I found leading by example, the best way to go ahead with this.
Initially, users were very prompt at downvoting and closing off questions very fast, without a comment in DataScience SE too, which is very bad for a new site(It was, then.), cause it again takes the same number of
reopen votes for reopening the question, after the OP has figured out what's wrong in their question and did the necessary edits.
So, an alternate approach is, first leave a friendly comment, like:
Welcome to <> SE. This question would most likely be closed cause < mention the reason properly and clearly >, and give the OP some time for revisiting the post, and properly edit it.
If the OP still doesn't edit/improve the post to fit into the community's policies, then by all means, go ahead and close it off.
Having said that, posts which are blatantly off-topic or which looks like
copy-pasted homework or which displays significant lack of effort(like, basic error googling, etc), should be closed down.
However, don't expect all the users to be as constructive as you are. It will take time for them to see how good constructive criticism is, and how it is, thereby improving the site as a whole.
PS: Some OP's who were given time to improve their first posts, are now regular users in DataScience SE, and are giving the same welcoming and moderation treatment to their fellow users now. And a site which was supposed to be dead long back, is now close to graduation :).
So, the culture will catch up, but it has to start with a person or two.