Outlier: An observation that falls well outside the overall pattern of the data.
1.5×IQR rule: Values below Q1 - 1.5×IQR or above Q3 + 1.5×IQR are considered outliers.
Interquartile Range (IQR): The difference between the third quartile (Q3) and first quartile (Q1).
- Arrange data in ascending order
- Find Q1 (first quartile) and Q3 (third quartile)
- Calculate IQR = Q3 - Q1
- Calculate boundaries: Q1 - 1.5×IQR and Q3 + 1.5×IQR
- Identify values outside these boundaries
Data is sorted: 10, 12, 15, 16, 18, 20, 22, 25, 28, 100
With 10 values (even), split into two halves: 10, 12, 15, 16, 18 and 20, 22, 25, 28, 100
Q1 = (12 + 15) ÷ 2 = 13.5
Q3 = (25 + 28) ÷ 2 = 26.5
IQR = Q3 - Q1 = 26.5 - 13.5 = 13
Lower boundary = Q1 - 1.5×IQR = 13.5 - 1.5×13 = 13.5 - 19.5 = -6
Upper boundary = Q3 + 1.5×IQR = 26.5 + 1.5×13 = 26.5 + 19.5 = 46
Any value < -6 or > 46 is an outlier
Value 100 > 46, so 100 is an outlier
Mean = (10+12+15+16+18+20+22+25+28+100) ÷ 10 = 256 ÷ 10 = 25.6
Median = (18+20) ÷ 2 = 19
Without 100: 10, 12, 15, 16, 18, 20, 22, 25, 28
Mean = (10+12+15+16+18+20+22+25+28) ÷ 9 = 156 ÷ 9 = 17.33
Median = 18 (middle value)
The value 100 is an outlier. The mean decreases from 25.6 to 17.33 when the outlier is removed (change of 8.27), while the median changes from 19 to 18 (change of 1). The mean is more affected by the outlier than the median.
• 1.5×IQR rule: Outliers are values outside [Q1 - 1.5×IQR, Q3 + 1.5×IQR]
• Mean sensitivity: Highly affected by outliers
• Median robustness: Less affected by outliers
Robust measure: A statistical measure that is not significantly affected by outliers.
Sensitive measure: A statistical measure that is significantly affected by outliers.
Range = 15 - 5 = 10
Q1 = 6.5 (average of 5 and 8), Q3 = 13.5 (average of 12 and 15)
IQR = 13.5 - 6.5 = 7
Range = 50 - 5 = 45
Q1 = 6.5 (average of 5 and 8), Q3 = 31 (average of 12 and 50)
IQR = 31 - 6.5 = 24.5
Range: Dataset A = 10, Dataset B = 45 (increase of 35)
IQR: Dataset A = 7, Dataset B = 24.5 (increase of 17.5)
The range increased by 350% (from 10 to 45)
The IQR increased by 250% (from 7 to 24.5)
Both measures are affected, but range is more sensitive
The IQR is more robust than the range because it only considers the middle 50% of data
The range is affected by both the minimum and maximum values
Dataset A: Range = 10, IQR = 7; Dataset B: Range = 45, IQR = 24.5. The outlier (50) increases the range by 350% and the IQR by 250%. The IQR is more robust than the range because it's only affected by the middle 50% of data.
• Range sensitivity: Affected by min and max values
• IQR robustness: Only affected by middle 50% of data
• Outlier impact: Range more sensitive than IQR
Standard deviation: A measure of the amount of variation or dispersion of a set of values.
Variance: The average of the squared differences from the mean.
Mean = (20+22+24+26+28) ÷ 5 = 120 ÷ 5 = 24
Deviations from mean: (20-24)², (22-24)², (24-24)², (26-24)², (28-24)²
Squared deviations: 16, 4, 0, 4, 16
Variance = (16+4+0+4+16) ÷ 5 = 40 ÷ 5 = 8
Standard deviation = √8 = 2.83
Mean = (20+22+24+26+60) ÷ 5 = 152 ÷ 5 = 30.4
Deviations from mean: (20-30.4)², (22-30.4)², (24-30.4)², (26-30.4)², (60-30.4)²
Squared deviations: 108.16, 70.56, 40.96, 19.36, 876.16
Variance = (108.16+70.56+40.96+19.36+876.16) ÷ 5 = 1115.2 ÷ 5 = 223.04
Standard deviation = √223.04 = 14.93
Dataset A: Standard deviation = 2.83
Dataset B: Standard deviation = 14.93
Change: 14.93 - 2.83 = 12.10
Percent increase: (12.10 ÷ 2.83) × 100% = 427.6%
The outlier (60) dramatically increases the standard deviation from 2.83 to 14.93
Standard deviation is highly sensitive to outliers because it squares the deviations
Mean changed from 24 to 30.4 (27% increase)
Standard deviation changed by 427.6% (much more sensitive)
Dataset A: Standard deviation = 2.83; Dataset B: Standard deviation = 14.93. The outlier increases the standard deviation by 427.6%. Standard deviation is highly sensitive to outliers because it involves squaring the deviations from the mean.
• Standard deviation sensitivity: Squares deviations, amplifying outlier effects
• Robust alternatives: IQR and MAD are less sensitive to outliers
• Outlier impact: Squared deviations make SD highly sensitive
outliers
outliers
sensitive
Outlier: An observation that falls well outside the overall pattern of the data.
Robust measure: A statistical measure that is not significantly affected by outliers.
Sensitive measure: A statistical measure that is significantly affected by outliers.
- Visual inspection: Look at plots (dot plots, box plots, histograms)
- 1.5×IQR rule: Values outside [Q1 - 1.5×IQR, Q3 + 1.5×IQR]
- Contextual analysis: Consider if value makes sense in the situation
• Most sensitive: Range, Standard deviation
• Moderately sensitive: Mean
• Least sensitive: IQR, Median
Real-world context: Applying statistical concepts to practical business situations.
Extreme event: An unusual occurrence that significantly affects business metrics.
Mean = (25+28+30+32+35+38+40+42+45+120) ÷ 10 = 435 ÷ 10 = 43.5
Median = (35+38) ÷ 2 = 36.5
Range = 120 - 25 = 95
Q1 = 30, Q3 = 42, IQR = 42 - 30 = 12
Without 120: 25, 28, 30, 32, 35, 38, 40, 42, 45
Mean = (25+28+30+32+35+38+40+42+45) ÷ 9 = 315 ÷ 9 = 35
Median = 35
Range = 45 - 25 = 20
Q1 = 29, Q3 = 41, IQR = 41 - 29 = 12
Mean: 35 → 43.5 (increase of 8.5, 24.3% increase)
Median: 35 → 36.5 (increase of 1.5, 4.3% increase)
Range: 20 → 95 (increase of 75, 375% increase)
IQR: 12 → 12 (no change)
Range is most affected (375% increase)
Mean is moderately affected (24.3% increase)
Median is least affected (4.3% increase)
IQR is unaffected by the outlier
For planning regular operations, median and IQR provide better insight
Range is inflated by the special event
Mean suggests higher sales than typical days
With outlier: Mean = 43.5k, Median = 36.5k, Range = 95k, IQR = 12k. Without outlier: Mean = 35k, Median = 35k, Range = 20k, IQR = 12k. The range is most affected (375% increase), followed by mean (24.3% increase). The IQR is unaffected, and the median shows minimal change.
• Range sensitivity: Most affected by outliers
• IQR robustness: Unaffected by extreme values
• Business decision-making: Use robust measures for regular planning
Data validation: Checking for accuracy and legitimacy of data points.
Outlier investigation: Determining if an outlier is legitimate or an error.
Sorted data: 15, 78, 82, 85, 88, 90, 92, 95, 100
With 9 values: Lower half: 15, 78, 82, 85; Upper half: 90, 92, 95, 100
Q1 = (78+82) ÷ 2 = 80, Q3 = (92+95) ÷ 2 = 93.5
IQR = 93.5 - 80 = 13.5
Lower boundary = 80 - 1.5×13.5 = 80 - 20.25 = 59.75
Upper boundary = 93.5 + 1.5×13.5 = 93.5 + 20.25 = 113.75
Since 15 < 59.75, it is an outlier
Mean = (15+78+82+85+88+90+92+95+100) ÷ 9 = 625 ÷ 9 = 69.44
Median = 88
Without 15: 78, 82, 85, 88, 90, 92, 95, 100
Mean = (78+82+85+88+90+92+95+100) ÷ 8 = 610 ÷ 8 = 76.25
Median = (88+90) ÷ 2 = 89
Potential causes: Absence, illness, misunderstanding instructions, data entry error
Should investigate before deciding to remove
If 15 is a legitimate score (student present, attempted exam), keep it
If it's due to data error or extenuating circumstances, consider removing
Always report both analyses when presenting results
The score 15 is an outlier (below 59.75). With outlier: Mean = 69.44, Median = 88. Without outlier: Mean = 76.25, Median = 89. The outlier significantly lowers the mean. Before removing, investigate the cause. If legitimate, keep it; if error, consider removal. Always report both analyses.
• Investigate first: Always determine cause before removing outliers
• Legitimate outliers: May be important to retain
• Transparent reporting: Show results with and without outliers
Outlier: An observation that falls well outside the overall pattern of the data.
1.5×IQR rule: Values below Q1 - 1.5×IQR or above Q3 + 1.5×IQR are considered outliers.
Robust measure: A statistical measure that is not significantly affected by outliers.
Sensitive measure: A statistical measure that is significantly affected by outliers.
Interquartile Range (IQR): The difference between Q3 and Q1, representing the middle 50% of data.
- Identify potential outliers: Use 1.5×IQR rule or visual inspection
- Calculate summary statistics: With and without outliers
- Investigate outliers: Determine if they are legitimate or errors
- Decide on action: Keep, remove, or analyze separately
- Report transparently: Show impact of outliers on conclusions
• Keep outliers: When they represent legitimate extreme values
• Remove outliers: When they are due to errors or impossible values
• Analyze separately: When they represent a different population
• Always investigate: Before making decisions about outliers
Mean: Highly sensitive to outliers
Median: Robust to outliers
Range: Very sensitive to outliers
IQR: Robust to outliers
Standard deviation: Highly sensitive to outliers
Analysis: Different summary statistics respond differently to outliers, affecting data interpretation.
- Mean: Affected by all values, including outliers
- Median: Unaffected by extreme values
- Range: Highly dependent on min/max values
- IQR: Focuses on central 50% of data
- Standard deviation: Squares deviations, amplifying outliers