You ever sit down to grade a 2017 AP Statistics exam and realize question 6 is the one that quietly separates the kids who get it from the ones who memorized formulas? Think about it: yeah. That last free-response item — the investigative task — is where the College Board throws something weird at you on purpose.
Here's the thing: if you're searching for 2017 AP Statistics free response answers question 6, you're probably a student, a teacher, or a parent who's suddenly very invested in regression trees. Or at least, you will be in a minute.
I've read through this question more times than I care to admit. And honestly, most answer keys online either skip the nuance or explain it like a textbook that's mad at you. So let's actually talk about it.
What Is the 2017 AP Stats Question 6
The 2017 AP Statistics free response question 6 is the investigative task from that year's exam. The data comes from a sample of U.Still, it's the one about predicting a county's median rent using things like population, poverty rate, and whether it's a "metro" or "non-metro" area. S. counties.
In plain language? Then they ask you to think hard about whether a linear model is even the right call. Day to day, they give you a scatterplot, some residual plots, and a regression output. And then — the twist — they ask you to compare two models and justify a choice.
The Setup They Gave
You got a dataset of 100 counties. The response variable was median monthly rent. Explanatory stuff included population (in thousands), poverty rate (percent), and a category for metro status.
The first model was a straight-up linear regression using all those predictors. The second was a model that included an interaction between metro status and population. That's the part that tripped a lot of people.
Why It's Called an Investigative Task
Unlike questions 1 through 5, question 6 isn't just "do the procedure." It's "here's a situation, now reason about it." The 2017 version tested whether you understand what a residual plot tells you, and whether you can spot when a model misses something important.
Why It Matters
Why does this question still get searched years later? In practice, not calculations. Because it's a near-perfect snapshot of what AP Stats actually wants from you. Reasoning.
Most students crush the math. They can write the regression equation. But when the College Board asks, "Explain why the residual plot suggests the linear model isn't great for non-metro counties," a lot of them freeze. They can find a slope. That's the gap.
And look — if you're a teacher building a unit on model comparison, this question is gold. Practically speaking, it shows kids that statistics isn't about one right number. It's about which story the data is telling, and whether your model listens.
In practice, the 2017 question 6 teaches you to check your assumptions before* you trust a prediction. Real talk: that habit is worth more than the exam score.
How It Works
Let's break down the actual question parts. The 2017 AP Statistics free response answers question 6 usually gets split into four or five subparts. Here's how the thinking goes.
Part (a): Describe the Relationship
They gave you a scatterplot of rent vs. Also, population, colored by metro status. You had to describe the association for metro and non-metro counties separately.
The short version is: for metro counties, rent tends to increase with population, with a positive linear-ish trend. For non-metro counties, the relationship is way weaker, maybe slightly positive, but basically flat and noisy.
Here's what most people miss — you had to say both* patterns, not just one. And you should've mentioned the clusters looked different. That's the whole setup for later.
Part (b): Interpret the Residual Plot
Next, a residual plot from the single linear model (all counties, no interaction). The residuals were plotted against predicted rent.
You were supposed to notice: for low predicted rents (mostly non-metro), residuals were scattered but with a weird spread. For high predicted rents (metro), things looked different. The plot showed the model systematically missed for one group.
Why does this matter? Because a good residual plot should look like random noise. This one didn't. It told you the straight line was forcing two different relationships into one.
Part (c): Compare the Two Models
Now the meat. Even so, model 1: rent = b0 + b1(pop) + b2(poverty) + b3(metro). Model 2 added b4(metropop).
The output showed the interaction term was significant. The residual standard error dropped. The residual plot for model 2 looked healthier.
Continue exploring with our guides on ap english literature and composition score calculator and how long is ap psychology exam.
So the answer isn't "model 2 is better because R-squared is bigger.Practically speaking, " It's: model 2 lets the slope for population depend on metro status, which matches what we saw in part (a). That's the reasoning the graders wanted.
Part (d): Justify a Prediction
Sometimes they'd ask you to predict rent for a specific county type. The key was using model 2 if you'd argued for it, and stating your assumption that the county fits the pattern in the data.
I know it sounds simple — but it's easy to miss that you must acknowledge the model is based on a sample of 100 counties. Extrapolation is a no-no.
Common Mistakes
This is the part most guides get wrong, because they list "errors" like "forgot the units.Consider this: " Sure. But the real misses are deeper.
One big one: students described the scatterplot as one relationship. Because of that, they wrote "as population goes up, rent goes up" and ignored the metro split. That lost points fast.
Another: in the residual part, they said "the model is bad" without saying why the plot showed it. "There's a pattern" isn't enough. You need "the spread differs by predicted value, suggesting non-constant variance or a missing predictor.
And the worst — picking model 2 just because the output looked more complicated. The 2017 AP Statistics free response answers question 6 rewards justification, not vibes. If you didn't connect the interaction to the visual split, you left points on the table.
Practical Tips
If you're prepping for this style of question — or teaching it — here's what actually works.
First, always read the visual before the numbers. The scatterplot in this question is the key. If you see two groups, say two groups. Don't average them in your head.
Second, when you interpret a residual plot, use this script: "A good model shows randomly scattered residuals. This one shows [specific thing], which means [specific problem]." That structure hits the rubric. The details matter here.
Third, for model comparison, tie it back. Don't say "model 2 is better." Say "model 2 includes an interaction that reflects the different slopes we saw earlier, and its residuals are more evenly spread.
And if you're a teacher? Give this question to students with the output hidden. Ask them what's weird. Just the scatterplot. You'll learn more in ten minutes than from a full review packet.
One more: don't cram the night before. The investigative task is a thinking exercise. You can't memorize your way through "explain why." You practice by reading real studies and poking at their models.
FAQ
What was the topic of 2017 AP Stats question 6? It was about using county data — population, poverty rate, metro status — to model median rent, and deciding whether a model with an interaction term fit better than a simple linear one.
Where can I find the official 2017 AP Statistics free response answers question 6? The College Board released the exam and scoring guidelines. Search their AP Statistics free-response archive for 2017. The scoring guide breaks down exactly what earned points.
Why is question 6 harder than the others? It's the investigative task. It combines everything — visuals, regression, model checking — but asks you to reason instead of just compute. That's the point.
How much is question 6 worth? It's 25% of the free-response section, which is half your exam score. So it's worth about 12.5% of your total AP Stats grade. Not nothing.
Do I need to know interaction terms for AP Stats? You won't
Do I need to know interaction terms for AP Stats?
You won't see them in every curriculum, but they’re fair game. The College Board expects you to recognize when a relationship between variables changes across groups — like how poverty rate affects rent differently in metro vs. non-metro areas. If your course didn’t cover interactions explicitly, you can still earn credit by describing the visual split and explaining that a model should reflect that difference. Justify your reasoning, and you’re golden.
Conclusion
Mastering AP Statistics investigative tasks comes down to one thing: thinking like a statistician, not just following steps. Start with visuals, ask targeted questions about patterns, and connect every model choice to evidence. Day to day, whether you’re a student drilling practice problems or a teacher designing lessons, focus on building the habit of explanation over guesswork. The exam rewards clarity and logic — so train yourself to see the story behind the data, not just the numbers on the screen.
