The tendency to cherry-pick stats, take them out of context and force them into an alternative narrative by placing them in close proximity to a collection of other stats, is a troublesome trend. By engaging in this very process, you've ultimately made the decision to ignore the many other potential narratives that exist within the original dataset.
Over on her blog, Storytelling With Data, Cole Nussbaumer illustrates how even a comparatively small dataset can contain a number of different stories, and that by emphasising or de-emphasising different aspects of the chart you can change the audience's perspective of the data:
The same visualisation method may have been used each time, but by simply highlighting different areas, or changing the scale of the y-axis, a number of different narratives unfold. However, if you condense an entire dataset down into one summarised statistic, you lose these rich, potential narratives and strip said statistic of any context. For example we could choose to simply state that:
as a handy nugget of information alongside an illustration, or we could visually show the audience the comparison and context. This may not even be the most interesting part of this data, but by pulling out this one aspect and hiding the rest from the audience, you're intentionally or unintentionally obscuring the truth.
As Cole points out in her exploration, even a small dataset can contain many narratives ripe for the picking, but:
Presenting the dataset in an accessible way will reveal a much deeper story to the audience than a cherry-picked, standalone statistic ever could. Show your audience, don't tell them.