Accuracy vs. Precision: Know What’s Worth Obsessing Over

Let’s face it, most of us did not get into GIS because we love obsessing over decimal places. Personally, I’m a truncate-er: no rounding, just cut it off. Which is either the wrong way or the right way, depending on who you ask.

But at some point, we all run into that classic question:

“Is this data accurate and precise? Wait, what’s the difference?”

Before you go down a rabbit hole trying to make your dataset “better,” it’s worth asking yourself:

“Do I even need to care about this right now?”

This week’s hack isn’t about pushing the right buttons or clicking faster; it’s about shifting your thinking so you don’t waste time chasing the wrong kind of correct.

Accuracy vs Precision

Accuracy - How close the data is to the real-world truth.  For example, the GPS point lands near to where the groundwater well is in the field.

Precision - How consistently or specifically the data is recorded. For example, every technician recorded the well location to 7 decimal places.  However, they used different GPS units, and the points form a tight cluster 45 feet from the groundwater well, not on it.

What this means is that data can can be precise but not accurate, or accurate but not precise, or neither, or both! What a mess!

Dartboard Analogy

• Accurate = You hit close to the center of the dartboard each time
• Precise = You hit about the same spot on the dartboard, but not the center
• Accurate and Precise = You hit the center of the dartboard every time
• Neither = You hit your friend standing next to the dartboard, and now you owe them a drink

Lazy Rule of Thumb

If you’re mapping for insight, go for accuracy. If you’re measuring or modeling, aim for precision.

Examples:

• Making a map for a report? Accuracy wins. You don’t need six decimal places.
• Modeling flood zones? Precision matters. That’s where meters (or centimeters) can really count.

Red Flags

• A shapefile with coordinates out to the 9th decimal place, collected with a handheld GPS - That’s very precise, but probably not accurate. Check for an accuracy column or metadata before trusting it.
• A data point “looks right” on the map but fails an internal tolerance check - It’s accurate enough for visualization, but not precise enough for modeling or analysis.    
• Someone says, “I copied the numbers exactly from the PDF.” - Possibly neither accurate nor precise , and now you have even more questions.

 GIS You Later!

Pro Tip: Don’t start cleaning a dataset until you know what the map is for. Sometimes “close enough” is enough.