AP Statistics confidence intervals cheat sheet

By Jude Wallis · Published

A confidence interval is statistic plus or minus margin of error, where the margin of error is a critical value times a standard error. Proportions use a z critical value and means use a t critical value. The margin of error shrinks with a larger sample and grows with a higher confidence level.

Confidence intervals appear in Unit 3 (proportions) and Unit 4 (means) of the Fall 2026 AP Statistics course. The course has no confidence interval for a regression slope, which was removed in the redesign.

The general form

statistic±(critical value)(standard error)\text{statistic} \pm (\text{critical value})(\text{standard error})
PieceMeaning
Statistic (point estimate)The sample value that estimates the parameter, such as p^\hat{p} or xˉ\bar{x}
Standard errorThe estimated standard deviation of the statistic
Critical valuez∗z^* or t∗t^*, set by the confidence level
Margin of error(critical value)(standard error)(\text{critical value})(\text{standard error})
WidthTwice the margin of error

Given only the endpoints, the point estimate is the midpoint and the margin of error is half the width. For the interval (0.294,0.386)(0.294, 0.386), the estimate is 0.294+0.3862=0.34\frac{0.294 + 0.386}{2} = 0.34 and the margin of error is 0.386−0.2942=0.046\frac{0.386 - 0.294}{2} = 0.046. See margin of error vs standard error.

Each interval in the course

ParameterIntervalCritical value
Proportion ppp^±z∗p^(1−p^)n\hat{p} \pm z^*\sqrt{\frac{\hat{p}(1-\hat{p})}{n}}z∗z^*
Difference in proportions p1−p2p_1 - p_2(p^1−p^2)±z∗p^1(1−p^1)n1+p^2(1−p^2)n2(\hat{p}_1 - \hat{p}_2) \pm z^*\sqrt{\frac{\hat{p}_1(1-\hat{p}_1)}{n_1} + \frac{\hat{p}_2(1-\hat{p}_2)}{n_2}}z∗z^*
Mean μ\muxˉ±t∗sn\bar{x} \pm t^*\frac{s}{\sqrt{n}}t∗t^* with df=n−1df = n - 1
Mean difference μd\mu_d (paired data)xˉd±t∗sdn\bar{x}_d \pm t^*\frac{s_d}{\sqrt{n}}t∗t^* with df=n−1df = n - 1, where nn is the number of pairs
Difference in means μ1−μ2\mu_1 - \mu_2(xˉ1−xˉ2)±t∗s12n1+s22n2(\bar{x}_1 - \bar{x}_2) \pm t^*\sqrt{\frac{s_1^2}{n_1} + \frac{s_2^2}{n_2}}t∗t^* with dfdf from technology

p^\hat{p} ("p-hat") is a sample proportion, xˉ\bar{x} ("x-bar") is a sample mean, and ss is a sample standard deviation. For a difference, state the order of subtraction. Before computing any interval, check its conditions, which are listed in the conditions for inference checklist. The steps are worked in how to calculate a confidence interval.

Critical values

Confidence levelz∗z^*
90%1.645
95%1.960
99%2.576

z∗z^* marks off the middle C%C\% of the standard normal distribution. For 95%, the area between −1.960-1.960 and 1.9601.960 is 0.95, which leaves 0.025 in each tail.

t∗t^* depends on both the confidence level and the degrees of freedom, so read it from the t-table or technology. For the same confidence level, t∗t^* is larger than z∗z^*, and it gets closer to z∗z^* as the degrees of freedom increase. With 95% confidence, t∗=2.262t^* = 2.262 for df=9df = 9 and t∗=2.045t^* = 2.045 for df=29df = 29.

Interpreting the interval and the confidence level

These are two different sentences.

What is being interpretedSentence pattern
The interval"We are [C]% confident that the interval from [low] to [high] captures the [parameter, in context, for the population]."
The confidence level"If many random samples of the same size were taken and a [C]% interval were built from each, about [C]% of those intervals would capture the [parameter in context]."
Statement to avoidWhat is wrong with it
"There is a 95% probability that the parameter is in this interval."The parameter is a fixed number. This one interval either contains it or does not. The 95% describes the method over many samples.
"95% of the data values are in the interval."The interval estimates a parameter. It does not describe individual values.
"95% of all sample means fall in this interval."The interval is centered at one sample's statistic, and other samples give other intervals.
"The sample proportion is in the interval."The statistic is always at the center. The interval is about the population parameter.

See what 95% confidence means and confidence level vs confidence interval.

What changes the margin of error

Change, with everything else held fixedMargin of error
Larger sample sizeSmaller
Sample size multiplied by 4Cut in half (exactly for z intervals, approximately for t intervals)
Higher confidence levelLarger
More variable data (larger ss)Larger
p^\hat{p} closer to 0.5Larger

For a proportion with p^=0.5\hat{p} = 0.5 and 95% confidence, the margin of error is 0.098 with n=100n = 100 and 0.049 with n=400n = 400. Keeping n=400n = 400, it is 0.041 at 90% confidence and 0.064 at 99% confidence.

The sample size needed for a margin of error MM when estimating a proportion is

n=(z∗M)2p^(1−p^)n = \left(\frac{z^*}{M}\right)^2 \hat{p}(1-\hat{p})

Use p^=0.5\hat{p} = 0.5 when no earlier estimate exists, and round up. For M=0.03M = 0.03 at 95% confidence, n=(1.960.03)2(0.25)=1067.1n = \left(\frac{1.96}{0.03}\right)^2(0.25) = 1067.1, so 1,068 are needed.

The margin of error covers sampling variability only. It does not account for bias from the way the data were collected. See why is my confidence interval so wide.

Using an interval to judge a claim

The values inside a confidence interval are the plausible values of the parameter at that confidence level.

SituationConclusion
The claimed value is inside the intervalThe claim is plausible. The data do not give convincing evidence against it.
The claimed value is outside the intervalThe data give convincing evidence that the parameter differs from the claimed value.
An interval for a difference contains 0No convincing evidence of a difference between the two parameters.
An interval for a difference is entirely positiveConvincing evidence that the first parameter is larger.
An interval for a difference is entirely negativeConvincing evidence that the first parameter is smaller.

A claimed value inside the interval is one of many plausible values, so the interval does not show the claim is true. See two-proportion confidence interval contains zero.

One-sample z-interval for a proportion

In a random sample of 400 adults from a large city, 136 say they walk to work. Construct and interpret a 95% confidence interval for the proportion of all adults in the city who walk to work.

  1. Conditions: the sample is random, 400 is less than 10% of the adults in a large city, and the observed counts 136 and 400−136=264400 - 136 = 264 are both at least 10.

  2. Point estimate: p^=136400=0.34\hat{p} = \frac{136}{400} = 0.34.

  3. Standard error: 0.34(0.66)400=0.000561=0.0237\sqrt{\frac{0.34(0.66)}{400}} = \sqrt{0.000561} = 0.0237.

  4. Margin of error: 1.96×0.0237=0.04641.96 \times 0.0237 = 0.0464.

  5. Interval: 0.34±0.04640.34 \pm 0.0464, which is (0.2936,0.3864)(0.2936, 0.3864).

The interval is about (0.294,0.386)(0.294, 0.386). We are 95% confident that the interval from 0.294 to 0.386 captures the proportion of all adults in the city who walk to work.

One-sample t-interval for a mean

A random sample of 25 batteries of one model has a mean life of 82.4 hours and a standard deviation of 6.5 hours. A boxplot of the 25 lifetimes shows no strong skewness and no outliers. Construct a 95% confidence interval for the mean life of all batteries of this model.

  1. Conditions: the sample is random, 25 batteries are less than 10% of all batteries of the model, and with n=25<30n = 25 < 30 the boxplot shows no strong skewness and no outliers.

  2. Degrees of freedom: 25−1=2425 - 1 = 24. The 95% critical value is t∗=2.064t^* = 2.064.

  3. Standard error: 6.525=6.55=1.3\frac{6.5}{\sqrt{25}} = \frac{6.5}{5} = 1.3.

  4. Margin of error: 2.064×1.3=2.6832.064 \times 1.3 = 2.683.

  5. Interval: 82.4±2.68382.4 \pm 2.683, which is (79.717,85.083)(79.717, 85.083).

The interval is about (79.72,85.08)(79.72, 85.08) hours. We are 95% confident that the interval from 79.72 to 85.08 hours captures the mean life of all batteries of this model.

Frequently asked questions

What does 95% confident mean?

It describes the method. If many random samples of the same size were taken and a 95% interval were built from each one, about 95% of those intervals would capture the parameter. For one computed interval, the parameter is either inside it or not.

How can I make a confidence interval narrower?

Take a larger sample or use a lower confidence level. A larger sample lowers the standard error. A lower confidence level lowers the critical value, at the cost of capturing the parameter less often.

When do I use a z critical value and when a t critical value?

Intervals for proportions use z∗z^*. Intervals for means use t∗t^*, because the standard error is built from the sample standard deviation ss. See z-interval vs t-interval.