[Just to make clear, though my answers should always be read this way: this is personal opinion]
You don't need to have any intention of answering a question to edit it.
the difficulties facing the initial user may not be as conspicuous because the editor may have inadvertently filled in some critical conceptual gaps in the OP.
Where feasible, it's important to have a question and answer that will be understood by and be useful to more people.
Mostly you should lean toward letting the OP's phrasing stand, but you can certainly encourage the OP to clarify. While some people would lean toward never fixing misconceptions in the question, if it's a side issue to the main question and fixing it is likely to clarify that misconception for the OP it may be okay. (It can sometimes be hard to be sure, though.)
The concern you mentioned is certainly possible so if you do edit, care must be taken - and you can always invite the OP to change anything back.
In the case of plots, and beyond subjective, aesthetic considerations, perhaps some subtle change in the presentation or the data may occur that defeats the purpose of the OP.
Such things are always a small risk with edits. If you're aware of the possibility, it's unlikely to be a major problem. Note that edits can be rolled back.
How should data points be posted to avoid presuming R familiarity or preference?
Certainly avoid editing to make anything R-specific; if it's already R-specific that's different. It's perfectly possible to edit to produce a cleanly-formatted ASCII table, for example, which will copy-paste (or easily be read) into a wide variety of packages.
e.g. something formatted like this:
A B
4 2
4 10
7 4
7 22
8 16
9 10
... is not R-specific but will read straight into R with a copy-paste (type OPdata=read.table(stdin(),header=TRUE)
then paste, press Enter and you have a data frame; or just read from clipboard
rather than stdin
)
So: plain ASCII text in a standard form (space delimited or comma-delimited, compatible with CSV) is desirable -- these can be read by almost anything. This is uniformly better that an image of numbers (since it contains the same information but can be used by answerers) ... but be very careful you don't change the data!
On this issue I got moderator guidance from @whuber advising to format them as text (usually comma delimited), or else include the data in the answer.
It's sensible advice; I'd add space-delimited to the mix of what's fine.
Should plots be always left untouched?
I don't have enough context to know what things you're proposing to fix, but usually you wouldn't alter a plot. Always is too strong.
There are some situations where I have edited a plot. One example is where there's a small plot with gigantic amounts of white-space around it (a plot smaller than 1/10 the size of a standard page (A4 or 8.5x11), say 3x2.5 inches sitting in a field of white pixels the size of a page. In that case I cut out almost all the white space.
Do you all agree that improving the grammar to make the question understandable is an OK reason for an edit in some cases?
Yes, almost always.
Is reformatting with latex to beautify equations OK without exceptions?
Nearly always, it's a good thing to do.
But without exception? That's hard to say, since exceptions may be things we haven't thought of. But in this case I can think of a few possible exceptions.
i. If I expect a post will move to SO I try to avoid LaTeX.
ii. If an equation can as easily be read as pseudocode (e.g. can very easily be turned into say a formula in R if left as is) I'm less inclined to edit it because there's another way to use it as it stands.
Your question is filled with requests for absolutes. That's not going to happen. Don't look for ironclad rules here. Just act sensibly, change things in ways that will clearly improve the post for as many people as you can, but with respect for the OP's intent.
If one is to edit (which is a very good thing to do), it's necessary to accept that:
You will occasionally make mistakes (of judgement, say)
Even when you act in very sensible ways, people will disagree occasionally; there can be legitimate difference of opinion
Don't get too bothered about these things unless there seem to be a lot of problems, in which case you may need to attempt to move your expectations toward the community norms more. Mistakes are not irretrievable -- as the saying goes, it's not rocket surgery.
Oh, and definitely put me down in the "the data were" column. "Datum" is singular, "data" is a plural noun.