin's transaction is tied to illegal activity, for example [1] claims 46%. However they are fatally flawed: they fail to consider the amplifying effect of mixers/tumblers that, by design, create thousands of automated mixing transactions per original transaction. So 1 illegal transaction will look like 1000 transactions in their analysis... facepalm
[1] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3102645