Upper and Lower Fence Calculator
Find Tukey-style inner fences, the interquartile range, and observations that fall outside the fence limits.
Inputs
Results
Outlier check
Sorted dataset detail
| Rank | Value | Classification |
|---|
How to use this upper and lower fence calculator
What this calculator does. This tool estimates the first quartile (Q1), median, third quartile (Q3), interquartile range (IQR), lower fence, and upper fence for a numeric dataset, then flags observations outside those fences as potential outliers. It uses the median-of-halves quartile convention and the familiar Tukey inner-fence rule. The standard multiplier is 1.5, so the lower fence is Q1 – 1.5 × IQR and the upper fence is Q3 + 1.5 × IQR. The result is an exploratory screening rule, not proof that an unusual value is erroneous. A legitimate rare observation can still fall beyond a fence. NIST describes the same 1.5 × IQR inner-fence idea in its guidance on box plots with fences.
When to use it. Use fences when you want a quick, resistant check for unusually low or high observations before deeper analysis; when you are reviewing a small batch of measurements for data-entry problems; when you are preparing a box-plot-style summary; or when you want an outlier rule that depends on the middle half of the data rather than the mean and standard deviation.
How to calculate. The calculator opens with a complete 12-value rainfall-style demonstration dataset, so the results and Excel workbook are available immediately. To analyze your own data:
- Replace the numbers in Dataset values with 2 – 50 finite values. You may use spaces, semicolons, line breaks, or a comma followed by whitespace as separators. Decimal points and scientific notation are accepted; decimal-comma input such as “1,5” is rejected because it is ambiguous with comma-separated data.
- Leave Use default multiplier (1.5) selected for the conventional inner fences. Clear it only when your method calls for a different factor, then enter that nonnegative number in Custom multiplier.
- Read the two fence values first, then review Q1, Median, Q3, IQR, the outlier summary, and the sorted table. Choose Download Excel to export the current validated model. Reset clears the demonstration dataset and calculated state; export remains unavailable until you enter a complete valid dataset again.
Input guide. Dataset values is required and accepts 2 – 50 real finite numbers with no unit imposed by the calculator; use the same measurement unit for every observation. A realistic entry is “12.4; 12.8; 13.1; 13.0; 18.9”. Higher or lower values affect quartiles only according to their ordered positions, while sufficiently extreme values may be flagged without greatly shifting the IQR. Mixing units, pasting thousands separators, or using decimal commas are common mistakes. Use default multiplier (1.5) is a required method choice: selected means the multiplier is exactly 1.5; clearing it activates the custom field. Custom multiplier is required only when the default box is cleared. For example, 3 produces wider fences than 1.5 and therefore usually flags fewer observations; a negative or nonfinite multiplier is rejected.
Output guide. Lower fence and Upper fence are threshold values in the same unit as the input data. Values strictly outside them are marked as potential outliers. Q1 and Q3 bound the middle half of the ordered sample under the median-of-halves convention. Median is the middle ordered value or the average of the two middle values. IQR is Q3 – Q1; an IQR of zero means the middle half has no spread, so the two fences may collapse to the same value. The live pills Values, Outliers, and Multiplier summarize the current sample size, flagged count, and rule. The Sorted dataset detail table lists each rank, numeric value, and classification as Inlier, Lower outlier, or Upper outlier. These are exact identities for the selected quartile convention and multiplier, not predictions.
Worked example. The opening dataset is 1.33, 1.96, 3.12, 2.20, 1.58, 2.04, 1.80, 6.32, 1.90, 3.84, 2.93, and 2.34. After sorting, the lower-half median is Q1 = 1.85 and the upper-half median is Q3 = 3.025. Therefore IQR = 3.025 – 1.85 = 1.175. With multiplier 1.5, the lower fence is 1.85 – 1.5 × 1.175 = 0.0875, and the upper fence is 3.025 + 1.5 × 1.175 = 4.7875. Only 6.32 lies outside those limits, so the first-open outlier count is 1.
How the fence method works
The interquartile range focuses on the central 50% of ordered observations, making it less sensitive to isolated extremes than the full range. Penn State's statistics material explains the IQR method for identifying outliers: establish Q1 and Q3, compute IQR, then place fences a selected multiple of IQR beyond those quartiles. Because quartile definitions vary among textbooks and software packages, two tools can occasionally return slightly different quartiles for the same small dataset. This calculator deliberately uses the median-of-halves convention because it reproduces the demonstrated calculation above.
Lower fence = Q1 – k × IQR
Upper fence = Q3 + k × IQR
Interpreting outliers carefully
A fence breach is a prompt to investigate, not an instruction to delete data. Check whether the observation came from a recording error, unit mismatch, sensor problem, or a real but unusual event. NIST's discussion of outliers and box-plot fences distinguishes inner and outer fences and shows why unusually distant points deserve special attention. If the sample is tiny, strongly discrete, or intentionally heavy-tailed, supplement the fence rule with domain knowledge and a view of the original observations rather than treating the label as a final statistical decision.