You need P(affecting outcome) x (value of positive outcome) > cost of voting.
Even assuming the value of a positive outcome is vastly bigger than the size of the world economy (e.g. $10^15, maybe $10^20), P(affecting outcome) is so fantastically low in any state with political leanings that you might as well not bother.
Littering in a public park is a separate issue - the effect of littering is cumulative, whereas the effect of voting is based on a threshold. Your litter makes the park incrementally worse. Donating blood does make a significant measurable difference - one particular person gets your unit of blood, and this helps them avoid death.
A dollar value can be assigned to all the things we've mentioned. Donating blood is worth multiple dollars, littering in a park is probably harmful to the tune of a few pennies. The dollar value of your vote is so small that double precision floats treat it as 0.
Mostly, there are two things to do before voting:
1) decide your position on multiple issues
2) think about how much the implementation of the system allows your positions to be communicated/counted (so some positions become more important because the system will allow them to have an effect)
both of these actions can be beneficial (and much so).
1 means thinking about life in society
2 means thinking about effective ways of agregating knowledge and preferences in a society
you could think of the elections as a holiday celebrating the fact that we live in a society, and inviting to think about it