At the same time, you don't have to agree with the political scientists who are constantly telling us that "running mates don't matter"--which really means that running mates don't have any real impact that can be detected with statistical analysis--to believe that Pence won't do much that's truly positive for Trump. Indiana is a safe Republican state; if it's sufficiently in play for Pence's presence on the ticket to possibly make a difference, Trump is in real trouble.
That doesn't follow, if you're saying what it is I think you meant to say. Suppose it's true that running mates have made no statistical difference in the past. Why would a statistically verifiable positive contribution spell trouble?
Here's what I was trying to say:
1) There's a fair amount of research by political scientists that concludes that running mates do not really make positive contributions to a presidential ticket--people vote for the presidential candidate, not the running mate. A running mate can be a negative at times (see Palin, Sarah--although McCain would almost certainly have still lost with a Not!Palin running mate), but they don't really provide any positive boost, according to the statistical research.
2) But, suppose you are skeptical about that research. Social science research isn't physics, after all. What, then, would traditional analysis (the sort we get from political journalists, pundits, et. al.) suggest are ways that Pence could help Trump win the election. One of the big "traditional" ways your running mate is supposed to help you (as a presidential candidate) is by giving your candidacy a boost in his/her home state. The classic "traditional" example of this is Johnson supposedly enabling Kennedy to win Texas back in 1960.
3) Pence, however, doesn't really offer Trump any help in that regard. He's from Indiana, traditionally a very safe Republican state. Aside from Obama narrowly carrying the state in 2008, every Republican from 1968 to the present has carried Indiana pretty easily--and Obama's win was a real outlier that Clinton is very, very unlikely to duplicate. So, the only way that Pence's presence on the ticket matters in Indiana is if there is a large enough swing nationwide to the Democrats to put the state in play; probably that would need to be at least a 4-5% swing compared with 2012. A 4-5% swing to the Democrats would make states like Ohio and Florida pretty safe Democratic wins, while probably flipping North Carolina and putting at least some of Arizona, Georgia, Missouri and South Carolina in contention. If that did happen, Trump would have much bigger problems than whether or not Pence would help him carry Indiana.