There is probably no hard data on all (or even most) the options you ask about, but this should be interesting:
“Money is fungible in a way that votes aren’t,” said Harvard political scientist and government professor Stephen Ansolabehere.
A hard question is how much, exactly, a single donation is worth. Some very rough estimates suggest it takes $40 on average to get a person to vote using traditional methods like canvassing, Ansolabehere said.
“If I give $200, that translates in some sense to five votes,” he said, “But campaigns are like really inefficient nonprofits. A lot depends on how much they’re spending on overhead, and the effectiveness of their methods.”
Indeed, there’s no magic formula to predict how much your cash will help. On average, current presidential candidates have spent about $10 for every popular vote they’ve earned to date, based on Federal Election Commission data and reported primary election results.
Of course, different campaigns see different “returns” on spending.
And they give some examples in which $20 in one campaign equals $4 in another campaign (both contemporary). Also
But the relationship between campaign spending and success with voters can be loose — and unpredictable. [...]
studies have shown that challengers get more voter “bang” for their campaign-spending buck than do incumbents, in part because lesser-known candidates have more to benefit from simply getting their name out.
I suspect the same is true for other things that money buy, i.e. there's probably a diminishing return for canvassing etc. as a campaign/candidate gets better known.
Interestingly, volunteering might not be as "fungible" as giving money:
That's what political scientists Ryan Enos and Eitan Hersh found in their paper "Party Activists as Campaign Managers," recently published in the American Political Science Review. As they note, modern presidential campaigns go out of their way to produce advertisements that are tailored for their local audience. They choose actors, backgrounds, themes, and language to appeal most effectively to Iowa farmers, Florida retirees, and Ohio factory workers. It's hard to beat the sophistication with which modern campaigns match an ad to its intended target.
But it's an entirely different story when it comes to recruiting campaign volunteers. Campaigns can't build canvassers the way they can manufacture television ads; they're dependent on the available pool of local volunteers. Who actually volunteers to give up their time to walk precincts and knock on doors in the weeks prior to an election? It's really not your average American. It's hard enough to get the average American to vote in most elections.
As Enos and Hersh find in their analysis of President Obama's campaign workers in 2012, those who will actually offer their labor tend to be somewhat wealthier, whiter, more educated, and more liberal than the electorate as a whole. They believe different things and prioritize issues differently than the people they're being sent out to persuade.
What's more, the volunteers often aren't from the same area to which they're being deployed. Volunteers can be very effective when they're canvassing in their local community, but quite often, those communities simply aren't competitive in the election. [...]
Does this actually end up hurting campaigns? That's difficult to say, but the effect is likely not large.
Note: the paper title as published is a bit different.
This seems to point to money buying personalized adds as ultimately more effective than canvassing, but I haven't seen a quantification of volunteer time (unit) as an equivalent adds-money value. And of course, with the right skills, one can volunteer for a big data operation.