TL;DR: The most common percent mistake is dividing by the wrong base—often the new value after a change, or the larger number, or 100 because the word percent contains “cent.” Re-anchor to the original or to the set the question named. A pretty 33% that used the new value is still the wrong 33%.
After-increase magnet
Price 80 → 100. Change 20. Percent increase is 20/80 = 25%, not 20/100 = 20%. The 100 is magnetic because it is money now. Now is not original. If the question is “what percent of the new price is the increase,” that is a different question (20%), and it should say so. Default increase questions want original in the denominator.
Students “check” 25% by taking 25% of 100 and getting 25, which is not 20, so they abandon 25% for 20%. They checked with the wrong whole. 25% of 80 is 20. Checks must use the same base as the claim.
After-discount magnet
Shirt was 50, now 40, $10 off. Off is 10/50 = 20%, not 10/40 = 25%. Sale tags that say “25% off” using the new price as a vibe are lying or confused. 10 is 25% of 40, which is “discount as a percent of what you pay,” a rare framing. Retail “% off” is of original.
If you reverse a 20% off to find original from 40, original is 40/0.80 = 50. Using 40 × 1.20 = 48 is the markup identity in the wrong store. Wrong base, wrong identity, nearby dollars.
100 as a security blanket
72 points on an 80-point test is 90%, not 72%. Dividing by 100 because percent “means out of 100” skips the actual whole of 80. Out of 100 is a display. Out of 80 is the measurement. Convert 72/80 first, then you may write 90/100 if you like the display.
Survey: 72 of 80 respondents is 90% of respondents, not 72% of a fantasy 100 respondents. Inventing 20 silent people to make a 100 is not a method.
Subset bases in word problems
“30% of the students who walk” is not 30% of the class unless all students walk. If 40 of 200 walk, 30% of walkers is 12 students, which is 6% of the class. Feeding 0.30 × 200 is the class base wearing a walker costume. Underline the clause after of.
Conditional percents are where wrong bases go to look advanced. Draw a tree: 200 → 40 walkers → 12. Trees beat slogans.
How to recover when the answer key disagrees
If you got 20% and the key says 25%, you probably used new. If you got 333% and the key says 30%, you swapped part and whole. If you got 4.5 and the key says 45, you dropped a factor of 10 (percent vs decimal). Those three diagnoses cover a shocking fraction of “the calculator is broken” emails.
Recompute the same numbers with labeled fields after you write the base in words. If the site matches your wrong 20%, you taught the site the wrong story. Fields are obedient. Bases are the work.
Teachers can name a weird base on purpose
Occasionally a problem wants percent of the new, or of the average, or of capacity (80 riders of a 100-capacity bus as 80% full, while a change from 70 riders is a different percent). Weird is legal if named. Your job is to detect whether the base is named weirdly, not to assume every percent is (new−old)/new because last week’s wrong answer was close to that. Close is how wrong bases survive.
Write “base = original $80” as a dedicated line. If you cannot fill that line, you are guessing. Guessing can hit 20% on an 80-to-100 problem and feel blessed. Blessed 20% still misses 25%.
Capacity, “full,” and three tempting wholes
80 riders, 100 seats, 70 riders yesterday. Percent full today = 80/100 = 80% (capacity whole). Change in riders = 10/70 ≈ 14.3% (yesterday whole). Share of today’s riders who boarded at the last stop needs a third whole (today’s 80). Three honest percents. One data set. If a dashboard shows a single 80% in red, ask which whole it used. Red 80% of capacity is not a 80% increase. Wrong-base dashboards are how agencies “panic up” or “talk down.”
Always keep a legend: whole = seats, or whole = yesterday, or whole = today’s riders. Legends are not decoration. They are the formula’s noun. Without a legend, 80% is a mood.
Refunds and the original as a magnet
You paid $80 after a 20% off $100 ticket. Refund policies may refund $80 (what you paid) or $100 (list) minus restocking of list or of paid. Restocking 15% of which whole? If 15% of list, $15, you get $85 back on an $80 payment, which is a gift or a glitch. If 15% of paid, $12, you get $68. Wrong-base refunds are customer-service math. Ask which whole before you spend the expected cash.
A quiz that moved from 70 to 84
Last quiz was 70 points. This quiz is 84. The change is 14 points. Percent increase uses last quiz as the whole: 14/70 = 20%. The magnet is 84, because it is the paper in your hand, and 14/84 is about 16.7%, which answers “what percent of the new score is the gain,” a question nobody asked. Students then “check” the 20% by taking 20% of 84 and getting 16.8, which does not return them to 70, so they abandon the correct 20% and adopt the wrong base that happens to feel checkable. The forward check is 70 times 1.20 = 84. If that lands on the new score, the original was the right whole.
A drop from 84 back to 70 is not a 20% decrease. The loss is still 14, but the whole is now the 84 you are falling from: 14/84, about 16.7%. Direction picks the starting number. The larger score is not automatically the base, and it is not automatically forbidden either.
Town population and the census you can see
A town had 12,000 residents in the last census and 13,800 now. Growth is 1,800 / 12,000 = 15%. A newsletter that divides 1,800 by 13,800 gets about 13% and calls the boom smaller because this year’s headcount is sitting on the mayor’s desk. This year’s headcount is the result, not the base, unless the sentence is “newcomers are what percent of today’s town.” That second sentence is about composition, not growth. Growth percents look backward to the earlier count. Composition percents look around at the current set. If the newsletter wants both, it must print both wholes, or readers will treat 13% and 15% as a dispute about arithmetic rather than a dispute about which noun was promised.
