Preferably, you only rate those sites you have actually been to.
Having (rightly) adopted a very "laissez faire" policy around "allowing" people both to rate whatever sites and and at whatever level they wish we didn't put in place a "rule" that one can only rate a site one has recorded as having visited.
Today I was looking at the full rated sites list sequenced by "rating" and was interested to note Taputapuātea Marae Complex among just 4 WHS running at an "average" rating of 5*. None has received more than 3 ratings and Taputapuātea has only received 1. Interested to discover who had visited this site and rated it so (Surprisingly in my eyes) highly, I then discovered that the rater has not actually done so (or at least has not recorded having done so and doesn't appear to have traveled in the area on the basis of his "travel map"). The 2 members who have recorded a visit haven't done a rating.
The rating MAY be because the individual has a particular "interest" in/knowledge about Taputapuātea and, perfectly legitimately within our "rules", wishes to record this using the "insights" regarding the value of the site which this "interest" has given him, or could arise from data entry "finger trouble" (I have found that it is rather easy to click on a site's stars when moving the cursor down the full list without meaning to!!).
I don't think that there are many other examples of this. These small numbers are unlikely to make much difference to our overall ratings for those WHS with many ratings and, in the case of WHS with very few ratings, can be identified quite easily by those who wish to! No doubt an analysis could be done to identify how many examples there are of "rated but not visited".
If it is "finger trouble" with data entry then an additional warning message "Rated but not visited - are you sure?" might help to prevent them - indeed Community members might actually be pleased to receive such a warning which would prevent them making date entry mistakes. On the other hand this might be regarded potentially as "supporting" ratings of unvisited sites when we would prefer this to be a relatively rare occurrence. Another possibility might be to "prevent" the rating of an "unvisited" site altogether (if someone felt VERY strongly about wanting to rate a site they hadn't visited, but perhaps knew a lot about, then they would just have to tick it as "Visited" - as they could now!). The 3rd option of course is to leave things as they are whilst, perhaps, doing the analysis of such occurrences as suggested above so as to keep an eye on the frequency!