I have been working on some site optimization for a while, and have run into a bit of a confusing issue. The biggest complaint from PageSpeed Insights for us, has pretty much always been "Defer parsing of Javascript". To fix this problem we started using LabJS to manage much of the javascript. Everything seemed great, and our Pagespeed scores moved to the mid 90s with most of the changes being things that were out of our control to change.
Recently however, the mobile report has suddenly started giving a much worse score when the mobile flag is set to true for testing. When I look at the "Defer parsing of Javascript" feedback, it shows much more javascript being parsed for mobile than for the normal report. What would cause this? Is this a problem with LabJS or simply the way we use it?
Here are links to the two reports.
I don't have a definitive answer on what's happening, but when you look at the list of what the report is complaining about, you have dozens of scripts from google and facebook (all appear to be like social icons, etc). My guess is that your mobile page is loading all of these in a blocking way (not using LABjs). Nothing in that list of resources looks like the typical kind of stuff you'd be loading with LABjs, so I think it might be a red herring.
I'd suggest taking a look at how you load those third-party resources (most likely, they load themselves). Maybe they have changed recently and are no longer doing a good job of loading asynchronously.