Tech companies often prefer job candidates with majors in STEM fields in their hiring practices. The day-to-day reality of the workplace, however, suggests they would be better served to focus on a different set of criteria.

Although researchers often spend little time discussing data preparation, it has the potential to massively alter a given study’s results. To ensure research remains useful, we need universal standards and better documentation.

The longstanding methods by which researchers establish statistical significance are based on a fallacy. To produce more meaningful results from data, it’s time we revise them.

More data can actually hurt when the collection methods are ignored.

Anyone with even a passing familiarity with data and statistics has heard the old maxim “correlation is not causation.” If that's true, though, why use statistics at all?

There are lies, damned lies and statistics. And then there are data visualizations.

Companies often want to apply grandiose, technical solutions to the problem of bad data. In fact, they would be better served to just take their spreadsheets more seriously.

Transparency within an organization is good for productivity and morale. Taken too far, though, it becomes surveillance and starts to have adverse effects.

Data may be the new oil, but that doesn’t make its collection inherently valuable. Organizations would do well to remain mindful of Goodhart’s Law to avoid letting a data-driven mindset warp their larger goals.

The STEM fields have been plagued by a crisis that threatens to erode public trust in the value of research. To solve it, we need to foster a more open and collaborative culture around research and publication.

When programming languages fall by the wayside, beware of putting them out of sight and out of mind — or they just might come back to bite you.

Empirical Bayes models are a vital part of the statistician’s toolbox, even if they can’t fully bridge the gap between Bayesian and Frequentist approaches.