Why Two Data Points Don't Show Trends: Understanding the Importance of Comprehensive Data Analysis in Workplaces and News

The full blog post

Two data points are not a trend. Two-data-point comparisons can be mathematically correct but practically meaningless.

This is true in workplaces and news articles like this one.

Multiple two-data-point comparisons (comparing last month to the previous month AND comparing it to the year before) don't paint the full picture the way a simple run chart would.

If a hospital's margin is "23% higher" than the year before, is that a difference between 1% and 1.23% or the difference between 10% and 12.3%?

Give me more data points. Better yet, create a chart that shows trends (or the lack thereof) over time. Otherwise, we're just celebrating (or bemoaning) every little up and now.

23% sounds like a big change. But that doesn't mean it's statistically meaningful. Was it down 27% the previous month? Possibly. Some metrics simply fluctuate around a stable average.

On NPR recently, the hourly news update covered economic indicators, including the truth and data points that say:

• Gas prices are UP 5 cents from a week ago
• Gas prices are DOWN 10 cents from a month ago

So, gas prices are both going UP and DOWN. It depends on which data point you use as a starting comparison -- and what point you might be trying to prove. What are those two facts "indicators" of?? What's the longer term trend??