I very much enjoyed Nate Silver's tour of late-entry presidential candidates, which featured well-told quicky reviews of nine such enterprises. If there's an empirical conclusion to be had -- and given how few examples there are and how few of those were really viable candidates comparable to a Chris Christie type of last-minute entry -- it's that it may hurt more in Iowa and New Hampshire than it does in early national polling. That makes sense; after all, the campaign looks a lot different in the early states than it does to the rest of us. If you're in California or Texas or New York, the campaign is mainly driven by "free media" coverage; debates loom large, and the press decides which candidates get the most air time. In Iowa and New Hampshire, however, voters are exposed to loads and loads of paid media and personal outreach by the campaigns, along with a local media environment that is heavily influenced by candidate campaigns. Of course, the local campaigns are fueled by what's happening nationally (through fundraising, but also through elite signaling and, yes, national media coverage).
The point is that the process "works" because what happens in Iowa and New Hampshire is in fact (partially? mostly?) a function of what's happening nationally, but for that to happen the candidates actually have to translate national support into local campaigns, and it's possible that late-entry candidates can't do that. It's also possible that what's happening here is that late-entry candidates are doing well on one indicator of national campaign success -- polling numbers -- but fall behind on other indicators, such as high-profile endorsements and money. Let me try to be clearer...it's possible that an Iowa-specific weakness of the late-entry strategy would be very hard to distinguish from a national ("invisible") weakness expressed through poor Iowa/New Hampshire results.
Beyond that, with Silver's quantitative follow-up still to come, I'm going to caution strongly against concluding too much. He's really only talking about either one or two clearly viable late-entry candidates: Fred Thompson in 2008 and Wesley Clark in 2004. Thompson, famously, didn't appear to be in it to win it. Clark attempted a non-viable strategy (skipping Iowa), and it's not entirely clear that a semi-famous general is really a viable candidate to begin with. There's also Ted Kennedy in 1980, but in my view we should be extremely cautious about extrapolating from early reform-era contests (1972 through roughly 1980), because the system was still evolving; it's also hard to know how challenging an incumbent is different than a more typical open nomination.