Deciding what to do with an outlier

Name:

Date:

Class:

Quick Questions

Shade in the A or B circle for each question.

1. A class survey shows one student's height as 1750 cm. What should you do first?

A Investigate: it is probably a typing error for 175 cm, so check the source and fix it

B Keep it, because every number in a data set must stay exactly as typed

2. A shop's sales data has one day with sales far higher than usual. The receipts confirm it was a real Christmas Eve rush. What do you do?

A Delete it, because it is much bigger than the other values

B Keep it, because it is a real extreme value, and mention it in your report

3. Which approach shows the effect of an outlier on your results?

A Calculate the mean with the outlier and again without it, then compare

B Calculate the mean once and hope the outlier does not matter

4. Your data shows one runner finishing a 100 m sprint in 0.9 seconds. The most likely explanation is:

A A real world record that you should keep without checking

B A typing error, such as 9.0 or 10.9, which you should check against the source

5. Your graph does not support your argument because of one unusual point. Is it okay to delete that point so the graph fits your story?

A No. Only remove data if you can show it is an error, and explain what you did

B Yes. Removing awkward points makes the conclusion clearer

Match the Pairs

Draw a line connecting each item on the left to its match on the right.

Left column		Right column
1=AVERAGE(B2:B30)-AVERAGEIF(B2:B30,"<>"&MAX(B2:B30))		ACompare the mean with and without the biggest value to see how much one outlier pulls the result
2=IF(B7>100,"Check source","OK")		BFlag a suspicious value so you can investigate it instead of deleting it
3If the outlier is a typing error, correct it using the original record (for example 950 should be 95)		CDeleting data just to fit your story is dishonest. The outlier was not checked first.
4Report the effect of a real outlier by finding the median, which is less affected by it, next to the mean		DA mark of 950 out of 100 is impossible, so it is almost certainly a typing error, not a real value.
5Delete row 7 because it ruins my trend line		E=MEDIAN(B2:B30)
6Mia scored 950 out of 100, so keep it as real data		FB7 = 95

Definitions

Investigate the outlier

Fill in the Blanks

Complete each line of code by filling in the missing parts.

=_______(B2:B9,">200")

=IF(B2_$F$1,"Check","OK")

=______(B2:B9)

=F2+___*(F2-F1)

=AVERAGE(B2:B9)_AVERAGEIF(B2:B9,"<200")

Spot the Error

Shade in the letter of the line that contains an error.

Question 1

A outlier ← 980

B check the source record

C IF typing error THEN delete it

D ELSE keep the value

E report the effect

Question 2

A Mean with outlier is 95

B Mean without it is 60

C The outlier raises the mean

D Report only the mean of 60

Question 3

A Age column: 15, 16, 15, 160

B 160 is a real age, so keep it

C Check the survey form

D Correct 160 to 16

Question 4

A Sprint times: 13, 14, 13, 9.6

B Check the timing record

C 9.6 was recorded correctly

D Change 9.6 to 13 to match

Question 5

A mean ← AVERAGE(B2:B9)

B median ← AVERAGE(B2:B9)

C Compare the two results

D Report the effect of B9

Answer Key

Deciding what to do with an outlier

Section A: Quick Questions

A — Investigate: it is probably a typing error for 175 cm, so check the source and fix it

B — Keep it, because it is a real extreme value, and mention it in your report

A — Calculate the mean with the outlier and again without it, then compare

B — A typing error, such as 9.0 or 10.9, which you should check against the source

A — No. Only remove data if you can show it is an error, and explain what you did

Section B: Match the Pairs

=AVERAGE(B2:B30)-AVERAGEIF(B2:B30,"<>"&MAX(B2:B30)) → Compare the mean with and without the biggest value to see how much one outlier pulls the result

=IF(B7>100,"Check source","OK") → Flag a suspicious value so you can investigate it instead of deleting it

If the outlier is a typing error, correct it using the original record (for example 950 should be 95) → B7 = 95

Report the effect of a real outlier by finding the median, which is less affected by it, next to the mean → =MEDIAN(B2:B30)

Delete row 7 because it ruins my trend line → Deleting data just to fit your story is dishonest. The outlier was not checked first.

Mia scored 950 out of 100, so keep it as real data → A mark of 950 out of 100 is impossible, so it is almost certainly a typing error, not a real value.

Section C: Definitions

Investigate the outlier — Check if an outlier is a typing error or a real extreme value. Fix errors, keep real values, and compare results with and without it. Never delete data just to fit your story.

Section D: Fill in the Blanks

=COUNTIF(B2:B9,">200")

=IF(B2>$F$1,"Check","OK")

=MEDIAN(B2:B9)

=F2+1.5*(F2-F1)

=AVERAGE(B2:B9)-AVERAGEIF(B2:B9,"<200")

Section E: Spot the Error

C — A typing error should be corrected using the source record, not deleted. It should read: IF typing error THEN fix it.

D — You should report the effect of the outlier by showing both means, with and without it. Showing only one hides the effect. It should be: Report both means.

B — No student is 160 years old, so it is most likely a typing error. It should say 160 is likely a typing error, which leads to checking and correcting it.

D — The record confirms 9.6 is a real extreme value. Real values must be kept, not changed to fit the pattern. It should be: Keep 9.6 and note it is real.

B — The median must be found with MEDIAN, not AVERAGE. Using AVERAGE twice gives the same result, so you cannot compare the effect of the outlier. It should be: median ← MEDIAN(B2:B9).