The author has decided what they want the findings to be and has shoehorned these into the report regardless.
This is exactly right. The premise of the research - and indeed of the think tank - seems to be that people are nice, not nasty, and they mostly agree with each other about stuff.
This is nonsense. It's nonsense in general, and it's nonsense specifically in regard to this research.
How do you decide if people are "polarised" or not? The research found that people had different opinions. As dino said, it makes no sense to say that on a question where 46% agree and 32% disagree, people are not polarised, whereas on one where 19% agree and 57% disagree, there is a "risk" of becoming polarised.
Obviously it depends how you define "polarised", but I'd say a situation where people are equally divided on an issue (eg the EU referendum) is more polarised than one where the majority of people agree about something.