A good data scientist understands that poring over heaps of data in front of a computer screening, using algorithms and conducting the analysis is the most important part of his/her job. However, a great data scientist in addition to what a good one does, also goes out in the open, talks to people to get a real practical feel and does ground work. That is why while the former trust marketing research using surveys completely, the latter are aware that non-sampling errors play a big part. That is what led to the wrongful predictions made by the Princeton Election Commission and the New York Times in the recent US Presidential elections. It is important to note how the data was actually collected. Analysis of the said data may not be enough, instead genuine background research is necessary to get the full context of the situation. Also all theories learnt and observed must actually be applied in day-to-day circumstances.

Source:https://hbr.org/2017/01/the-best-data-scientists-get-out-and-talk-to-people