How to Determine If a Function Is Even or Odd: The f(−x) Test, Worked Examples, and Three Failure Cases

A function is even if f(−x) equals f(x) for every x in its domain, odd if f(−x) equals −f(x) for every x, and neither if either identity fails anywhere. The condition that matters most is domain symmetry: if x = 2 is allowed but x = −2 isn't, the function can't be even or odd, no matter how tidy the formula looks. Everything below follows from those three rules — the algebra, the shortcuts, and the places where a promising-looking function quietly disqualifies itself.

What Even and Odd Mean in Plain Terms

Start with the definition, because every shortcut later depends on it. A function is even when flipping the input's sign leaves the output untouched, so f(−3) = f(3), f(−10) = f(10), and so on for every permitted input. Graphically, that means mirror symmetry across the y-axis: fold the coordinate plane along the vertical axis and the curve lands on itself.

A function is odd when flipping the input's sign flips the output's sign as well, so f(−3) = −f(3). Rotate that graph 180° about the origin and it maps onto itself. The classic reference is f(x) = x³, where f(2) = 8 and f(−2) = −8, which satisfies the rule exactly.

A function is neither when at least one input breaks the rule. And there's a fourth category worth naming: functions that are both even and odd are exactly those with f(x) = 0 for every x in the domain. Proof takes two lines — if f(−x) = f(x) and f(−x) = −f(x) simultaneously, then f(x) = −f(x), so 2f(x) = 0.

Plain English first, notation second. "Even" describes how the output behaves when you negate the input, not anything about the exponents. The names come from power functions: x² is even, x³ is odd, and every even power belongs to the even family while every odd power belongs to the odd family. That naming coincidence causes a lot of the confusion later.

The Two-Rule Test: f(−x) = f(x) or f(−x) = −f(x)

The mechanical test has a fixed order, and doing it out of order is where most errors creep in. Substitute −x everywhere x appears, simplify completely, then ask a single question: does the result equal the original f(x), its exact negative, or neither?

Simplify before you judge. For f(x) = x⁴ + 2x² + 5, substitution gives (−x)⁴ + 2(−x)² + 5 = x⁴ + 2x² + 5, which is visibly the original. For f(x) = x³ − 4x, you get −x³ + 4x = −(x³ − 4x), the exact negative. For f(x) = x² + x, you get x² − x, which matches neither f(x) = x² + x nor −f(x) = −x² − x.

Test at a point before you test with algebra. If f(x) = x² + x, then f(1) = 2 and f(−1) = 0. Since 0 is neither 2 nor −2, the function is neither, and you've established that in under ten seconds without any simplification. A single numerical counterexample is a complete disproof; a single numerical match proves nothing.

Remember that −f(x) means negate the entire expression, not just the first term. The most common student error I see reported in tutoring contexts is flipping only the leading term's sign. For f(x) = x³ − 4x, −f(x) = −x³ + 4x, with both terms negated. Miss a term and you'll mislabel the function.

Building a Reference Table for Common Function Types

Recognizing whole families saves enormous time in exams and in calculus work, and the classification follows from exponent parity plus a short list of exceptions.

Notice the table has a dependency baked in: a polynomial is even or odd exactly when all its terms share a parity. Mixed degrees kill the symmetry unless the minor powers cancel each other, which does happen — see the cancellation case further down.

Constants deserve a separate mention. Every constant function f(x) = c is even, including c = 0, because negating the input changes nothing. Students writing "zero is neither" on a quiz are producing exactly the result the both-case proves wrong.

The Step-by-Step Procedure, With Domain Checked First

Use one fixed sequence every time, because drifting into substitution before checking the domain produces phantom answers on functions like 1/x.

1: Write down the domain. List every x value the function accepts. For 1/x, the domain is all reals except 0. For √(x − 2), it's x ≥ 2. 2: Test domain symmetry first. Ask whether x in the domain forces −x in the domain. For 1/x, yes: 2 → −2 and −2 → 2. For √(x − 2), no: 2 is allowed but −2 gives √(−4), which is undefined. 3: Substitute −x for x in the formula, using parentheses around every instance: f(−x) = (−x)³ − 4(−x), not −x³ − 4x. 4: Expand and simplify completely, combining like terms and resolving signs. 5: Compare to f(x) and to −f(x), writing −f(x) out explicitly so you don't guess at it. 6: Conclude with even, odd, neither, or both — and note whether the domain allowed the conclusion to exist at all.

Step 2 is where careless work fails. If you skip it, √(x − 2) looks like it fails the parity test at the point level, and you may write "neither" as though algebra decided it. The domain decided it. A function whose domain isn't symmetric about zero cannot be even or odd because the very statement f(−x) = f(x) references points outside the function's existence.

Run the procedure on a negative exponent to see it earn its keep. Take f(x) = 1/x². Domain: all reals except 0, symmetric about 0. Substitute: 1/(−x)² = 1/x², equal to f(x). Even. Now take g(x) = 1/x. Substitute: 1/(−x) = −1/x, exactly −g(x). Odd. Neither required any graph paper.

Worked Examples: One Even, One Odd, One Neither

Concrete arithmetic fixes the rules better than definitions, so work these three by hand before trusting shortcuts.

Even: f(x) = 3x⁴ − 2x² + 11. Substituting gives 3(−x)⁴ − 2(−x)² + 11 = 3x⁴ − 2x² + 11. Check numerics: f(2) = 48 − 8 + 11 = 51, and f(−2) = 48 − 8 + 11 = 51. Because even powers absorb the negation and each coefficient is untouched, this family of functions is always even.

Odd: g(x) = x⁵ − 6x³ + x. Substituting gives (−x)⁵ − 6(−x)³ + (−x) = −x⁵ + 6x³ − x = −(x⁵ − 6x³ + x). Check: g(2) = 32 − 48 + 2 = −14, and g(−2) = −32 + 48 − 2 = 14, which is −(−14). Every sign flips, as required.

Neither: h(x) = x² + 3x. Substituting gives x² − 3x, which isn't h(x) and isn't −x² − 3x. Check at a point: h(2) = 10, h(−2) = −2, and −h(2) = −10. Since −2 matches neither 10 nor −10, neither it is.

Now check the endpoints of a restricted domain, which the three examples above quietly dodge. Take f(x) = x² defined only on [0, 5]. Substituting is clean; x² is even everywhere. But 2 is in the domain and −2 isn't, so f(−2) is undefined and the equation f(−2) = f(2) never gets to be true. Restricted to symmetric intervals like [−5, 5], x² is even again. The formula didn't change; the domain did.

Three Failure Cases That Break Careless Answers

The definitions are simple. These three scenarios are where otherwise-correct procedures go wrong.

Failure 1: The unsymmetric domain you didn't check. Take f(x) = x² on [0, 5] from above, or g(x) = x³ on [1, 10]. Algebra says "even" and "odd." The domain says neither test even applies. Most textbook exercises work on all real numbers, which trains you to skip step 2 until an exam or an applied problem punishes the habit.

Failure 2: The function that looks odd but is neither. Consider f(x) = sin x + cos x. The textbook favorites sin and cos are respectively odd and even, so mixing them might seem to preserve oddness. It doesn't. At x = π/4, f = √2 ≈ 1.414. At x = −π/4, f = −√2 + √2, which is 0. And −f(π/4) = −1.414. Three different numbers. Odd-looking and odd are different things.

Failure 3: The cancellation that rescues a mixed polynomial. Here's the one that surprises people. Take f(x) = (x² − 4)/(x² − 4) on its domain, which excludes x = ±2. Everywhere it's defined, f(x) = 1. A constant is even. But write it as x²/x² − 4/4 and the parity looks mixed, and the domain, all reals except ±2, is symmetric. The simplification reveals f(x) = 1 for all permitted x, so f is even. Simplifying before testing isn't stylistic; it changes answers. Similarly, (x³ + x²)/(x + 1) simplifies via factoring to x² for x ≠ −1, making it even on its symmetric domain despite the misleading numerator.

How Graphs and Symmetry Shortcuts Speed Up Checks

The visual test is fast and usually confirms what the algebra predicted, and a plotted check catches sign errors the algebra might have hidden.

Y-axis mirror: If the left half of the curve is the right half reflected across the y-axis, suspect even. Parabolas, absolute value, and cosine all pass. Origin rotation: If rotating the graph half a turn about (0, 0) lands it on itself, suspect odd. Cubics of the form a x³ + bx, sine, and the identity line all pass. Point check: Pull one x value from each quadrant, compare outputs, and treat a mismatch as decisive. A match of one pair settles nothing. Zero on the graph: Odd functions pass through (0, 0) if zero is in the domain; even functions need not. If the curve crosses the y-axis above or below the origin, it can't be odd. Piecewise folds: If the graph has a vertical jump held in place, the mirror or rotation may fail even though each piece looks symmetric on its own.

The table shortcut earns its speed with a caveat. A polynomial of mixed degrees is neither — unless the mixed terms cancel through factoring. Confirm before you trust the shortcut on anything but clean sums of like-parity terms.

There's also a composability rule worth memorizing: products and compositions of even and odd functions follow behavior you can predict. Even × even = even, odd × odd = even, even × odd = odd, and composing any function with an even one gives an even function. Take f(x) = x³ times g(x) = cos x. Odd × even = odd, and the test confirms it: (−x)³·cos(−x) = −x³·cos x.

What Symmetry Does for Calculus and Integrals

Non-experts tend to think of symmetry as a graphing convenience. It's actually a computation shortcut. Integrals over symmetric intervals collapse when symmetry cooperates: for an odd function f, the integral from −a to a of f(x) dx equals 0, because the positive and negative areas cancel exactly. For an even function, that integral equals 2 times the integral from 0 to a. Roughly half the volume of standard calculus technique practice relies on these two identities, and both are direct consequences of the algebra.

Use the shortcut when the interval is symmetric and the integrand's parity is already established. A common exam trap: an odd integrand on an interval like [1, 4] that isn't symmetric. The odd parity doesn't help. The integral there is not zero unless the integrand happens to have a symmetric antiderivative line up nicely; here it generally doesn't, and the shortcut fails silently.

Derivatives also respect parity with a swap: the derivative of an even function is odd, and the derivative of an odd function is even. Check with f(x) = x² (even) and f'(x) = 2x (odd). That cross-over explains why the pattern holds across polynomials: the derivative of x⁴ is 4x³, even to odd.

Taylor series exploit the same structure. A function with a purely even series — cos x = 1 − x²/2! + x⁴/4! − … — is even. One with a purely odd series — sin x = x − x³/3! + x⁵/5! − … — is odd. If you know the function's series, you've already proven its parity.

Pitfalls, Ambiguous Notations, and the Zero Function

The most-missed cases cluster around four ideas, and each of them breaks a tactic that works everywhere else.

The zero function. Both even and odd, because f(−x) = 0 = f(x) and also 0 = −0. It's the only both-case on the reals. Domain-bounded formulas. A formula plus a restricted domain is a different function; the domain decides first. Vertical shifts. Adding a constant breaks oddness unless the constant is inside the shift. f(x) = x³ + 1 isn't odd because it fails at x = 0: 1 vs. −1. Absolute value of x. Even despite the piecewise writing. The concern is behavior, not the notation style. Fractional exponents. f(x) = x^(1/3) is odd on all reals (the real cube root takes negatives), while x^(1/2) doesn't accept negative inputs and has an unsymmetric domain. Pair the test with the domain.

Check the FAQ for the question that catches even careful students: nothing in the definition requires the function to be a polynomial, and nothing requires it to be continuous or defined at every point. Parity is a property of outputs at matched input pairs. If every permitted pair satisfies the identity, the label applies.

Checking a Function in About Thirty Seconds

Most readers want the fast route after reading the theory, so here it is compressed without the theory that usually rides along.

1: Glance at the domain. Nonsymmetric? Stop; neither. 2: Test one x value. Mismatch on either identity? Stop; neither. 3: Substitute −x algebraically and simplify. 4: Compare to f(x) and to the fully negated form. 5: Sanity-check against the table or a sketch if time permits.

The shortcut catches the fast failures and skips the algebra on them; anything that passes needs the full simplification. Try it on f(x) = x/(x² + 1), a function common in calculus exercises. Domain is all reals (denominator never zero). At x = 2, f = 2/5 = 0.4; at x = −2, f = −2/5 = −0.4. Matches −f(2). Algebra: –x/(x² + 1) = −(x/(x² + 1)). Odd, in under thirty seconds.

FAQ

Does every function have to be either even or odd?

No. Most functions are neither, and "neither" is a fully valid conclusion in exactly the sense that a number being neither positive nor negative is valid. The both-case reduces to f(x) = 0 on the entire domain, proven by setting f(x) = −f(x). In a typical random collection of functions, neither is by far the most common outcome.

Can a function switch between even and odd on different parts of its domain?

Not for a single function over its whole domain. Parity is a global property, so a single failing input disqualifies it. A piecewise function can be even on one interval by coincidence, such as f(x) = x² for x < 0 and f(x) = x⁴ for x > 0, which happens to be even overall at x⁴'s pace on the positive side but still qualifies. The piecewise rule must hold for every matched pair, and "matched" means the pair connects through the full domain.

What's the difference between an even function and an even power?

An even function satisfies f(−x) = f(x); an even power means an exponent like 2 or 4. Functions with only even powers are usually even, but counterexamples exist: f(x) = x²/(1 + x) is neither despite containing only an even power in the numerator, because substituting −x gives x²/(1 − x), which fails both tests. Power parity is a clue, not a proof.

Is sin²x even or odd?

Even. At x = π/3, sin x = √3/2 so sin²x = 3/4. At x = −π/3, sin x = −√3/2 and squaring gives the same 3/4. The general proof: sin(−x) = −sin x, so squaring both sides gives sin²(−x) = sin²x. Composition with the even squaring function is what absorbs the sign.

Can a function be odd without passing through the origin?

Yes, if the origin isn't in its domain. f(x) = 1/x is odd and undefined at x = 0, so it has no point at (0, 0). If zero is in the domain and the function is odd, then the definition forces f(0) = 0: set x = 0 to get f(−0) = −f(0), so f(0) = −f(0), which is only possible when f(0) = 0. The common claim that odd functions pass through the origin is a consequence only when the domain includes zero.

How is this different from the even-odd property of integers?

The names are borrowed and nothing more. "Even function" and "odd function" don't mean the input or output is even; they mean the output behaves in a specific, sign-flipping way. A function like f(x) = x² returns both even and odd numbers as outputs, but is itself an even function. Don't cross-map the terminology.

What if f(−x) = f(x) for some x but fails for others?

Then the function is neither, unless the failing cases lie outside the domain. The definition is a for-all statement, so a single exception anywhere in the domain is fatal. Checking one x value is a quick disproof tool and never a proof tool.

Can I test parity without simplifying the formula?

At the precalculus level, simplifying first is safer and shows your reasoning. At the calculus level, an odd integrand over a symmetric interval means integral = 0 without further simplification. The one thing never safe is skipping the domain check, which simplifying can't fix.