Decision Errors

Authors

Affiliations

Nate Yomogida, B.S., SPT

Doctor of Physical Therapy

B.S. in Kinesiology

Chloë Kerstein, B.S., SPT

Doctor of Physical Therapy

B.A. in Neuroscience

Type I Error

Overview

• Occurs when the null hypothesis is true and you mistakenly reject it¹.

Significance Level

• α refers to the chance of making a Type I Error (significance level)
  • Lower α results in lower chance of Type I Error
• Quantifying Type I error: p-value indicates the chance of making a type I Error
  • p-value: 0.20 = 20% α
  • p-value: 0.05 = 5% α

Prevention

• Lower α is a way to prevent type I Error
• Setting a lower signifiance level (p-value cutoff) decreases α

Type II Error

Overview

• Type II Error occurs when the research hypothesis is true, but you erroneously conclude it is false (accept the null hypothesis)
• This occurs when the results are not extreme enough to be considered statistically significant¹.

Risk

• Extremely low significance levels increase β and chance to make a type II error¹

Significance Level

• β refers to the probability of making a type II error¹.
• Higher p-cutoff score decreases β (chance of making type II error)¹

Prevention

• Setting a lenient signifiance level (\(p < 0.10\) or \(p < 0.20\)) is a way to decrease chance of type II error (β)¹

Caution

Do not confuse β when referring to type II error with “Standardized regressing coefficient (β)”¹

Balancing Type I and II Error

As seen above, if you decrease your p-value cutoff score, you are increasing α but decreasing β¹. Thus it is impossible to fully prevent the chance of making Type I and type II error. Finding a balance between these two errors is the key to a good study.

References

1.

Aron A, Coups EJ, Aron E. Statistics for Psychology. 6th ed. Pearson; 2013.