It's different between web and iOS. For which are you concerned?

Why is it different for web and iOS?

Because I implemented it smarter the second time but haven't had the chance to overhaul the web version yet. In the iOS app, wrong character reviews aren't submitted on their own if you know the character much better than the word, because Skritter theorizes that you just forgot that it was that character in that word, not that you'd miss that character on its own.

So the word-level review info is the same whether you mark one or both characters wrong in either the iOS app or the web app, but in the web app, both characters will also get reviews submitted, whereas in the iOS app, one or both of the characters may not have wrong reviews submitted, whereas correct reviews _are_ submitted.

It doesn't actually schedule the word differently based on whether the first, second, or both characters are marked wrong, but each review scheduling has an element of randomness in it, so if you look at the scheduling you get from resubmitting it a different way, you'll see a slightly different review interval.