Topic 5.2 The Normal Distribution, Revisited: revisit the normal model and z-scores in the context of distributions of statistics, finding proportions and using the standard normal as the basis for later inference.

What this topic is asking

The College Board (Topic 5.2) wants you to revisit the normal model and z-scores, now applied to distributions of statistics, finding proportions and percentiles with the standard normal so that you are ready to use it on sampling distributions and in inference.

The normal model, applied to statistics

The mechanics are identical to Topic 1.10; what changes is the object being described. There, the normal model described a population of measurements (heights, scores). Here it describes the distribution of a statistic over all samples, so the "mean" and "standard deviation" in the z-score are the mean and standard deviation of the sampling distribution, not of raw data. Recognizing that the same standardizing step works for a statistic is the whole purpose of revisiting it.

Proportions and percentiles

These are the same forward-and-backward moves as Topic 1.10: value to z to area, and area to z to value. The reason to drill them again is that the critical z-values that appear here, $1.645$, $1.96$, $2.576$, are exactly the ones that define $90\%$, $95\%$, and $99\%$ confidence intervals later, so building fluency now pays off immediately in Unit 6.

Why this refresher sits in Unit 5

It might seem odd to revisit Topic 1.10 mid-course, but the placement is deliberate. The next topics (the central limit theorem, and the sampling distributions of $\hat{p}$ and $\bar{x}$) all conclude that, under the right conditions, a statistic is approximately normal with a specific mean and standard deviation. The moment you have that, every probability question about the statistic, "how likely is a sample proportion above $0.5$?", "what value does the sample mean exceed only $5\%$ of the time?", becomes a normal-model area or percentile calculation, exactly the skills revisited here. So Topic 5.2 is the toolkit you pick up just before you need it. It also re-establishes the discipline of interpreting in context: a z-score of $2$ should be read as "this value is $2$ standard deviations above the center of the distribution, so it is in the upper few percent," not left as a bare number. Carrying that interpretive habit into sampling distributions keeps your later answers about how unusual a statistic is grounded and full-credit.

When the normal model is appropriate

As in Topic 1.10, the normal model is only valid when the distribution is approximately normal, and the upcoming topics give the precise conditions under which a sampling distribution qualifies (the large-counts condition for proportions, and the central limit theorem or a normal population for means). Until those conditions are checked, you should not assume a statistic is normal. The honest workflow, established here and used throughout inference, is: confirm (via a stated condition) that the sampling distribution is approximately normal, identify its mean and standard deviation, then standardize and use the standard normal to find the area or percentile you need. This sequence, check normality, get the parameters, standardize, read the area, is the spine of every calculation in the rest of the course, which is why Topic 5.2 makes sure the normal-model machinery is sharp before the sampling-distribution topics deploy it.

Try this

Q1. A statistic is approximately normal with mean $100$, sd $8$. Find the z-score of $112$ and the proportion above it. [2 points]

• Cue. $z = (112 - 100)/8 = 1.5$; $P(Z > 1.5) \approx 1 - 0.9332 = 0.0668$, about $6.7\%$.

Q2. What z-score marks the $97.5$th percentile, and why is it useful later? [1 point]

• Cue. $z \approx 1.96$; it is the critical value for a $95\%$ confidence interval, used throughout Unit 6 onward.