How To Lie With Statistics

Sampling Bias

• Every sample has some bias.
• A truly representative sample eliminates all bias sources.
• For a good random sample, every member of the group should have an equal selection chance.

Deceiving Averages

• “Average” is vague. Know if it’s mean, median, or mode.
• Outliers can inflate an “average,” skewing perceptions.

Incomplete Data

• Knowing the range is as vital as the average.
• Laws of averages only matter with numerous trials.
• Charts can mislead. Always check Y-axis values and units.

Insignificant Differences

• Differences matter only if they impact outcomes.
• Minor variances often get undue attention.

Misleading Graphs

• Graphs can misrepresent while technically being accurate.
• Adjusting scales can exaggerate trends. Beware!

Distorted Comparisons

• Ensure comparisons are consistent.
• Context matters. “27% of doctors smoke X” can be misleading without proper context.
• Distinguishing between correlation and causation is essential.

Mistaking Correlation for Causation

• Just because Y follows X doesn’t mean X caused Y.
• Correlations can be coincidental or deceptive.

Manipulating Data

• Maps can mislead due to disproportionate areas.
• Percentages can create illusions of precision.
• The base of calculations can shift perceptions.

Interrogating Statistics When faced with data, ask:

1. Who’s providing it? Check for biases and potential conflicts of interest.
2. How was it gathered? Look for potential sampling issues or biases.
3. What’s not being shown? Omissions can lead to false conclusions.
4. Has the topic shifted? Ensure conclusions align with the presented data.

Remember: Data doesn’t lie, but interpretations might. Stay critical and informed.