According to Howarth, et al. 2011, from 4 wells they sampled, flow back after hydraulic fracturing lead to a release of methane equal to 0.6-3.2% of lifetime production of the well. Fracking wells released methane equal to 3.6-7.9% of lifetime production from all sources, while non-fracking wells released 1.7-6.2% of lifetime production.
Petron, et al. 2014 come to similar conclusions, though they don't provide data for lifetime individual well production to allow comparisons. They measured from an aircraft and found 10ppb to 100ppb increases in methane concentration in the air above wells at the Denver-Julenberg fracking site. They note that their estimated methane release is 2-3 times higher than the EPA's estimates (i.e. the estimates under which the fracking company filed its environmental impact statement). I don't like this study as much due to lack of information about what stage of production the wells are at (if they fracked two years ago and are still pumping, then this is just due to leaky pipes, not due to the fracking process).
In any case, there is some significant evidence that fracking is releasing more methane than it is expected to. The first paper adds some discussion of prevention measures and such.
I will say this, it barely matters in the long run if the methane is leaking or being pumped into your furnace and burned. Methane isn't going to last too long in an oxygen atmosphere, and its all going to end up as CO$_2$ sooner or later.
Crucially, there are no consistent baseline data on methane concentrations in groundwater in the USA collected before shale gas exploitation began, which makes it very difficult to assess and deal with the observed problems of methane in groundwater.
$\endgroup$