Should I delete the high score before I average?

Delete a score only when it is not a real observation, such as a mistyped 400 where the paper shows 40. A genuine high paper stays in the mean. If it distorts the story, report the median beside the mean rather than erasing the row. State the count you used so a later reader can rebuild the sum from the sheet without guessing which papers you kept.

Does a high spike always leave the mean above the median?

A single high spike pushes the mean up and leaves the median on a middle rank, so the mean usually sits above the median. A low spike does the opposite and pulls the mean down. Two spikes, one high and one low, can cancel in the mean and still leave a wide spread. Check both ends of the ordered list before you describe the direction of the gap.

How large a gap between mean and median is worth mentioning?

On a short list, a gap of a few points on the same scale as the scores is already worth a clause. In the five-score example the gap was 4.6 against scores that otherwise lived between 7 and 15. There is no universal cutoff that fits every unit. If swapping the extreme value moves the mean by more than the decision can ignore, put both summaries in the sentence.

Can I average percentages the same way I average points?

Only when each percent is one observation and each observation should pull equally. A 90 percent on a 10-point quiz and a 70 percent on a 100-point final are not equal rows. Averaging those two percents treats them as twins. Convert back to points, or apply weights that match the points at stake, before you call the result a course average. Keep the unit, the count, and the name of the formula in the same sentence as any figure you copy into a report.

What should I store so the mean can be checked later?

Store the sum, the count, the minimum, and the maximum, not the mean alone. From the sum and the count anyone can rebuild the average. From the minimum and the maximum a reader can see whether a spike was even possible. A file that keeps only the average cannot be audited when a score is challenged weeks later and the raw list is gone. Keep the unit, the count, and the name of the formula in the same sentence as any figure you copy into a report.

What a Mean Hides When One Score Spikes

What 40 points do to five scores

Five quiz scores of 12, 7, 9, 15, and 10 sum to 53. Divide by five and the mean is 10.6. Replace the last 10 with 40 and the sum becomes 83, so the mean rises to 16.6. Four scores are still 15 or lower, yet the published average now sits above almost every ordinary paper. The six-point jump came from one observation. Publishing 16.6 as the typical result describes the spike, not the cluster where the class actually sat. Put the median, or the count of scores below the mean, next to that average before it appears on a summary sheet.

Leave those four ordinary scores in place and imagine the spike is a makeup exam rather than a first sitting. The list is still 12, 7, 9, 15, and 40, and the equal share is still 16.6. If the makeup replaces a zero that was never a real attempt, average the four completed papers instead: they sum to 43, and 43 divided by 4 is 10.75. That is a different question from the mean of every recorded mark. Say which list you averaged. A reader cannot tell a dropped zero from a real 40 if the sheet prints only one number labeled average.

The equal share and the middle paper

Sort 7, 9, 12, 15, and 40 and the middle paper is 12, while the equal share is 16.6. The gap of 4.6 points is the spike showing up in one summary only. The Mean Calculator shows the share, the sum, and the count together, so the pull is easy to see: 40 of the 83 points sit in a single row. Do not quote 16.6 in a sentence that claims most students scored near it. Quote 12 as the middle position and mention that one paper is 40. A reader who has both numbers can decide whether the equal share is the story they needed.

A pay list makes the same split. Five hourly rates of 16, 17, 18, 19, and 62 dollars sum to 132, so the mean is 26.40. The middle rate after sorting is 18. Someone reading 26.40 might think a typical shift pays that much, yet four of the five rates are 19 or less. If you are describing a typical worker, use 18 and state the range from 16 to 62. If you are splitting a payroll of 132 dollars across five people, 26.40 is the right figure, and you should call it a share of the total rather than a typical wage.

Row share against share of the sum

On the spiked quiz the extreme paper is 1 of 5 rows, which is 20 percent of the list, but 40 of 83 points, which is about 48 percent of the sum. Row share and sum share should be read together. A score that owns nearly half the total will move an average even when it is rare in the list. Write both percents in the note under the chart: one row, about half the points. Readers then see why the mean sits so far from the ordinary scores. Without those two percents, 16.6 looks like a consensus the papers never gave.

Try a milder spike so the habit is not reserved for wild numbers. Scores of 70, 72, 74, 76, and 98 sum to 390, and the mean is 78. The high paper is still one fifth of the rows. It contributes 98 of 390 points, about 25 percent of the sum, and it lifts the mean from the middle score of 74 up to 78. Four points is a smaller jump, and it still matters on a scale where 80 is a common cut. If a boundary sits at 80, say that the mean cleared it and the middle paper did not.

Why a class of 30 shrinks one spike

Put the same 40 into a class of 30 whose other 29 scores average 10.6. Those 29 scores sum to about 307. Add 40 and the total is about 347, so 347 divided by 30 is about 11.6. The spike now moves the mean by roughly one point, not six. Class size is the shock absorber. A one-point move may be small enough to leave the headline mean in place, with the 40 listed as the maximum. Still check the middle of the ordered list. If that middle sits near 10 or 11, the mean and the center finally agree closely enough for a short report.

The reverse case is a section of eight. Seven scores at 11 and one score at 40 sum to 117, and the mean is 14.625, which you might round to 14.6. If the ordered middle pair are both 11, the median is 11, and the mean sits about 3.6 points higher. In a section that small, one paper is 12.5 percent of the rows. Do not send 14.6 to a department summary without the maximum and the section size on the same line. Small groups make every spike look like a change in the whole room, and the count is what stops that misreading.

A reporting line that keeps the spike visible

Use one sentence with three numbers: the mean, the median, and the maximum. For the five-score list that sentence is easy to write. The mean is 16.6, the median is 12, and the highest score is 40, on five papers. A reader can recover the story without opening the spreadsheet. If you have room for a fourth number, add how many scores fall below the mean. Here that count is four. Four of five below the average is a plain warning that the average is not the center of the pile. Use this sentence whenever the mean and the median disagree by enough to change a decision.

Round with a purpose, and do not average averages from groups of different sizes. Reporting 16.6 as 17 and the median as 12 still leaves a gap of about five points, so the rounding did not create the disagreement. A section of 8 with a total of 117 and a section of 30 with a total of about 347 do not combine by averaging 14.6 and 11.6. That shortcut gives 13.1 and treats the small section as equal to the large one. Add the totals and divide by 38. About 464 divided by 38 is about 12.2, and that is the equal share of the combined papers.

Times and totals where the mean stays honest

Five delivery times of 12, 14, 15, 16, and 18 minutes sum to 75, and the mean is 15 minutes. The middle time is also 15. No single stop owns the total. The longest is 18 of 75 minutes, about 24 percent, and the shortest is 12, about 16 percent. The mean is a fair summary of a typical stop and a fair share of the total time. Staffing a route from 15 minutes is reasonable if the stops resemble this list. Keep the range, 12 to 18, next to the mean, because a promise of 15 minutes at every door is more than an average can support.

Decide what a blank means before you trust any share. A missed quiz left blank is not a zero unless the written rule scores the miss as zero. Typing a blank as 0 pulls the mean down the way a 40 pulls it up. Five scores of 12, 7, 9, 15, and a true zero sum to 43, and the mean is 8.6. If the fifth paper was never sat, the mean of the four sat papers is 10.75. Write the rule in the header of the sheet. Then the equal share matches the decision you actually made about what counts as an observation.

Open the worksheet

The matching page is the Mean Calculator. It opens with a worked example, names the formula, and keeps the steps next to the result. Use it when you want the arithmetic on your own numbers rather than the argument on this page.