Common Significance Levels: 0.05, 0.01, and 0.10
You'll learn why researchers typically use α = 0.05, 0.01, or 0.10, and how to choose the right threshold for your analysis.
Why These Specific Values?
You've learned that the significance level (α) determines when you reject the null hypothesis. But why do we almost always see 0.05, 0.01, or 0.10? These values aren't magical—they're conventions that balance the risk of false positives with practical decision-making needs.
The Standard: α = 0.05
The most common choice is α = 0.05 (5% risk of Type I error). This strikes a middle ground: you're willing to accept a 1-in-20 chance of incorrectly rejecting H₀. Think of it as "reasonably strict but not extreme."
Most fields use 0.05 as the default when consequences of false positives are moderate.
The Stricter Choice: α = 0.01
When false positives are particularly costly—like in medical trials or quality control—researchers use α = 0.01 (1% risk). This is like raising the burden of proof: you need very strong evidence before declaring an effect real.
Example: Testing a new drug's safety requires stronger evidence than testing a website's color scheme.
The Lenient Choice: α = 0.10
In exploratory research or when missing a real effect is more costly than a false alarm, α = 0.10 (10% risk) is sometimes used. This is more permissive, suitable for pilot studies or screening for promising leads.
How to Choose
Consider:
- Field conventions (what does your discipline expect?)
- Cost of false positives vs false negatives
- Sample size (smaller samples may justify relaxed α)
- Exploratory vs confirmatory research
Key Takeaway: α = 0.05 is the conventional default, but choose 0.01 when false positives are costly or 0.10 for exploratory work—always match your significance level to the real-world consequences of your decision.