stats manipulation

In Spring
2015 a researcher published results of a study on the benefits of
eating chocolate to help lose weight. This article spurred many news
media reports on the benefits of chocolate, for example http://www.dailymail.co.uk/femail/article-3018945/New-study-reveals-eating-chocolate-doesn-t-affect-Body-Mass-Index-help-LOSE-weight.html in the Daily Mail. I personally saw a segment on the Today Show reporting the original article’s findings.

In
May, the author released news that his published student was not based
on science but essentially “junk”. His point is individuals including
the media need to be skeptical to what they read and make sure the
results are based on good science. Too often data can be manipulated in
a way to prove a particular argument especially in media coverage. http://www.npr.org/sections/thesalt/2015/05/28/410313446/why-a-journalist-scammed-the-media-into-spreading-bad-chocolate-science

What
are your thoughts on these manipulations? What can you do to help
distinguish between statistical conclusions that are likely to be valid
and those that are seriously flawed? Would you have believed the
original study?