Cherry picking data is easy to do, but it can lead to spurious results. often this is more of a problem with the media focusing on a particular issue and making exaggerated claims about the findings. in some cases, however, ther selection of data in the first place can skew the data. It is not always clear whether this is intentional or unintentional data manipulation and statics can give variable results depending on what analyses you apply to the data.
This Scientific American article does a good jobg of looking at the potentially very scarey conclusion that there is an increase in innfant morbidity on the US west-coast following Fukushima.
It seems that there is not an overall trend in infant deaths, so the previous analysis of data is misleading. Good thing about science is that it is skeptical and errors such as this can be picked up.
Skeptical kinkster musing on whatever takes my fancy!