Worksheets · AP Statistics

AP Statistics formula sheet

Every formula a 2027 AP Statistics student uses, unit by unit, each with a worked example and the calculator command. You get an official formula sheet and tables on the exam; this page explains those formulas and shows how to use them.

AP Statistics Formula Sheet You get an official formula sheet and tables on the exam. This one shows how to use each formula, with a worked example.

Sigma Prep

Describing one variableUnit 1

Mean
xˉ=∑xin\displaystyle \bar{x} = \frac{\sum x_i}{n}Example: for 4, 7, 9, 12, xˉ=4+7+9+124=324=8\displaystyle \bar{x} = \frac{4 + 7 + 9 + 12}{4} = \frac{32}{4} = 8. Calculator: 1-Var Stats.
Median
The middle value of the sorted data; with an even count, the mean of the two middle values.Example: for 3, 5, 8, 10, the median is 5+82=6.5\displaystyle \frac{5 + 8}{2} = 6.5.
Standard deviation (sample)
sx=∑(xi−xˉ)2n−1\displaystyle s_x = \sqrt{\frac{\sum (x_i - \bar{x})^2}{n - 1}}Example: for 4, 7, 9, 12 (mean 8), the squared deviations are 16, 1, 1 and 16, so sx=343≈3.37\displaystyle s_x = \sqrt{\frac{34}{3}} \approx 3.37.
Interquartile range
IQR=Q3−Q1\displaystyle \text{IQR} = Q_3 - Q_1Example: Q1=12\displaystyle Q_1 = 12 and Q3=20\displaystyle Q_3 = 20 give IQR=20−12=8\displaystyle \text{IQR} = 20 - 12 = 8.
Outliers: 1.5 × IQR rule
Below Q1−1.5⋅IQR\displaystyle Q_1 - 1.5 \cdot \text{IQR} or above Q3+1.5⋅IQR\displaystyle Q_3 + 1.5 \cdot \text{IQR}Example: with Q1=12\displaystyle Q_1 = 12, Q3=20\displaystyle Q_3 = 20 and IQR=8\displaystyle \text{IQR} = 8, the fences are 12−1.5(8)=0\displaystyle 12 - 1.5(8) = 0 and 20+1.5(8)=32\displaystyle 20 + 1.5(8) = 32, so a value of 35 is an outlier.
Outliers: 2 SD rule
Outside xˉ±2sx\displaystyle \bar{x} \pm 2s_xA value more than 2 standard deviations from the mean.Example: with xˉ=50\displaystyle \bar{x} = 50 and sx=6\displaystyle s_x = 6, values below 50−2(6)=38\displaystyle 50 - 2(6) = 38 or above 50+2(6)=62\displaystyle 50 + 2(6) = 62 are outliers.
z‑score
z=x−μσ\displaystyle z = \frac{x - \mu}{\sigma} · z=x−xˉsx\displaystyle z = \frac{x - \bar{x}}{s_x}How many standard deviations a value is from the mean; the second form is for a sample.Example: a score of 88 in a class with mean 76 and standard deviation 8 has z=88−768=1.5\displaystyle z = \frac{88 - 76}{8} = 1.5, so it is 1.5 standard deviations above the mean.

Changing units (linear transformation)Unit 1

If every value x\displaystyle x is multiplied by b\displaystyle b and then has a\displaystyle a added, each new value is y=a+bx\displaystyle y = a + bx.

The new values
y=a+bx\displaystyle y = a + bxExample: Celsius to Fahrenheit multiplies by 1.8 and adds 32, F=32+1.8C\displaystyle F = 32 + 1.8C, so 20 ∘C\displaystyle 20\,^\circ\text{C} becomes 32+1.8(20)=68 ∘F\displaystyle 32 + 1.8(20) = 68\,^\circ\text{F}.
Center and position
yˉ=a+bxˉ\displaystyle \bar{y} = a + b\bar{x}The median, quartiles and percentiles change the same way: multiply by b\displaystyle b, then add a\displaystyle a.Example: a mean temperature of 20 ∘C\displaystyle 20\,^\circ\text{C} becomes 32+1.8(20)=68 ∘F\displaystyle 32 + 1.8(20) = 68\,^\circ\text{F}, and a median of 15 ∘C\displaystyle 15\,^\circ\text{C} becomes 32+1.8(15)=59 ∘F\displaystyle 32 + 1.8(15) = 59\,^\circ\text{F}.
Spread
sy=∣b∣ sx\displaystyle s_y = |b|\, s_x · IQRy=∣b∣ IQRx\displaystyle \text{IQR}_y = |b|\, \text{IQR}_xThe standard deviation, IQR and range are multiplied by ∣b∣\displaystyle |b|. Adding a\displaystyle a does not change them.Example: a standard deviation of 5 ∘C\displaystyle 5\,^\circ\text{C} becomes 1.8(5)=9 ∘F\displaystyle 1.8(5) = 9\,^\circ\text{F}; adding 32 does not change it.

Two‑way tables and probabilityUnit 2

Conditional relative frequency
count in the celltotal of the given row or column\displaystyle \frac{\text{count in the cell}}{\text{total of the given row or column}}Example: of the 45 plants of one variety, 8 did not fruit: 845≈0.178\displaystyle \frac{8}{45} \approx 0.178.
Equally likely outcomes
P(E)=outcomes in Eall outcomes\displaystyle P(E) = \frac{\text{outcomes in } E}{\text{all outcomes}}Example: an even number on a fair die: P(even)=36=0.5\displaystyle P(\text{even}) = \frac{3}{6} = 0.5.
Complement
P(not A)=1−P(A)\displaystyle P(\text{not } A) = 1 - P(A) · P(at least one)=1−P(none)\displaystyle P(\text{at least one}) = 1 - P(\text{none})Example: 4 independent serves, each in with probability 0.7: P(at least one in)=1−(0.3)4=0.9919\displaystyle P(\text{at least one in}) = 1 - (0.3)^4 = 0.9919.
Addition rule
P(A∪B)=P(A)+P(B)−P(A∩B)\displaystyle P(A \cup B) = P(A) + P(B) - P(A \cap B)Example: P(A)=0.5\displaystyle P(A) = 0.5, P(B)=0.4\displaystyle P(B) = 0.4 and P(A∩B)=0.2\displaystyle P(A \cap B) = 0.2 give P(A∪B)=0.5+0.4−0.2=0.7\displaystyle P(A \cup B) = 0.5 + 0.4 - 0.2 = 0.7.
Mutually exclusive events
P(A∩B)=0\displaystyle P(A \cap B) = 0 · P(A∪B)=P(A)+P(B)\displaystyle P(A \cup B) = P(A) + P(B)The events cannot happen together.Example: a bus cannot be both early and late. With P(early)=0.25\displaystyle P(\text{early}) = 0.25 and P(late)=0.35\displaystyle P(\text{late}) = 0.35, P(early or late)=0.25+0.35=0.6\displaystyle P(\text{early or late}) = 0.25 + 0.35 = 0.6.
Conditional probability
P(A∣B)=P(A∩B)P(B)\displaystyle P(A \mid B) = \frac{P(A \cap B)}{P(B)}Example: P(A∩B)=0.12\displaystyle P(A \cap B) = 0.12 and P(B)=0.40\displaystyle P(B) = 0.40 give P(A∣B)=0.120.40=0.30\displaystyle P(A \mid B) = \frac{0.12}{0.40} = 0.30.
Multiplication rule
P(A∩B)=P(A)⋅P(B∣A)\displaystyle P(A \cap B) = P(A) \cdot P(B \mid A)Example: 60% of orders are online and 25% of online orders are returned: P(online and returned)=0.6(0.25)=0.15\displaystyle P(\text{online and returned}) = 0.6(0.25) = 0.15.
Independent events
P(A∣B)=P(A)\displaystyle P(A \mid B) = P(A) · P(A∩B)=P(A)⋅P(B)\displaystyle P(A \cap B) = P(A) \cdot P(B)Knowing that B\displaystyle B happened does not change the chance of A\displaystyle A. Then the probabilities multiply.Example: two independent serves, each in with probability 0.7: P(both in)=0.7(0.7)=0.49\displaystyle P(\text{both in}) = 0.7(0.7) = 0.49.

Random variables and the binomial distributionUnit 2

Mean (expected value)
μX=E(X)=∑xi P(xi)\displaystyle \mu_X = E(X) = \sum x_i \, P(x_i)Example: X=0,1,2\displaystyle X = 0, 1, 2 with probabilities 0.5, 0.3, 0.2: μX=0(0.5)+1(0.3)+2(0.2)=0.7\displaystyle \mu_X = 0(0.5) + 1(0.3) + 2(0.2) = 0.7.
Standard deviation of X
σX=∑(xi−μX)2 P(xi)\displaystyle \sigma_X = \sqrt{\sum (x_i - \mu_X)^2 \, P(x_i)}Example: for the same X\displaystyle X, the squared deviations are 0.49, 0.09 and 1.69, so σX=0.49(0.5)+0.09(0.3)+1.69(0.2)=0.61≈0.781\displaystyle \sigma_X = \sqrt{0.49(0.5) + 0.09(0.3) + 1.69(0.2)} = \sqrt{0.61} \approx 0.781.
Binomial setting
Two outcomes, a fixed number of trials n\displaystyle n, independent trials, and the same probability of success p\displaystyle p on each.Example: the number of sixes in 10 rolls of a fair die is binomial with n=10\displaystyle n = 10 and p=16\displaystyle p = \frac{1}{6}.
Binomial probability
P(X=x)=(nx)px(1−p)n−x\displaystyle P(X = x) = \binom{n}{x} p^x (1 - p)^{n - x}Example: with n=5\displaystyle n = 5 and p=0.3\displaystyle p = 0.3, P(X=2)=10(0.3)2(0.7)3=0.3087\displaystyle P(X = 2) = 10(0.3)^2(0.7)^3 = 0.3087. Calculator: binompdf(5,0.3,2)\displaystyle \text{binompdf}(5, 0.3, 2).
Binomial: at most, at least
P(X≤k)=binomcdf(n,p,k)\displaystyle P(X \le k) = \text{binomcdf}(n, p, k) · P(X≥k)=1−binomcdf(n,p,k−1)\displaystyle P(X \ge k) = 1 - \text{binomcdf}(n, p, k - 1)Example: with n=21\displaystyle n = 21 and p=0.40\displaystyle p = 0.40, P(X≥9)=1−binomcdf(21,0.40,8)=0.4763\displaystyle P(X \ge 9) = 1 - \text{binomcdf}(21, 0.40, 8) = 0.4763.
Binomial mean and SD
μX=np\displaystyle \mu_X = np · σX=np(1−p)\displaystyle \sigma_X = \sqrt{np(1 - p)}Example: a 70% shooter takes 77 free throws: μX=77(0.70)=53.9\displaystyle \mu_X = 77(0.70) = 53.9 made and σX=77(0.7)(0.3)≈4.02\displaystyle \sigma_X = \sqrt{77(0.7)(0.3)} \approx 4.02.

The normal distributionUnit 2

Empirical rule (68‑95‑99.7)
About 68%, 95% and 99.7% of values lie within 1, 2 and 3 standard deviations of the mean.Example: with μ=100\displaystyle \mu = 100 and σ=15\displaystyle \sigma = 15, about 95% of values are between 100−2(15)=70\displaystyle 100 - 2(15) = 70 and 100+2(15)=130\displaystyle 100 + 2(15) = 130.
Normal probability
z=x−μσ\displaystyle z = \frac{x - \mu}{\sigma} · P(a<X<b)=normalcdf(a,b,μ,σ)\displaystyle P(a < X < b) = \text{normalcdf}(a, b, \mu, \sigma)Standardize and use the table, or find the area with the calculator.Example: heights with μ=64\displaystyle \mu = 64 and σ=2.5\displaystyle \sigma = 2.5 inches: z=68−642.5=1.6\displaystyle z = \frac{68 - 64}{2.5} = 1.6, and P(X>68)=normalcdf(68,1E99,64,2.5)=0.0548\displaystyle P(X > 68) = \text{normalcdf}(68, \text{1E99}, 64, 2.5) = 0.0548.
Inverse normal
z=invNorm(area to the left)\displaystyle z = \text{invNorm}(\text{area to the left}) · x=μ+zσ\displaystyle x = \mu + z\sigmaA value (a percentile or cutoff) from an area.Example: the 90th percentile when μ=500\displaystyle \mu = 500 and σ=100\displaystyle \sigma = 100: z=invNorm(0.90)=1.2816\displaystyle z = \text{invNorm}(0.90) = 1.2816, so x=500+1.2816(100)≈628.2\displaystyle x = 500 + 1.2816(100) \approx 628.2.

Sampling distributionsUnits 2–4

Sample proportion
μp^=p\displaystyle \mu_{\hat{p}} = p · σp^=p(1−p)n\displaystyle \sigma_{\hat{p}} = \sqrt{\frac{p(1 - p)}{n}}Example: p=0.4\displaystyle p = 0.4 and n=150\displaystyle n = 150 give σp^=0.4(0.6)150=0.04\displaystyle \sigma_{\hat{p}} = \sqrt{\frac{0.4(0.6)}{150}} = 0.04.
Large counts condition
np≥10\displaystyle np \ge 10 and n(1−p)≥10\displaystyle n(1 - p) \ge 10Then the sampling distribution of p^\displaystyle \hat{p} is approximately normal.Example: 150(0.4)=60\displaystyle 150(0.4) = 60 and 150(0.6)=90\displaystyle 150(0.6) = 90, both at least 10.
10% condition
n≤0.10N\displaystyle n \le 0.10NWhen sampling without replacement, the sample is at most 10% of the population.Example: a sample of 150 from a school of 4,000 students: 150≤0.10(4000)=400\displaystyle 150 \le 0.10(4000) = 400.
Sample mean
μxˉ=μ\displaystyle \mu_{\bar{x}} = \mu · σxˉ=σn\displaystyle \sigma_{\bar{x}} = \frac{\sigma}{\sqrt{n}}Example: σ=12\displaystyle \sigma = 12 and n=36\displaystyle n = 36 give σxˉ=1236=2\displaystyle \sigma_{\bar{x}} = \frac{12}{\sqrt{36}} = 2.
Central limit theorem
xˉ\displaystyle \bar{x} is approximately normal if the population is normal, or if n≥30\displaystyle n \ge 30 (more when the population is very skewed).Example: a right-skewed population and a sample of 40: 40≥30\displaystyle 40 \ge 30, so xˉ\displaystyle \bar{x} is approximately normal.
Difference of proportions
μp^1−p^2=p1−p2\displaystyle \mu_{\hat{p}_1 - \hat{p}_2} = p_1 - p_2 · σp^1−p^2=p1(1−p1)n1+p2(1−p2)n2\displaystyle \sigma_{\hat{p}_1 - \hat{p}_2} = \sqrt{\frac{p_1(1 - p_1)}{n_1} + \frac{p_2(1 - p_2)}{n_2}}Example: p1=0.5\displaystyle p_1 = 0.5, n1=100\displaystyle n_1 = 100, p2=0.4\displaystyle p_2 = 0.4, n2=120\displaystyle n_2 = 120: σp^1−p^2=0.5(0.5)100+0.4(0.6)120≈0.0671\displaystyle \sigma_{\hat{p}_1 - \hat{p}_2} = \sqrt{\frac{0.5(0.5)}{100} + \frac{0.4(0.6)}{120}} \approx 0.0671.
Difference of means
μxˉ1−xˉ2=μ1−μ2\displaystyle \mu_{\bar{x}_1 - \bar{x}_2} = \mu_1 - \mu_2 · σxˉ1−xˉ2=σ12n1+σ22n2\displaystyle \sigma_{\bar{x}_1 - \bar{x}_2} = \sqrt{\frac{\sigma_1^2}{n_1} + \frac{\sigma_2^2}{n_2}}Example: σ1=10\displaystyle \sigma_1 = 10, n1=25\displaystyle n_1 = 25, σ2=12\displaystyle \sigma_2 = 12, n2=36\displaystyle n_2 = 36: σxˉ1−xˉ2=10225+12236=8≈2.83\displaystyle \sigma_{\bar{x}_1 - \bar{x}_2} = \sqrt{\frac{10^2}{25} + \frac{12^2}{36}} = \sqrt{8} \approx 2.83.

Confidence intervalsUnits 3–4

General form
statistic±(critical value)(standard error)\displaystyle \text{statistic} \pm (\text{critical value})(\text{standard error})Example: a sample proportion of 0.62 with a standard error of 0.03, at 95% confidence: 0.62±1.96(0.03)=0.62±0.059\displaystyle 0.62 \pm 1.96(0.03) = 0.62 \pm 0.059.
Critical value z*
z∗=invNorm(1+C2)\displaystyle z^* = \text{invNorm}\left(\frac{1 + C}{2}\right)z∗=1.645\displaystyle z^* = 1.645 for 90%, 1.960\displaystyle 1.960 for 95% and 2.576\displaystyle 2.576 for 99% confidence.Example: at 95% confidence, z∗=invNorm(0.975)=1.960\displaystyle z^* = \text{invNorm}(0.975) = 1.960.
One‑proportion z‑interval
p^±z∗p^(1−p^)n\displaystyle \hat{p} \pm z^* \sqrt{\frac{\hat{p}(1 - \hat{p})}{n}}Example: 120 of 300 gives p^=0.40\displaystyle \hat{p} = 0.40 and SE=0.4(0.6)300=0.0283\displaystyle \text{SE} = \sqrt{\frac{0.4(0.6)}{300}} = 0.0283, so 0.40±1.960(0.0283)=0.40±0.055\displaystyle 0.40 \pm 1.960(0.0283) = 0.40 \pm 0.055, from 0.345 to 0.455. Calculator: 1-PropZInt.
Sample size for a margin
n=(z∗)2 p^(1−p^)(ME)2\displaystyle n = \frac{(z^*)^2 \, \hat{p}(1 - \hat{p})}{(\text{ME})^2}Round up. With no estimate of p\displaystyle p, use p^=0.5\displaystyle \hat{p} = 0.5.Example: a margin of 0.03 at 95%: n=1.962(0.5)(0.5)0.032=1067.1\displaystyle n = \frac{1.96^2 (0.5)(0.5)}{0.03^2} = 1067.1, so 1,068 people.
Two‑proportion z‑interval
(p^1−p^2)±z∗\displaystyle (\hat{p}_1 - \hat{p}_2) \pm z^* p^1(1−p^1)n1+p^2(1−p^2)n2\displaystyle \sqrt{\frac{\hat{p}_1(1 - \hat{p}_1)}{n_1} + \frac{\hat{p}_2(1 - \hat{p}_2)}{n_2}}Example: p^1=0.55\displaystyle \hat{p}_1 = 0.55 and p^2=0.45\displaystyle \hat{p}_2 = 0.45, each from 200: SE=0.0497\displaystyle \text{SE} = 0.0497, so 0.10±1.960(0.0497)=0.10±0.098\displaystyle 0.10 \pm 1.960(0.0497) = 0.10 \pm 0.098, from 0.002 to 0.198. Calculator: 2-PropZInt.
Margin of error and n
Multiplying n\displaystyle n by k\displaystyle k divides the margin of error by k\displaystyle \sqrt{k}; a higher confidence level makes it wider.Example: 4 times the sample size halves a margin of 0.06 to 0.064=0.03\displaystyle \frac{0.06}{\sqrt{4}} = 0.03.
Critical value t*
t∗=invT(1+C2,df)\displaystyle t^* = \text{invT}\left(\frac{1 + C}{2}, \text{df}\right) · df=n−1\displaystyle \text{df} = n - 1Example: 95% with n=16\displaystyle n = 16: t∗=invT(0.975,15)=2.131\displaystyle t^* = \text{invT}(0.975, 15) = 2.131.
One‑sample t‑interval
xˉ±t∗sn\displaystyle \bar{x} \pm t^* \frac{s}{\sqrt{n}} · df=n−1\displaystyle \text{df} = n - 1Example: n=16\displaystyle n = 16, xˉ=52.3\displaystyle \bar{x} = 52.3, s=4.8\displaystyle s = 4.8: 52.3±2.131⋅4.816=52.3±2.56\displaystyle 52.3 \pm 2.131 \cdot \frac{4.8}{\sqrt{16}} = 52.3 \pm 2.56, from 49.74 to 54.86. Calculator: TInterval.
Paired t‑interval
xˉd±t∗sdn\displaystyle \bar{x}_d \pm t^* \frac{s_d}{\sqrt{n}} · df=n−1\displaystyle \text{df} = n - 1Work with the differences; n\displaystyle n is the number of pairs.Example: 10 pairs with xˉd=3.2\displaystyle \bar{x}_d = 3.2 and sd=2.5\displaystyle s_d = 2.5, so t∗=invT(0.975,9)=2.262\displaystyle t^* = \text{invT}(0.975, 9) = 2.262: 3.2±2.262⋅2.510=3.2±1.79\displaystyle 3.2 \pm 2.262 \cdot \frac{2.5}{\sqrt{10}} = 3.2 \pm 1.79, from 1.41 to 4.99. Calculator: TInterval on the differences.
Two‑sample t‑interval
(xˉ1−xˉ2)±t∗s12n1+s22n2\displaystyle (\bar{x}_1 - \bar{x}_2) \pm t^* \sqrt{\frac{s_1^2}{n_1} + \frac{s_2^2}{n_2}}Degrees of freedom from technology.Example: xˉ1=78\displaystyle \bar{x}_1 = 78, s1=8\displaystyle s_1 = 8, n1=25\displaystyle n_1 = 25 and xˉ2=72\displaystyle \bar{x}_2 = 72, s2=10\displaystyle s_2 = 10, n2=30\displaystyle n_2 = 30: SE=6425+10030=2.43\displaystyle \text{SE} = \sqrt{\frac{64}{25} + \frac{100}{30}} = 2.43, df=52.9\displaystyle \text{df} = 52.9 and t∗=2.006\displaystyle t^* = 2.006, so 6±4.87\displaystyle 6 \pm 4.87, from 1.13 to 10.87. Calculator: 2-SampTInt.

Significance testsUnits 3–4

Test statistic
statistic−null valuestandard error of the statistic\displaystyle \frac{\text{statistic} - \text{null value}}{\text{standard error of the statistic}}Example: p^=0.575\displaystyle \hat{p} = 0.575, a null value of 0.5 and a standard error of 0.025 give 0.575−0.50.025=3.0\displaystyle \frac{0.575 - 0.5}{0.025} = 3.0.
One‑proportion z‑test
z=p^−p0p0(1−p0)n\displaystyle z = \frac{\hat{p} - p_0}{\sqrt{\frac{p_0(1 - p_0)}{n}}}Example: H0 ⁣:p=0.5\displaystyle H_0\!: p = 0.5 and Ha ⁣:p>0.5\displaystyle H_a\!: p > 0.5, with 230 of 400: z=0.575−0.50.025=3.00\displaystyle z = \frac{0.575 - 0.5}{0.025} = 3.00 and the P-value is 0.0013. Calculator: 1-PropZTest.
Pooled (combined) proportion
p^c=x1+x2n1+n2\displaystyle \hat{p}_c = \frac{x_1 + x_2}{n_1 + n_2}The successes in both groups over both sample sizes.Example: 60 of 150 and 45 of 150: p^c=60+45150+150=105300=0.35\displaystyle \hat{p}_c = \frac{60 + 45}{150 + 150} = \frac{105}{300} = 0.35.
Two‑proportion z‑test
z=p^1−p^2p^c(1−p^c)(1n1+1n2)\displaystyle z = \frac{\hat{p}_1 - \hat{p}_2}{\sqrt{\hat{p}_c(1 - \hat{p}_c)\left(\frac{1}{n_1} + \frac{1}{n_2}\right)}}Example: p^1=0.40\displaystyle \hat{p}_1 = 0.40 and p^2=0.30\displaystyle \hat{p}_2 = 0.30, each from 150, with p^c=0.35\displaystyle \hat{p}_c = 0.35: z=0.100.0551=1.82\displaystyle z = \frac{0.10}{0.0551} = 1.82, and the two-sided P-value is 0.069. Calculator: 2-PropZTest.
One‑sample t‑test
t=xˉ−μ0sn\displaystyle t = \frac{\bar{x} - \mu_0}{\frac{s}{\sqrt{n}}} · df=n−1\displaystyle \text{df} = n - 1Example: Ha ⁣:μ>50\displaystyle H_a\!: \mu > 50 with n=16\displaystyle n = 16, xˉ=52.3\displaystyle \bar{x} = 52.3, s=4.8\displaystyle s = 4.8: t=2.31.2=1.92\displaystyle t = \frac{2.3}{1.2} = 1.92 with 15 df, and the P-value is 0.037. Calculator: T-Test.
Paired t‑test
t=xˉd−0sdn\displaystyle t = \frac{\bar{x}_d - 0}{\frac{s_d}{\sqrt{n}}} · df=n−1\displaystyle \text{df} = n - 1Work with the differences; n\displaystyle n is the number of pairs.Example: 10 pairs with xˉd=3.2\displaystyle \bar{x}_d = 3.2 and sd=2.5\displaystyle s_d = 2.5, Ha ⁣:μd>0\displaystyle H_a\!: \mu_d > 0: t=3.20.791=4.05\displaystyle t = \frac{3.2}{0.791} = 4.05 with 9 df, and the P-value is 0.0014. Calculator: T-Test on the differences.
Two‑sample t‑test
t=xˉ1−xˉ2s12n1+s22n2\displaystyle t = \frac{\bar{x}_1 - \bar{x}_2}{\sqrt{\frac{s_1^2}{n_1} + \frac{s_2^2}{n_2}}}Degrees of freedom from technology.Example: xˉ1=78\displaystyle \bar{x}_1 = 78, s1=8\displaystyle s_1 = 8, n1=25\displaystyle n_1 = 25 and xˉ2=72\displaystyle \bar{x}_2 = 72, s2=10\displaystyle s_2 = 10, n2=30\displaystyle n_2 = 30, two-sided: t=62.43=2.47\displaystyle t = \frac{6}{2.43} = 2.47, and the P-value is 0.017. Calculator: 2-SampTTest.
Decision
P-value≤α\displaystyle \text{P-value} \le \alpha: reject H0\displaystyle H_0Otherwise, fail to reject H0\displaystyle H_0.Example: a P-value of 0.037 at α=0.05\displaystyle \alpha = 0.05: 0.037≤0.05\displaystyle 0.037 \le 0.05, so reject H0\displaystyle H_0. There is convincing evidence that μ>50\displaystyle \mu > 50.
Type I error
P(Type I error)=α\displaystyle P(\text{Type I error}) = \alphaThe chance of rejecting H0\displaystyle H_0 when H0\displaystyle H_0 is true.Example: at α=0.05\displaystyle \alpha = 0.05, a true null hypothesis is rejected 5% of the time.
Power
power=1−P(Type II error)\displaystyle \text{power} = 1 - P(\text{Type II error})The chance of rejecting H0\displaystyle H_0 when a particular alternative is true. It grows with n\displaystyle n, with α\displaystyle \alpha, and with the distance from the null value.Example: P(Type II error)=0.20\displaystyle P(\text{Type II error}) = 0.20 gives power =1−0.20=0.80\displaystyle = 1 - 0.20 = 0.80.

Chi‑square tests for two‑way tablesUnit 3

Expected count
expected=(row total)(column total)table total\displaystyle \text{expected} = \frac{(\text{row total})(\text{column total})}{\text{table total}}Example: a row total of 60, a column total of 50 and a table total of 200: 60(50)200=15\displaystyle \frac{60(50)}{200} = 15.
Chi‑square statistic
χ2=∑(observed−expected)2expected\displaystyle \chi^2 = \sum \frac{(\text{observed} - \text{expected})^2}{\text{expected}}Add over every cell of the table.Example: a 2-by-2 table with counts 30, 20, 20, 30 and every expected count 25: χ2=4⋅(30−25)225=4.0\displaystyle \chi^2 = 4 \cdot \frac{(30 - 25)^2}{25} = 4.0.
Degrees of freedom
df=(rows−1)(columns−1)\displaystyle \text{df} = (\text{rows} - 1)(\text{columns} - 1)Example: a 3-by-4 table has df=(3−1)(4−1)=6\displaystyle \text{df} = (3 - 1)(4 - 1) = 6. A 2-by-2 table has df=1\displaystyle \text{df} = 1, and χ2=4.0\displaystyle \chi^2 = 4.0 gives a P-value of 0.0455. Calculator: χ2\displaystyle \chi^2-Test.
Conditions
Random samples or random assignment, n≤0.10N\displaystyle n \le 0.10N when sampling without replacement, and every expected count greater than 5.Example: the smallest expected count in the table is 15, which is greater than 5.
Homogeneity or independence
Homogeneity: one variable in two or more populations or treatments. Independence: two variables in one population.Example: favorite sport in random samples of 9th and 12th graders is homogeneity; sport and grade in one random sample of the school is independence.

RegressionUnit 5

Least‑squares line
y^=a+bx\displaystyle \hat{y} = a + bxFind a\displaystyle a and b\displaystyle b with technology.Example: y^=12.5+3.2x\displaystyle \hat{y} = 12.5 + 3.2x at x=10\displaystyle x = 10 gives y^=12.5+3.2(10)=44.5\displaystyle \hat{y} = 12.5 + 3.2(10) = 44.5. Calculator: LinReg(a+bx).
Residual
residual=y−y^\displaystyle \text{residual} = y - \hat{y}Observed minus predicted.Example: an observed y=47\displaystyle y = 47 at x=10\displaystyle x = 10 has residual 47−44.5=2.5\displaystyle 47 - 44.5 = 2.5: the line underpredicts it by 2.5.
Through the means
yˉ=a+bxˉ\displaystyle \bar{y} = a + b\bar{x}The line passes through the point (xˉ,yˉ)\displaystyle (\bar{x}, \bar{y}).Example: with xˉ=8\displaystyle \bar{x} = 8, yˉ=12.5+3.2(8)=38.1\displaystyle \bar{y} = 12.5 + 3.2(8) = 38.1.
Interpreting the slope
b\displaystyle b is the predicted change in y\displaystyle y for each increase of 1 in x\displaystyle x.Example: b=3.2\displaystyle b = 3.2 points per hour of study: each extra hour predicts a score 3.2 points higher.
Correlation
−1≤r≤1\displaystyle -1 \le r \le 1The sign gives the direction and the size the strength of a linear association; r\displaystyle r has no units.Example: r=−0.8\displaystyle r = -0.8 is a strong, negative linear association.
Coefficient of determination
r2\displaystyle r^2The fraction of the variation in y\displaystyle y explained by the linear model with x\displaystyle x.Example: r=−0.8\displaystyle r = -0.8 gives r2=0.64\displaystyle r^2 = 0.64: 64% of the variation in y\displaystyle y is explained by the line.

Calculator commands are the TI-84 names; Desmos in Bluebook and other graphing calculators have the same functions.

© 2026 Sigma Prep · sigmaprep.ioFree to print and share

Is there a formula sheet on the AP Statistics exam?

Yes. You get College Board’s formula sheet and tables on both sections, so the work is knowing which formula fits, checking its conditions and saying what the result means.

  • The official sheet gives formulas, not choices. It does not say whether a question needs a one-proportion z-test or a two-sample t-interval; the procedure is yours to name.
  • Conditions earn points. On the free response, random sampling or assignment, the 10% condition and large counts (or the normal condition for means) are checked in context before any number.
  • The calculator does the arithmetic. normalcdf, invNorm, the interval and test menus and LinReg give the numbers, but the answer still states the formula with values and a conclusion in context.

Questions about the AP Statistics formula sheet

Yes. Every AP Statistics student gets College Board’s formula sheet and statistical tables (standard normal, t and chi-square) for both sections of the exam, and a graphing calculator is allowed throughout. The official sheet lists the formulas; it does not say when to use each one, which conditions to check, or how to interpret the result. This page does that, with a worked example for every formula.

No. This is Sigma Prep’s own sheet: our own order and wording, a worked example under every formula, and the calculator command. The official formula sheet and tables are printed in College Board’s course and exam description, and you get them on exam day.

Rules and conditions you are expected to know rather than look up: the 1.5 × IQR and 2-standard-deviation outlier rules, the empirical rule (68%, 95%, 99.7%), the large-counts, 10% and normal conditions, the expected-count condition for chi-square, how a change of units moves the center and spread, and how sample size changes a margin of error. All of them are on this sheet with an example.

A graphing calculator with statistical functions, allowed on both sections. On the digital exam the Desmos calculator in Bluebook is also available. The commands on this sheet are the TI-84 names (normalcdf, invNorm, binomcdf, invT, 1-PropZTest, 2-SampTTest, LinReg); other graphing calculators have the same functions. On the free response, name the procedure and show the formula with your values as well as the calculator result.

The geometric distribution, the mean and standard deviation of a sum or difference of random variables, the chi-square goodness-of-fit test and inference for the slope of a regression line are no longer in AP Statistics, and the correlation and slope are now found with technology rather than by formula. None of them is on this sheet.

Yes. Download the PDF or print it straight from the page. There is no account and no sign-up, and you can hand out as many copies as you like. To practice the formulas, the free unit practice tests have exam-style multiple choice and FRQs with answer keys and scoring guides.

AP is a registered trademark of the College Board, which was not involved in making this sheet and does not endorse it. Written by Sigma Prep; it does not reproduce the official AP Statistics formula sheet.