# What is expected of a reviewer when reviewing answers

I'm fairly new to the review queues (maybe a couple weeks now) and have had the opportunity to review a number of Low Quality posts and First Posts (Q or A). I was recently reading Should there be two first-post review queues? and the author pointed out that they skip many of the "first answers" because they are in areas that are unfamiliar. In the past, when reviewing answers, I've essentially looked for things that are bad: non-answers, jibberish/spam (and recommended deleting them), or if it was reasonably good, but needed help (i.e. formatting/grammar) just make an edit. There have been a few cases which were well outside my area of expertise which I gave an honest effort to read. If the answer looked plausible (i.e. not jibberish/spam that included math symbols) then I said "Looks ok". If I couldn't tell if it was jibberish/spam+math symbols, then I'd skip.

My question: When reviewing an answer (first post or low quality) is it the reviewers job/duty to do a "peer review" specifically for correctness of the solution? I.e. if the user made an apparently honest effort at a solution but the solution is wrong, should I recommend that it be deleted? or just comment and point out the error (recommending that the Answerer delete their own post--or fix it if possible). If this is the case (that I should read also for correctness) then I will likely skip 10-20% more reviews than I do presently.

EDIT: Follow-up question--please let me know if I should instead post this edit as a separate question. (This is sparked by Robert Soupe's answer.) If the answer is incorrect (and not likely fixable), I can certainly downvote the question and make a comment. But how does the review process terminate? Clicking "looks ok" feels like an endorsement of the answer (which I don't want to do if it is incorrect), but there is no option under "recommend delete" that has the flavor "this answer is wrong." If I click "skip" then I send it off to someone else to review (who may be stuck in the same puzzle I was in).

• I described my approach here. No, you do not have to check correctness. Deleting total nonsense is good, though. – user147263 May 1 '15 at 3:36
• Good question. Hitting the skip button is always an excellent option if you are not sure what to do. I do it on occasion. If you are sure of what to do, for example, an answer is actually a question, then close it. Or say an answer is excellent but extremely curt, leave it open, or make your call. I comment here because this is not really an answer, but I think the skip option should be introduced into the conversation. – J. W. Perry May 1 '15 at 3:48
• "if the user made an apparently honest effort at a solution but the solution is wrong" It depends on the magnitude of the error. Something small, like a missing exponent, might call for a comment or even an edit. If their mistake is more fundamental, that correcting the answer would require a complete rewrite, maybe that's what a downvote should be for. – Robert Soupe May 1 '15 at 16:05
• "Looks OK" is Stackexchange parlance for being readable, coherent, ostensibly on-topic, and a [valid answer if it turns out to be correct]. There is no implication of endorsing or evaluating the correctness or general quality of the post. The evaluations are done through up/downvotes and comments. – zyx May 25 '16 at 20:48

No, I don't think you're expected to check for correctness. On Low Quality Posts review, the options are:

• Looks OK
• Edit
• Recommend Deletion
• Skip

So it "looks OK," not "is correct." Also notice that if you choose to recommend deletion, "This is incorrect" is not an option.

If you do notice the answer is incorrect but it can be easily fixed then that's what the edit option is for, e.g., someone writes $x < y$ where they should have written $x \leq y$. But if the answer is wrong because of a fundamental misunderstanding of the principles involved, leading to incorrect conclusions, then that's what comments and downvotes are for. If you have to put words in the author's answer to make it correct, then you're going too far.

As J. W. already mentioned, there is the skip option. I just skipped an answer about the Cauchy-Riemann equations. I know of those equations, but not enough about them to even say if the answer at least looks okay.

Of course this represents what I've gathered in my even shorter time reviewing at this level. I have probably failed to take into account the intentions of those who actually wrote up the review prompts.

• This kind of highlights the internal struggle I have. If the post is incorrect and not fixable, then I am sort of stuck. Is it low quality (if it is incorrect and not fixable)? I think yes? But, if I skip, that just passes the task off to someone else. But as you point out, "recommend deletion" doesn't give you the choice of "this is not correct." – TravisJ May 1 '15 at 17:01
• If the flaw is fundamental, then my option is usually downvote, comment, and vote to delete (as "no comment needed" - the options are merely a convenience for common deletion grounds IMO). Should the author find a new approach, posting a new answer would be best. – Lord_Farin May 2 '15 at 10:23
• @Lord_Farin I forget: how many downvotes cause automatic deletion? – Robert Soupe May 2 '15 at 16:45
• @RobertSoupe Deletion from the review queue happens if the answer gathers either 3 Delete votes from users with 20K or any combination of 6 votes "Recommend Deletion" and "Delete". The latter happens much more often than the former, because there are relatively few 20K users. – user147263 May 2 '15 at 19:34
• @2mkgz Thank you very much for that information. Let's say on an answer you and Hagen vote Delete, then TravisJ and I vote Recommend Delete. That would need just one 20K user voting Delete or two other users voting Recommend Deletion to make deletion happen, right? – Robert Soupe May 3 '15 at 15:16
• Yes, that is correct. The condition is (D=3 or D+RD=6). – user147263 May 3 '15 at 17:04