For example, the concern that voting for your favorite candidate might help elect your least favorite choice (prevalent in vote-for-one plurality races with three or more candidates) is reduced by IRV. With plurality voting, voters need to know the latest polls to determine who the likely front-runners are, to plan how to strategically use their one vote, and so on. While every voting method can have possibilities of strategic voting in some situations, IRV is less prone to strategic manipulation than plurality or two-round runoffs.
Prof. Nicolaus Tideman in his latest book Collective Decisions and Voting used real-world election data to analyze the resistance to strategy of various voting methods. On a scale of 10, with 10 being perfectly resistant (which no system is), plurality voting got a 6.3, two-round runoffs got 8.1 and IRV got 9.7.
As to facts about real use of IRV in government elections, Australia has used it for generations. Here is an analysis of IRV use in the federal House of Representatives of Australia from 1949 - 2007.
IRV is also simple for voters to use (over 99.9% of IRV ballots were valid in Burlington's five-candidate mayoral races with IRV in 2006 and 2009, with no invalid ballots in the low-income/low education wards.
The main point in response to concerns about strategy, is that with IRV, strategy is less of an issue than with plurality elections or two-round runoffs. IRV also introduces NO new paradoxes or pathologies that do not already exist (and worse) under either two-round runoffs or plurality elections.
Concerns about whether voters need an advance degree to fully understand IRV are misplaced. Do voters fully understand how a plain old telephone works (how the switching works, etc. to reach the intended person)? No. But they can use it just as effectively as an electrical engineer. So too with IRV, there is convincing evidence that voters with less education use IRV as effectively as anyone else.