Mutually Exclusive Survey Questions: Fix Overlaps

A survey question is mutually exclusive when every respondent can truthfully pick only one of its answer options. If a respondent could defensibly choose two options, the options overlap, and every count, percentage and cross-tabulation built on that question is quietly wrong. This guide defines the term, separates it from “exhaustive” and “select all that apply”, walks through a computed audit of a faulty set of ranges, and gives checks you can run in Excel, R or Python.

What does mutually exclusive mean in a survey question?

Options are mutually exclusive when only one response option can be true for a single respondent. That is the definition given in the Encyclopedia of Survey Research Methods (Lavrakas, 2008), and Pew Research Center’s Writing Survey Questions guidance says the same thing as a design rule: the response categories should not overlap.

The encyclopedia’s own illustration is a commuting question with the options “less than 15 minutes”, “15 to 30 minutes”, “30 minutes to one hour” and “one hour or longer”. A respondent who commutes for 30 minutes could choose either the second or the third option, so the options are not mutually exclusive.

The test is simple: take any plausible answer and ask whether it belongs to exactly one option. An answer that fits two options is an overlap. An answer that fits none is a gap, which is a separate defect (see the next section).

Mutually exclusive vs exhaustive vs select-all-that-apply

Three properties are often blurred together. Pew’s guidance asks closed-ended questions to have a list of options that is exhaustive (it includes all reasonable responses) and whose options are mutually exclusive.

Property	Plain-language test	What breaks if it fails
Mutually exclusive	No answer fits two options	One respondent can be coded two ways, so category counts add up to more than the number of respondents
Exhaustive	Every plausible answer fits at least one option	Respondents with a valid answer have nowhere to go, so they skip the item, guess or pick the nearest option
Select all that apply	The respondent is invited to choose several options	Nothing, if designed on purpose: each option is then analysed as its own yes/no variable rather than as one category variable

A select-all question is therefore not a violation of mutual exclusivity; it is a different question type, covered in survey question types. The rule applies to single-answer items such as the closed-ended formats described in closed-ended questions in research and the ordered scales in ordinal survey question examples. Do not confuse the survey term with the probability term for events that cannot occur together; that sense is treated in independent vs mutually exclusive events.

Worked example: finding overlapping and missing ranges

This example is an illustrative composite, not real survey data. A questionnaire asks, “How many articles did you publish last year?” with three ranges: 0-5, 5-10 and 10-15. Ten respondents answer 0, 3, 5, 5, 7, 10, 10, 12, 15 and 16.

Place each answer. The 0 and 3 fit only 0-5. Each 5 fits both 0-5 and 5-10. The 7 fits only 5-10. Each 10 fits both 5-10 and 10-15. The 12 and 15 fit only 10-15. The 16 fits nothing.
Count by option. 0-5 holds 0, 3, 5, 5 = 4. 5-10 holds 5, 5, 7, 10, 10 = 5. 10-15 holds 10, 10, 12, 15 = 4.
Add them up: 4 + 5 + 4 = 13 placements for 10 respondents. Four respondents are double-placed (5, 5, 10, 10) and one is unplaced (16): 10 + 4 – 1 = 13.

Rewrite the options as 0-4, 5-9, 10-14 and 15 or more. Now 0 and 3 go to 0-4 (2), 5, 5 and 7 to 5-9 (3), 10, 10 and 12 to 10-14 (3), and 15 and 16 to 15 or more (2). That is 2 + 3 + 3 + 2 = 10 placements for 10 respondents, with no overlap and no gap.

The whole-number trap

Ranges such as 0-4 and 5-9 work only when answers are whole numbers. If the question measures something that can take decimal values, an answer of 4.5 fits neither option, which creates a gap. A convention that avoids this (our own rule of thumb, not a published standard) is to define each range as including its lower bound and excluding its upper bound: [0, 5), [5, 10), [10, 15). Every value then belongs to exactly one range. When respondents see the labels, use wording that matches, such as “5 up to but not including 10”, or restrict the answer to whole numbers.

Common mistakes: age, income and frequency bands

Overlap most often appears in banded numeric questions. A vendor guide to reviewing response options gives the age example “18 – 25” followed by “25 – 35”, which overlap at 25, and corrects it to 18-24, 25-34, 35-44, 45-54 and 55-64, with “Under 18” and “65 or older” at the ends of the list (Snap Surveys, vendor guidance).

Shared endpoints: “$25,000-$50,000” followed by “$50,000-$75,000”. The respondent at exactly $50,000 fits both.
Verbal bands that touch: the commuting example above, where “15 to 30 minutes” and “30 minutes to one hour” both contain 30 minutes.
Open ends missing: a top option such as “10 or more” is needed, or the highest respondents have no option.
Two ideas in one option: an option like “Agree because of cost” mixes a rating with a reason, which also makes the item double-barrelled; see double-barrelled questions.
Overlapping categories in a categorical list: options such as “Student” and “Graduate student” are both true for the same person. Merge them, or define one as a subset and say so in the question.

How to report response categories (APA)

Report the exact wording of the options in the method section or an appendix so readers can see that the categories were exclusive and exhaustive, and give the number of respondents (n) for each category together with the total. A suggested sentence (our wording, not a quoted rule): “Publication count was asked as a single-choice item with four ranges (0-4, 5-9, 10-14, 15 or more); n = 10.” For design decisions, you can cite Pew’s page in APA 7th edition webpage format with a retrieval date:

Pew Research Center. (n.d.). Writing survey questions. Retrieved October 29, 2026, from https://www.pewresearch.org/writing-survey-questions/

See questionnaire design for how option design fits into the wider instrument.

Check it in Excel, R or Python

These checks assume whole-number answers and ranges written as low and high values with inclusive endpoints, sorted by low value. They flag overlaps (the next range starts at or before the end of the current one) and gaps (the next range starts more than one unit after the end).

Excel

With lows in A2:A4 and highs in B2:B4, enter this in C2 and fill down one row less than the number of ranges, to compare each range with the next:

=IF(A3<=B2,"overlap",IF(A3>B2+1,"gap","ok"))

To count how many ranges a given answer fits (it should be exactly 1), put the answer in E2:

=COUNTIFS($A$2:$A$4,"<="&E2,$B$2:$B$4,">="&E2)

R

lows  <- c(0, 5, 10)
highs <- c(5, 10, 15)
n <- length(lows)
overlap <- lows[-1] <= highs[-n]
gap     <- lows[-1] >  highs[-n] + 1
x <- c(0, 3, 5, 5, 7, 10, 10, 12, 15, 16)
fits <- sapply(x, function(v) sum(lows <= v & v <= highs))
sum(fits)   # 13 placements for 10 respondents

Python

def audit(bins, values):
    bins = sorted(bins)
    overlaps = [(a, b) for a, b in zip(bins, bins[1:]) if b[0] <= a[1]]
    gaps = [(a[1] + 1, b[0] - 1) for a, b in zip(bins, bins[1:]) if b[0] > a[1] + 1]
    hits = [[i for i, (lo, hi) in enumerate(bins) if lo <= v <= hi] for v in values]
    return overlaps, gaps, hits

vals = [0, 3, 5, 5, 7, 10, 10, 12, 15, 16]
o, g, h = audit([(0, 5), (5, 10), (10, 15)], vals)
print(len(o), sum(map(len, h)))   # 2 overlaps, 13 placements

Running the Python version on the corrected bins (0, 4), (5, 9), (10, 14), (15, 99) returns no overlaps, no gaps and 10 placements for 10 respondents.

FAQ

Are all survey questions mutually exclusive?

No. Single-answer closed-ended questions should be. Select-all-that-apply questions deliberately let respondents choose several options, and they are analysed as separate yes/no items.

What is the difference between mutually exclusive and exhaustive?

Mutually exclusive means no answer fits two options. Exhaustive means every plausible answer fits at least one option. A good single-answer list is both, which is why “Other” or “Don’t know” options and open-ended top and bottom bands are common.

How do you fix overlapping answer choices?

Find every value that belongs to two options, then redraw the boundaries so that each value has one home: 0-4 and 5-9 instead of 0-5 and 5-10 for whole numbers, or half-open ranges for measured values. Re-run the placement count and confirm it equals the number of respondents.

Does overlap matter if the answer is only an estimate?

It can still weaken the data. A respondent whose true answer sits on a shared boundary has two valid options, so you cannot tell later which one that answer was meant to be. Clean boundaries cost nothing at design time.

Sources

Checked 2026-10-29.

Pew Research Center. Writing Survey Questions. https://www.pewresearch.org/writing-survey-questions/
Lavrakas, P. J. (Ed.). (2008). Mutually exclusive. In Encyclopedia of Survey Research Methods. Sage. https://doi.org/10.4135/9781412963947.n312
Snap Surveys. Carefully review survey response options (vendor guidance, used for the age-band example). https://www.snapsurveys.com/blog/carefully-review-survey-response-options/

Mutually Exclusive Survey Questions: Definition, Examples and How to Fix Overlaps