AP Statistics, Unit 4
Exam words: Inference for Quantitative Data: Means
The 18 words a written answer in this unit needs. For each one: what it means, a wording that loses the mark and why, and a sentence that uses it the way a full answer would.
Study these words as a deck
Flashcards, Learn, Test and Match in Cards, with your Still learning marks kept, and the booklet to print on A4 or US Letter. Sign in to open it.
The words
population mean
The mean of a variable over the entire population, written mu; a fixed parameter.
Loses the mark: “the mean of the sample”. That is x-bar, a statistic; mu is the mean for the whole population.
Example: The population mean fill of all bottles from the machine is supposed to be 500 mL.
standard deviation of the sample mean
The typical distance of x-bar from mu: sigma over the square root of n.
Loses the mark: “sigma divided by n”. Divide by the square root of n, not by n.
Example: With sigma = 4 mL and n = 16 bottles, the standard deviation of the sample mean is 4 divided by the square root of 16, which is 1 mL.
Central Limit Theorem
For large n, the sampling distribution of x-bar is approximately Normal, whatever the population's shape.
Loses the mark: “the population becomes normal”. The theorem is about the sampling distribution of x-bar; the population keeps its shape.
Example: Although wait times are skewed right, the Central Limit Theorem says x-bar for samples of 40 waits is approximately Normal.
population standard deviation
The standard deviation of a variable over the whole population, written sigma; rarely known.
Loses the mark: “s is the population standard deviation”. s is the sample standard deviation, a statistic; sigma is the population parameter.
Example: Because the population standard deviation is unknown, we estimate it with s and use a t distribution.
t distribution
A symmetric, bell-shaped distribution centered at 0, with heavier tails than the standard Normal.
Loses the mark: “use z with s”. When s replaces sigma, the standardized mean follows a t distribution, not the standard Normal.
Example: Because sigma is unknown, the t distribution with 14 degrees of freedom is used for the mean battery life.
degrees of freedom
The number selecting which t or chi-square distribution applies; n - 1 for a one-sample t.
Loses the mark: “df = n for a one-sample t”. One-sample t procedures use n - 1 degrees of freedom.
Example: With 15 batteries tested, the degrees of freedom are 15 - 1 = 14.
heavier tails
More area far from the center than the standard Normal, so t* exceeds z*.
Loses the mark: “thinner tails than the normal”. The t distribution has more area in its tails, so its critical values are larger, not smaller.
Example: With only 5 degrees of freedom, the t distribution's heavier tails give t* = 2.571 for 95% confidence, compared with z* = 1.96.
t* critical value
The multiplier from a t distribution with the right degrees of freedom setting the margin.
Loses the mark: “use 1.96 for a mean”. 1.96 is a z* value; for a mean with sigma unknown use t* with n - 1 degrees of freedom.
Example: For 95% confidence with df = 14, the t* critical value is 2.145.
one-sample t interval for a mean
A confidence interval for mu: x-bar plus or minus t* times the standard error.
Loses the mark: “z interval for the mean”. With sigma unknown, use a t interval with n - 1 degrees of freedom.
Example: A 95% one-sample t interval for the mean battery life is 10.4 plus or minus 2.145(0.284), or 9.8 to 11.0 hours.
standard error of the mean
The estimated standard deviation of x-bar: s divided by the square root of n.
Loses the mark: “s divided by n”. Divide s by the square root of n.
Example: With s = 1.1 hours and n = 15, the standard error of the mean is 1.1 divided by the square root of 15, about 0.284 hours.
Normal/Large Sample condition
Check x-bar is approximately Normal: Normal population, large n, or no strong skew or outliers.
Loses the mark: “n is at least 30 so the population is normal”. A large n makes the sampling distribution of x-bar approximately Normal; the population keeps its shape.
Example: With n = 15, the Normal/Large Sample condition is met because a dotplot of the battery lives shows no strong skew or outliers.
one-sample t test for a mean
A test of H0: mu = mu0 using t and a p-value from the t distribution.
Loses the mark: “z test with s”. With sigma unknown, the statistic follows a t distribution, so use a t test.
Example: A one-sample t test of H0: mu = 500 mL against Ha: mu < 500 mL checks whether the machine underfills bottles.
t statistic
The test statistic for a mean: how many standard errors x-bar lies from mu0.
Loses the mark: “divide by the standard deviation to get t”. Divide by the standard error, s over the square root of n, not by s alone.
Example: With x-bar = 498.6, s = 3.2 and n = 20, the t statistic is (498.6 - 500) divided by (3.2 over the square root of 20), about -1.96.
hypothesized mean
The population mean stated in H0, written mu0, which the t statistic measures distance from.
Loses the mark: “H0: x-bar = 500”. Write H0 about mu, the population mean, not about the sample mean.
Example: The hypothesized mean is mu0 = 500 mL, the fill amount printed on the label.
paired data
Two measurements linked by subject, such as before and after, analyzed through their differences.
Loses the mark: “two-sample t test on before and after”. Before and after scores on the same subjects are paired; analyze the differences with a one-sample t procedure.
Example: Each runner's times in both shoe types are paired data, so we analyze the 12 differences.
mean difference
For paired data, the population mean of the differences within pairs, written mu_d.
Loses the mark: “mu1 - mu2 for paired data”. Paired data have one parameter, the mean of the differences, with df equal to the number of pairs minus 1.
Example: The mean difference in sprint time (old shoe minus new shoe) for the 12 runners is 0.31 seconds.
paired t test
A one-sample t procedure applied to the differences from paired data, with n - 1 df.
Loses the mark: “degrees of freedom from both groups”. A paired t test has df equal to the number of pairs minus 1.
Example: A paired t test on the 12 differences gives t = 2.8 with 11 degrees of freedom.
two-sample t test
A test or interval for mu1 - mu2 from two independent samples, with unpooled standard error.
Loses the mark: “two-sample t test on paired data”. Differences within pairs call for a paired t test; a two-sample test is for independent groups.
Example: A two-sample t test comparing independent groups of 30 drivers and 25 cyclists gives t = 2.1 and a p-value of 0.04.