Is 'fair statistics' a thing? Given that statistics can often be abused to deliberately present 'facts' to support a pre-existing viewpoint. (Lies, damned lies and statistics).
And given confirmation bias.
Is there an established method of choosing which statistics to use, which ensures (or at least endeavours) to be as fair as possible? And does this area of enquiry have a name?
An example
If you want to compare the relative safety of flying vs. driving, you could choose from:


*

*fatalities per km

*fatalities per journey

*fatalities per time

*injuries per... 


etc.
Each one of these would yield different results which would 'pull the blanket' one way or the other. So how do you choose?
 A: I think statistics should always focus on their use. In your example, the overall question to compare the safety of flying with driving is hardly useful: You cannot choose to drive from New York to Paris in France by car. In turn, you cannot choose to go to your next supermarket by plane.
But you can decide if you want to travel from Madrid to Paris by either plane or car. This question is much more useful. Also, comparisons between both become automatically fairer because now you compare where both are more comparable. 
However, you might find that the actual choice between both by the subjects in your sample might depend on factors like cost, travelling time and luggage. You may find that these factors could also be related to security preferences. You could then try to adjust for these confounding factors by some model if you want to compare only the technical security itself. Or you don't consider these factors as confounders because you are interested in the security of the transportation including the different usages in terms of safety.
In the end, fair statistics is possible by isolating closely the factors involved from the factors you want to have a statement about.
Another requirement is to choose the measurable variable that represents best the point of interest: Is safety understood in terms of life hazards or in terms of injuries? There, it depends on what you are afraid of.
Also, there are questions that cannot be answered by statistical means due to lack of measurable variables.
In the end, fair statistics is possible by isolating closely the factors involved from the factors you want to have a statement about.
A: There is an area of enquiry named 'research methods' (students are often required to take a module in research methods prior to undertaking a research project as part of their course).  Its scope is wide, but one aspect is to draw a distinction between a 'research question', which indicates in general terms a  topic of interest, and a 'research hypothesis', which is sufficiently specific and precise to be suitable (if of a quantitative nature) for statistical testing. 
In terms of your example, the question 'Is flying safer than driving?' might be acceptable as a research question, but certainly not (even if reformulated as a statement rather than a question) as a research hypothesis.  An example of a good research hypothesis might be: 'The rate of fatalities per journey is lower for flying than for driving when travelling between European cities separated by between 100 and 1,000 miles'. 
The practical problem of course is that people often quote statistics when debating or arguing about questions that (like many research questions) are too vague to be resolved by statistical evidence, or indeed to be resolved at all.  Seeking 'fair statistics' doesn't really help here.  The choice is between trying to help people to see that the question is vague and perhaps identify a more precise question that is important to them, or perhaps judging that the debate is fruitless and declining to participate.
