GMAT Data Insights lesson · Weighted Average
GMAT Data Insights: weighted averages and combined rates
Combining rates across groups on GMAT Data Insights: weight each rate by its count, and what leaving a group out does.
The core idea
A combined rate is total (rate × count) ÷ total count. It lands nearer the bigger group's rate. The plain average of the rates is right only when the groups are the same size.
Core ideas
Turn rates back into counts: 95% of 1,200 responses is 1,140 approvals. Add the counts, add the totals, divide.
A combined rate is never just the average of the rates unless the groups are the same size. Each group "votes" in proportion to its count.
A small case. Two classes sit the same test.
| Class | Students | Pass rate | Students who passed |
|---|---|---|---|
| Red | 300 | 80% | 0.80 × 300 = 240 |
| Blue | 100 | 60% | 0.60 × 100 = 60 |
| Both | 400 | ? | 300 |
Combined pass rate: 300 ÷ 400 = 75%. The plain average of 80% and 60% is 70%. The answer is nearer 80% because Red has three times as many students.
- For each group, multiply its rate by its count: that gives the "hits" (passes, approvals, hectares irrigated).
- Add the hits.
- Add the counts.
- Divide total hits by total count.
Two quick checks before you compute:
- The combined rate always lies between the lowest and the highest group rate.
- It lies nearer the bigger group's rate. If an option is outside the range, or nearer the smaller group, it's wrong.
Combined rate in a table
Why it works. A rate is hits ÷ count, so putting two groups together means adding their hits and adding their counts. That makes each group's rate count in proportion to its share of the total, which is why the result lies between the two rates and nearer the bigger group's.
How to spot it: a table with a rate column and a size column, and "the percent of the two farms' total land that is irrigated".
Example. Two farms. What percent of their combined land is irrigated?
| Farm | Land (ha) | Irrigated | Irrigated land (ha) |
|---|---|---|---|
| Amberfield | 100 | 50% | 50 |
| Greenacre | 190 | 75% | 142.5 |
| Both | 290 | ? | 192.5 |
- Turn each rate into hectares: 50% of 100 = 50; 75% of 190 = 142.5.
- Add the hectares: 50 + 142.5 = 192.5.
- Add the land: 100 + 190 = 290.
- 192.5 ÷ 290 ≈ 66.4%.
Trap: (50% + 75%) ÷ 2 = 62.5% treats the farms as equal in size. That would be right only if both had the same land.
Try it. Two clinics. Lake has 400 patients, 30% of them vaccinated. Hill has 600 patients, 55% of them vaccinated. What percent of all the patients are vaccinated?
Show the answer
- Lake: 30% of 400 = 120. Hill: 55% of 600 = 330.
- Vaccinated: 120 + 330 = 450, out of 400 + 600 = 1,000 patients.
- 450 ÷ 1,000 = 45%.
- Check: 45% is between 30% and 55%, and nearer Hill's 55%, the bigger clinic. ✓ (The plain average, 42.5%, is wrong.)
Edge cases.
- Equal sizes: the plain average of the rates is right.
- The distances are in the inverse ratio of the sizes. In the class example, 75% is 5 points from Red's 80% and 15 points from Blue's 60%, and 5 : 15 = 1 : 3 is the reverse of Red : Blue = 300 : 100.
- So a combined rate can give you the sizes. If you know the three rates, the distances give the ratio of the group sizes.
GMAT-level example. A company makes 50,000 units a month at two plants. The defect rates are below. How many units does Plant B make?
| Plant | Units | Defect rate |
|---|---|---|
| A | ? | 2% |
| B | ? | 5% |
| Both | 50,000 | 3.2% |
- Defects in all: 3.2% of 50,000 = 1,600.
- Let A make a units, so B makes 50,000 − a. Then 0.02a + 0.05(50,000 − a) = 1,600.
- 0.05 × 50,000 = 2,500, so 2,500 − 0.03a = 1,600, and 0.03a = 900.
- a = 900 ÷ 0.03 = 30,000. Plant B makes 50,000 − 30,000 = 20,000 units.
- Check: 2% of 30,000 = 600 and 5% of 20,000 = 1,000, and 600 + 1,000 = 1,600 ✓.
A faster way, using the distances: 3.2% is 1.2 points from A and 1.8 points from B. The sizes are in the reverse ratio, A : B = 1.8 : 1.2 = 3 : 2, so B makes 2/5 of 50,000 = 20,000.
Check yourself. One group of 200 has a 40% pass rate, and another group of 300 has a 90% pass rate. What is the combined pass rate?
Check your answer
40% of 200 = 80 and 90% of 300 = 270. (80 + 270) ÷ 500 = 350 ÷ 500 = 70%.
Rates on a chart with counts given
Why it works. A chart of rates hides the group sizes: a bar for 10 people looks the same as a bar for 10,000. The counts in the note put the sizes back. Rate × count turns each bar into hits, and hits and counts are the things that add up.
How to spot it: a bar chart of percents, with the group sizes given in a note or the question.
Example. A bank surveyed customers at two branches. The chart shows the approval rate; Main Sq. had 1,200 responses and Mill Ln. had 1,000. What was the approval rate across both branches?
- Main Sq.: 95% of 1,200 = 1,140 approvals.
- Mill Ln.: 45% of 1,000 = 450 approvals.
- Total: 1,140 + 450 = 1,590 approvals out of 1,200 + 1,000 = 2,200 responses.
- 1,590 ÷ 2,200 ≈ 72.3%, so about 72%.
Main Sq. gives 1,200 of the 2,200 responses, about 55%, so the result leans a little toward its 95%.
Trap: the plain average, (95% + 45%) ÷ 2 = 70%, gives Mill Ln. as much say as Main Sq. Using only the bigger branch's rate (95%) ignores Mill Ln. altogether.
Try it. A chart shows the share of survey responses that were "yes" in three age groups: 40% for ages 18–34 (500 responses), 60% for ages 35–54 (300 responses) and 75% for ages 55 and over (200 responses). What percent of all responses were "yes"?
Show the answer
- Yes answers: 40% of 500 = 200; 60% of 300 = 180; 75% of 200 = 150.
- Total: 200 + 180 + 150 = 530, out of 500 + 300 + 200 = 1,000 responses.
- 530 ÷ 1,000 = 53%.
- Check: the plain average of the three bars is about 58.3%. The real rate is lower because the biggest group has the lowest rate. ✓
Edge cases.
- Sizes given as shares work as weights. If a note says a branch gave 55% of the responses, use 0.55 as its weight, with no counts needed.
- No sizes anywhere? Then the combined rate can't be found, only a range between the lowest and highest rate (see what can be determined).
- Check what the bars show. Bars of counts add directly; bars of rates need weights.
GMAT-level example. An airline's on-time rate at three airports is shown below. Alder handles 25% of its flights, Birch 45% and Cedar 30%. What is the on-time rate across all three? If Birch's rate rose to 88%, by how many percentage points would the overall rate rise?
- Weight each rate by its share of flights: 0.25 × 92 = 23; 0.45 × 78 = 35.1; 0.30 × 85 = 25.5.
- Add: 23 + 35.1 + 25.5 = 83.6%.
- Birch's rate rises by 88 − 78 = 10 points, and Birch carries 45% of the weight, so the overall rate rises by 0.45 × 10 = 4.5 percentage points, to 88.1%.
- Check with 1,000 flights: 230 + 351 + 255 = 836 on time, 83.6% ✓. At 88%, Birch's 450 flights give 396 on time instead of 351, so 45 more, which is 4.5% of 1,000 ✓.
The trap answers are 85% (the plain average of 92, 78 and 85) and a 10-point rise (forgetting that Birch is only 45% of the flights).
Check yourself. In the bank example, Main Sq. had 1,200 responses at 95%. How many responses at 45% would Mill Ln. need for the combined rate to be exactly 70%?
Check your answer
1,140 + 0.45m = 0.70 × (1,200 + m). So 1,140 − 840 = 0.70m − 0.45m, which gives 300 = 0.25m, and m = 1,200. With equal sizes, the combined rate is the plain average of 95% and 45%, which is 70%.
Leaving a group out
Why it works. The combined rate is total hits ÷ total count, so leaving a group out means taking its hits and its count off both totals. A group below the combined rate drags it down, so removing it lifts the rate. The bigger the group and the farther its rate is from the combined rate, the bigger the jump.
How to spot it: "If the South depot were excluded, the on-time rate for the remaining depots would be ___".
Remove the group's count from the total and its rate × count from the numerator, then divide again. A big group with a low rate moves the combined rate a lot.
Example. Parcels and on-time rates at three depots. What is the on-time rate for all three, and for the depots other than South?
| Depot | Parcels | On time | On-time parcels |
|---|---|---|---|
| North | 2,000 | 90% | 1,800 |
| South | 1,500 | 60% | 900 |
| East | 500 | 80% | 400 |
| All three | 4,000 | 77.5% | 3,100 |
- All three: 3,100 ÷ 4,000 = 77.5%.
- Without South, subtract its parcels from both totals: (3,100 − 900) ÷ (4,000 − 1,500) = 2,200 ÷ 2,500 = 88%.
- Without East instead: (3,100 − 400) ÷ (4,000 − 500) = 2,700 ÷ 3,500 ≈ 77.1%. East is small and close to the average, so it barely matters.
Adding a group works the same way in reverse. If a fourth depot with 1,000 parcels at 70% joined the three, the rate would be (3,100 + 700) ÷ (4,000 + 1,000) = 3,800 ÷ 5,000 = 76%.
Whichever way the group moves, rebuild both totals (hits and counts) and divide once more.
Trap: averaging the two remaining rates, (90% + 80%) ÷ 2 = 85%, drops the weights again. North has four times East's parcels, so the true answer (88%) sits much nearer North's 90%.
Try it. In the depot table, what is the on-time rate for the depots other than North?
Show the answer
- Take North's 1,800 on-time parcels and its 2,000 parcels off the totals: 3,100 − 1,800 = 1,300 on time, out of 4,000 − 2,000 = 2,000.
- 1,300 ÷ 2,000 = 65%.
- Check: North's 90% is above the combined 77.5%, so removing it must lower the rate. ✓
Edge cases.
- A group at exactly the combined rate changes nothing when you remove it.
- Removing a group above the combined rate lowers it, as with North.
- Rates alone aren't enough. To take a group out you need its count, or its share of the total.
GMAT-level example. A company's sales leads and conversion rates by region. Dropping which one region would raise the combined conversion rate the most, and to what?
| Region | Leads | Converted |
|---|---|---|
| North | 1,000 | 16% |
| South | 400 | 8% |
| East | 600 | 20% |
| West | 2,000 | 11% |
- Sales: North 160, South 32, East 120, West 220. Total: 532 sales from 4,000 leads, so 532 ÷ 4,000 = 13.3%.
- Only regions below 13.3% raise the rate when dropped: South (8%) and West (11%).
- Without South: (532 − 32) ÷ (4,000 − 400) = 500 ÷ 3,600 ≈ 13.9%.
- Without West: (532 − 220) ÷ (4,000 − 2,000) = 312 ÷ 2,000 = 15.6%.
- Dropping West raises the rate the most, to 15.6% (up 2.3 points, against about 0.6 for South).
- Check: North, South and East have 160 + 32 + 120 = 312 sales from 1,000 + 400 + 600 = 2,000 leads ✓.
The trap answer is South, the region with the lowest rate. It is small, so it moves the total little. West is only a little below average, but it holds half the leads.
Check yourself. In the depot example, a fourth depot with 1,000 parcels joins. What on-time rate would it need for the four depots together to reach 80%?
Check your answer
80% of 5,000 parcels = 4,000 on time. The three depots have 3,100, so the new one needs 900 of its 1,000: 90%.
How hard questions disguise it
For harder questions (L4–L5)
At L4 and L5 the weights are hidden. The group sizes come as shares in a second chart or as a part-to-whole split, or the question gives the combined rate and asks for a group's size, which is a weighted average run backwards. A combined percent change across groups is weighted by each group's starting value, not averaged. Some questions show every group's rate rising while the combined rate falls, because the mix of group sizes moved toward the low-rate group. And when no sizes are given at all, the answer is often that the combined rate can't be determined. Turn every rate back into a count, or a share of the total, before you combine anything.
Practice
Practice weighted average questions in timed sets on the weighted averages practice page.
Common traps
Averaging rates without weights
50% of 100 hectares and 75% of 190 hectares combine to 66.4%, not 62.5%.
Wrong or missing weights
Every group counts, in proportion to its size: multiply each rate by its own count.