AP Calculus BC is the only AP course where the logistic differential equation dP/dt = kP(M − P) is assessed directly, and the logistic model is the cleanest example of a separable ordinary differential equation on the free-response section. Candidates reading this who are still in the AB–BC decision window should know that logistic questions almost never appear on the AB exam; the logistic family lives in Unit 7 of the BC course framework and tends to show up as one of the 45-minute FRQ prompts in the second half of the paper, or as a two-part MCQ stem that connects a verbal description to a differential equation. This article walks through how the rubric scores each row of a logistic FRQ, the four question shapes the College Board recycles, and the preparation strategy that converts a logistic stem from a guessed-answer gamble into a scored row.
The logistic differential equation and the row-by-row rubric logic
Every logistic FRQ on AP Calculus BC begins with a differential equation of the form dP/dt = kP(M − P), where P represents a population, k is the relative growth constant, and M is the carrying capacity. The first scored row almost always asks the candidate to do one of three things: identify the equilibrium values of P, write the differential equation in a particular form, or state what the long-run behaviour of P must be. The rubric is unforgiving on the equilibrium row because it requires two values, both stated explicitly. Candidates who write only P = 0 lose the second half of the row; the stable equilibrium is the non-zero value P = M, and the unstable equilibrium is P = 0, and the rubric language has long treated the two as paired answers rather than alternatives.
The second scored row in a typical logistic FRQ is the separable-ODE setup. The candidate is expected to write ∫ dP/[P(M − P)] = ∫ k dt, or an equivalent form such as ∫ 1/[P(M − P)] dP = ∫ k dt. The partial-fraction decomposition row, which is itself a separate scoring conversation on other AP Calculus FRQs, often appears inside the logistic stem as a one-step sub-skill. A candidate who writes the partial fractions as 1/[P(M − P)] = (1/M)(1/P) + (1/M)(1/(M − P)) earns the decomposition row, and the constants of integration cancel against each other when the two sides are added. For candidates who struggle with the partial-fraction setup, a substitution u = M − P turns the right-hand integral into a negative copy of the left, and the resulting equation solves without partial fractions at all; the rubric accepts either path as long as the antiderivative is correct.
The third row is the closed-form solution, and the rubric language is more specific than candidates expect. The logistic solution is conventionally written as P(t) = M / (1 + Ae^(−kMt)), where A is a constant determined by the initial condition. A candidate who writes the answer as P(t) = M / (1 + Ce^(−kMt)) scores the same row; the variable name on the constant does not matter, and the rubric does not deduct for using c, C, A, or any other letter. What the rubric does check is whether the exponent contains the product kM, not just k. Forgetting the factor of M on the exponent is one of the two most common logistic errors on the BC exam, and it costs a full row even when the rest of the solution is structurally correct.
The fourth row is the initial-condition row, where the candidate plugs P(0) = P₀ into the closed form and solves for the constant. The algebra reduces to A = (M − P₀) / P₀, and the rubric awards the row for any equivalent expression. Candidates who simplify to A = M/P₀ − 1 also score the row. The fifth row is the long-run behaviour, which on the AP Calculus BC exam is almost always phrased as a limit question: lim(t→∞) P(t) = ?. The rubric expects M, with a brief justification that the denominator tends to infinity as t grows. A candidate who answers M without the justification still typically earns the row on the FRQ, but a multiple-choice version of the same question can flip the credit if the distractor list contains the half-life of the population, the inflection time, or the value of t at which P = M/2; the safe move is to write the limit and the one-sentence reason.
How the BC rubric treats the constant of integration on the logistic row
The +C row on a logistic FRQ behaves the same way as on every other AP Calculus FRQ: a single arbitrary constant appears, and the constant of integration is absorbed into the constant determined by the initial condition. The rubric does not award a separate row for the appearance of +C, and a candidate who writes two different arbitrary constants loses nothing as long as the final answer carries only one. This is the cleanest possible setting in which to drill the +C rule, because the logistic derivation produces two antiderivatives that get added together, and the +C values combine into a single constant before the closed form is even reached.
Four logistic FRQ shapes the College Board recycles
The first shape is the pure dP/dt = kP(M − P) stem with no context, where the candidate is given k and M as numbers and asked to find the explicit solution. The second shape is the population-in-context stem, where P represents a fish population, a bacterial culture, a city's population, or the spread of a piece of information, and the constants are extracted from a verbal paragraph. The third shape is the graph-matching stem, where the candidate is shown a sigmoidal curve and asked which differential equation produces it; the correct differential equation is the one whose equilibria are the lower and upper horizontal asymptotes. The fourth shape, and the one that has appeared most often in the last several administrations, is the multi-part stem that starts with a logistic differential equation and ends with a logistic regression question on the calculator-active section.
For the graph-matching stem, the scoring is unusually generous because the candidate is given visual information instead of algebraic information. The rubric awards the row for matching the upper asymptote to M, the lower asymptote to 0, the inflection point to M/2 in the y-direction, and the slope of the curve at the inflection point to kM²/4. A candidate who can read the asymptotes off the graph usually gets the differential equation on the first try, and the row is scored without any integration at all. The trap in the graph-matching stem is the distractor dP/dt = kP, which produces an exponential curve that starts in the same place and looks similar in the first quadrant, but which never flattens at M. The rubric treats the choice between exponential and logistic as a one-row decision, and the row is scored by recognising the horizontal asymptote.
For the population-in-context stem, the candidate has to read k and M out of prose, and the prose is written in deliberately ambiguous units. A line such as 'the population grows at a rate proportional to both the current population and the difference between the carrying capacity and the current population' is the verbatim rubric language for dP/dt = kP(M − P), and a candidate who writes the differential equation in that form scores the setup row. The next line of the stem usually provides a numerical value for k, and a separate line provides M. The candidate is expected to translate each verbal phrase into a numeric constant without losing the variables, and the row is scored on the translation rather than on the integration. For candidates who struggle with verbal-to-symbolic translation, the safest move is to write the differential equation in symbolic form first, then substitute the numerical values at the end of the row.
The calculator-active logistic regression stem
The BC exam allows calculator use on part of the paper, and logistic questions can appear on the calculator-active FRQ in the form of a logistic regression. The candidate is given a data set and asked to find the logistic curve of best fit, then use the curve to answer a follow-up question. The regression coefficients do not score a separate row; the row is scored on the answer the candidate extracts from the calculator, typically a carrying capacity, a half-population time, or a long-run population projection. The candidate who knows where to find the logistic regression on the calculator, and who can read the coefficient list in the form y = c / (1 + ae^(−bx)), scores the row in under a minute.
Scoring the equilibrium row: stable versus unstable
The equilibrium row is where candidates lose the most points on logistic FRQs, and the reason is the rubric's insistence on labelling each equilibrium as stable or unstable. The equilibrium P = 0 is unstable because any small positive perturbation grows away from zero, and the equilibrium P = M is stable because any small positive or negative perturbation returns to M. A candidate who writes the two equilibria without the labels loses half the row. A candidate who writes only one equilibrium loses the full row. The rubric language for the equilibrium row typically reads 'P = 0 and P = M; the first is unstable and the second is stable', and any equivalent phrasing that pairs each value with its classification scores the row.
The classification is usually justified by a one-line argument, and the rubric does not require a full phase-line analysis. The justification can be as short as 'P = 0 is unstable because dP/dt > 0 for any small positive P', and the rubric awards the justification row for the sign argument alone. Candidates who try to draw a phase line on the FRQ lose time and gain no additional credit; the rubric treats the sign of dP/dt on each side of the equilibrium as a complete justification, and the phase-line drawing is at most a visual aid. For BC candidates preparing for the FRQ section, drilling the sign-of-derivative argument on each side of the equilibrium is the most efficient way to lock in the row.
The third type of question that touches the equilibrium row is the 'classify the equilibrium' MCQ, where the candidate is given a phase portrait of a one-dimensional differential equation and asked whether the equilibrium is stable, unstable, or semi-stable. The logistic differential equation produces a phase portrait with one stable and one unstable equilibrium, and the arrows on the phase line point away from P = 0 and toward P = M. The MCQ answer is 'unstable at P = 0 and stable at P = M', and the distractor 'stable at P = 0 and unstable at P = M' is the most common wrong answer, catching candidates who confuse the direction of the arrows on a downward-opening parabola with the direction of the arrows on an upward-opening parabola.
Phase-line drawings versus sign-table arguments
On the BC exam, the candidate can score the equilibrium row with either a phase-line drawing or a sign-table argument, and the rubric accepts both. A sign-table argument lists the sign of dP/dt in each of the three intervals (−∞, 0), (0, M), and (M, ∞), and concludes that the arrows on a phase line point right in (0, M) and left in (M, ∞), which means P = M attracts trajectories from both sides. A phase-line drawing sketches the three intervals on a horizontal line and draws the arrows directly. The sign-table approach is faster on the FRQ because the candidate can write the table in three lines of work and conclude the row in under 90 seconds, leaving more time for the integration that follows.
Carrying capacity, inflection point, and the half-population time
The logistic curve has three geometric features that the BC exam tests directly: the lower asymptote at P = 0, the upper asymptote at P = M, and the inflection point at P = M/2. The inflection point is the time at which the population is growing fastest, and the value of dP/dt at the inflection point is kM²/4, which is the maximum growth rate of the population. A candidate who is asked for the time of fastest growth on a logistic FRQ is being asked for the time at which P = M/2, and the rubric scores the row for stating both the population value and the implicit reference to d²P/dt² = 0. The second derivative row is sometimes a separate scored row on the BC exam, and the candidate is expected to compute d²P/dt² from the differential equation rather than from the closed form, because the differential equation is what carries the model.
The half-population time is the time at which P = M/2, and the value of t at that point is ln(A) / (kM), where A is the constant determined by the initial condition. The rubric typically awards the row for any expression equivalent to that value, and a candidate who writes the answer as t = (1/(kM)) ln((M − P₀)/P₀) scores the row. The half-population time appears in two common FRQ contexts: the 'when does the population reach half of carrying capacity' question, and the 'what is the maximum growth rate of the population' question. Both questions reduce to the same algebra, and the rubric treats them as interchangeable rows.
The half-population time is also the median of the logistic distribution, and the connection to the normal distribution is a high-value piece of exam trivia. The BC exam does not test the connection directly, but a candidate who recognises that the logistic curve is the cumulative distribution function of the logistic distribution can answer follow-up questions about quantiles in a single line. The first quartile of the logistic distribution is at t = (1/(kM)) ln(3), the third quartile is at t = (1/(kM)) ln(1/3), and the interquartile range is (2/(kM)) ln(3). None of this is required for the AP exam, but it is the kind of background that turns a stuck FRQ into a scored answer in under two minutes.
Worked example: a logistic FRQ end-to-end
Consider a stem that reads: 'A population P(t) satisfies dP/dt = 0.04P(200 − P), with P(0) = 20. Find P(t), the long-run population, and the time at which the population is growing fastest.' The first row is the differential equation, given. The second row is the separation: ∫ dP/[P(200 − P)] = ∫ 0.04 dt. The partial-fraction decomposition gives 1/[P(200 − P)] = (1/200)(1/P) + (1/200)(1/(200 − P)). The antiderivative on the left is (1/200) ln|P| − (1/200) ln|200 − P|, which simplifies to (1/200) ln|P/(200 − P)|. Setting this equal to 0.04t + C and solving gives P/(200 − P) = Ae^(8t), and rearranging gives P(t) = 200A e^(8t) / (1 + A e^(8t)) = 200 / (1 + (1/A) e^(−8t)). The constant 1/A is the constant in the rubric's preferred form, and the initial condition P(0) = 20 gives 1/A = 9, so the closed form is P(t) = 200 / (1 + 9 e^(−8t)). The long-run population is 200, and the time of fastest growth is t = ln(9)/8.
Common pitfalls and how to avoid them
The first pitfall is the exponent trap. The closed-form exponent is −kMt, not −kt, and a candidate who writes e^(−kt) on the FRQ loses a full row even when the rest of the solution is structurally correct. The fix is to remember that the separable step produces a factor of M on the right-hand side, and that factor survives into the exponent. The second pitfall is the equilibrium label. The rubric expects the candidate to label P = 0 as unstable and P = M as stable, and a candidate who writes only the values without the labels loses half the row. The fix is to write the classification in the same line as the value, so the row reads 'P = 0 (unstable) and P = M (stable)'.
The third pitfall is the partial-fraction sign. The decomposition 1/[P(M − P)] = (1/M)(1/P) + (1/M)(1/(M − P)) has a positive sign on the second term, not a negative sign, and a candidate who flips the sign loses the decomposition row. The fix is to check the decomposition by combining the fractions: (1/M)[(M − P) + P] / [P(M − P)] = (1/M)(M) / [P(M − P)] = 1/[P(M − P)], which confirms the sign. The fourth pitfall is the initial-condition algebra. The constant A is determined by (M − P₀)/P₀, and a candidate who writes A = P₀/(M − P₀) has the reciprocal and loses the row. The fix is to plug t = 0 into the closed form and solve for A in one line, rather than trying to remember the formula.
The fifth pitfall is the limit row. The long-run population is M, but the rubric awards the row only if the candidate writes the limit notation. A candidate who writes 'the population approaches 200' without the limit symbol loses the row on a strict grader. The fix is to write lim(t→∞) P(t) = 200 on the FRQ, even if the candidate is sure the answer is just 200. The sixth pitfall is the integration by parts trap. The logistic differential equation is separable, not integrable by parts, and a candidate who tries to integrate dP/dt = kP(M − P) by parts is going to spend five minutes producing a wrong answer. The fix is to recognise the separable form on sight and to apply the partial-fraction decomposition without delay.
Time-budget advice for the logistic FRQ
The logistic FRQ is one of the longer prompts on the BC exam, and the time budget is tight. A candidate who can complete the separation in three minutes, the partial-fraction decomposition in two minutes, the antiderivative in three minutes, the initial-condition plug in two minutes, and the limit row in one minute has spent eleven minutes on a question that is worth nine points. The remaining four minutes of the fifteen-minute budget should be spent on the equilibrium row and the inflection-point row, both of which can be scored in two minutes each with the right preparation. The candidate who gets stuck on the partial-fraction decomposition should skip to the long-run limit, score the M row, and come back to the integration if time permits.
Preparation strategy: drilling the logistic family
The most efficient preparation strategy for the logistic family on the BC exam is to drill four question types: the pure separable logistic, the population-in-context logistic, the graph-matching logistic, and the calculator-active logistic regression. Each question type is worth roughly 25% of the logistic score, and a candidate who can score all four on the FRQ section is in strong shape for a 5. The pure separable logistic can be drilled using released FRQs from prior administrations, and the College Board has published several logistic stems over the last decade that are accessible through the AP Classroom question bank. The population-in-context logistic can be drilled by writing differential equations from verbal descriptions, and a candidate who can translate 'grows at a rate proportional to the population and the difference between the carrying capacity and the population' into dP/dt = kP(M − P) in under 60 seconds has the row locked in.
The graph-matching logistic can be drilled by drawing sigmoidal curves and asking which differential equation produces them. The candidate should be able to identify the upper asymptote, the lower asymptote, the inflection point, and the maximum growth rate from the graph, and to convert each feature into a parameter of the differential equation. The calculator-active logistic regression can be drilled on the graphing calculator itself, and the candidate should know how to enter a data set, run a logistic regression, and read off the carrying capacity and the half-population time from the regression output. A candidate who can complete all four drills in under three hours of practice is well-prepared for the logistic section of the BC exam.
For candidates who want a structured preparation programme, the BC course framework in Unit 7 of differential equations includes the logistic model as a sub-skill, and the framework's learning objectives specify the four question types above. The framework also specifies that the logistic model is one of two population models assessed on the BC exam, the other being the exponential model dP/dt = kP. A candidate who has drilled both models, and who can tell at a glance whether a given population is bounded or unbounded, is in a strong position for the FRQ section. The exponential model is shorter and faster to score, but it carries fewer points per row because the differential equation is simpler and the rubric is correspondingly shorter.
Comparison: logistic versus exponential on the BC exam
The table below summarises the differences between the logistic and exponential models as they appear on the BC exam.
| Feature | Exponential dP/dt = kP | Logistic dP/dt = kP(M − P) |
|---|---|---|
| Equilibria | P = 0 only | P = 0 (unstable) and P = M (stable) |
| Long-run population | 0 or ∞ | M |
| Closed form | P(t) = P₀ e^(kt) | P(t) = M / (1 + Ae^(−kMt)) |
| Maximum growth rate | Unbounded | kM²/4 at P = M/2 |
| Typical FRQ length | 3–4 parts | 4–5 parts |
| Row count on rubric | 5–6 | 7–9 |
Calculator strategy on the logistic regression row
The BC exam allows the graphing calculator on part of the paper, and the logistic regression row is one of the highest-leverage places to use it. The candidate who knows how to enter a data set, run a logistic regression, and read off the carrying capacity can score the row in under 90 seconds, leaving the remaining time for the integration that follows. The regression output on a TI-84 or equivalent calculator lists the logistic equation in the form y = c / (1 + a e^(bx)), and the carrying capacity is c, the half-population coefficient is a, and the relative growth rate is b. A candidate who can match the calculator's output to the rubric's preferred form scores the row without any algebra at all.
The trap on the calculator row is the variable mismatch. The calculator's logistic regression uses a, b, and c in a specific order, and the rubric's preferred form uses A, k, and M in a different order. The candidate who reads the calculator output as 'a = 0.04, b = 200, c = 8' instead of 'A = 0.04, M = 200, k = 8' has the right answer in the wrong order, and the rubric deducts a row for the mislabelling. The fix is to write the calculator output in symbolic form before plugging in the numbers, and to match each symbol to the rubric's symbol rather than to the calculator's symbol. The candidate who writes 'c corresponds to M' on the FRQ scores the row with no ambiguity.
The second trap on the calculator row is the regression-format choice. The graphing calculator offers logistic, exponential, linear, logarithmic, power, quadratic, cubic, and quartic regressions, and the candidate who picks the wrong regression type loses the row. The fix is to look at the scatterplot before running the regression: if the scatterplot is sigmoidal, the regression is logistic; if the scatterplot is linear, the regression is linear; if the scatterplot is curved with a single turning point, the regression is quadratic. A candidate who takes 30 seconds to inspect the scatterplot before running the regression avoids the trap entirely.
Practising the calculator row under timed conditions
The calculator row on the BC exam is worth one or two points, and the time budget is roughly three minutes. A candidate who can complete the data entry, the regression, and the readout in under 90 seconds has 90 seconds of slack, which can be used to check the answer or to score an additional row elsewhere on the FRQ. The drill for the calculator row is to enter a synthetic data set, run the logistic regression, and read off the carrying capacity and the half-population time, all under timed conditions. The candidate who can do this drill five times in a row without error is ready for the exam.
Reading the logistic stem in under four minutes
The first four minutes of a logistic FRQ are the most important, because they determine whether the candidate scores the setup rows and whether the integration that follows is on the right track. A candidate who spends the first four minutes reading the stem, identifying the differential equation, writing the equilibria, and stating the long-run behaviour has scored three or four rows without any integration at all, and the remaining time can be spent on the algebra. The first-minute task is to read the stem and identify the variables: which symbol represents the population, which symbol represents time, and which symbols represent the parameters. The second-minute task is to write the differential equation in symbolic form, using the verbal description to set up dP/dt = kP(M − P).
The third-minute task is to identify the equilibria and label them. The candidate should write 'P = 0 (unstable) and P = M (stable)' in a single line, and the row is scored. The fourth-minute task is to state the long-run population as a limit, and the row is scored. By the end of the fourth minute, the candidate has scored three or four rows, and the integration that follows is worth an additional three to five rows. The candidate who skips the four-minute setup and goes straight to the integration loses the easy rows and gains no time on the hard rows, because the integration takes the same amount of time regardless of whether the setup rows have been scored.
The four-minute read is the single highest-leverage technique on the logistic FRQ, and a candidate who has practised the technique on five or six prior stems can execute it on the exam without thinking. The drill is to read a logistic stem, identify the variables, write the differential equation, identify the equilibria, and state the long-run behaviour, all in under four minutes. The candidate who can do this drill in under four minutes on five consecutive stems is ready for the BC exam. The four-minute read is also a useful diagnostic: a candidate who cannot complete the read in four minutes is missing a piece of conceptual understanding, and the piece can be identified by looking at which step is taking too long.
Diagnostic questions for the four-minute read
If the variable identification step is slow, the candidate needs to drill the verbal-to-symbolic translation of population-model stems. If the differential-equation step is slow, the candidate needs to drill the recognition of the logistic family versus the exponential family. If the equilibrium step is slow, the candidate needs to drill the sign-of-derivative argument. If the limit step is slow, the candidate needs to drill the long-run behaviour of the closed form. The candidate who has identified the slow step can drill that step in isolation, and the four-minute read becomes faster with each practice session.
For a candidate targeting a 5 on the BC exam, the logistic family is one of the highest-return areas to drill, because the rubric is generous on the setup rows and the integration is standard. AP Courses' one-to-one AP Calculus BC programme analyses each student's logistic FRQ error patterns against the BC rubric and turns a 5 target into a concrete preparation plan built around the four question types in this article.