Data Detective Agency

Every scatter plot is a crime scene: points cluster, trends emerge, outliers hide in plain sight. Learn to read the evidence, fit the line, question the model — and crack the case of the two-way table.

Grade 8 · Unit 7 8.SP.1 Scatter plots & association 8.SP.2 Fitting a line 8.SP.3 Using linear models 8.SP.4 Two-way tables
Part 1

The Evidence Board

Four open cases, four scatter plots. Read each one for direction, form, strength, and outliers. On the study-hours case, you're the analyst: slide your own trend line into place and watch the fit meter judge your work.

2
30
Part 2

Case Files

Five sealed case files. Lock in your verdict first — then the reveal opens the evidence.

Case File 1 of 5
Part 3

The Detective's Machines

Three machines in the crime lab: read any scatter plot, fit and use any line, and interrogate any two-way table.

Four checks, every time: direction, form, strength, outliers. Run them on each open case.

Fit informally, use precisely, interpret honestly — and know where the model stops working.

Categorical data can't scatter — it tabulates. Relative frequencies turn raw counts into verdicts.

Part 4

Detective's Exam

Ten questions to earn your badge. Wrong answers get a hint naming the exact machine or case that proves the right idea.

📦 Word Box

Scatter plot
A graph of paired data — one dot per case, positioned by its two measurements.
Positive association
As x rises, y tends to rise — the cloud climbs left to right.
Negative association
As x rises, y tends to fall — the cloud descends left to right.
No association
A shapeless cloud: knowing x tells you nothing about y.
Linear association
The cloud hugs a straight line.
Nonlinear association
The pattern curves — real, but no straight line captures it.
Cluster
A tight group of points — often a subgroup hiding inside the data.
Outlier
A point far from the overall pattern. Investigate it; don't automatically delete it.
Trend line (line of fit)
A straight line drawn through the heart of the cloud, balancing points above and below.
Slope of the model
The predicted change in y for each 1-unit change in x — the story's rate.
Intercept of the model
The predicted y when x = 0 — the story's starting value.
Two-way table
A grid counting cases by two categorical variables at once.
Relative frequency
A count turned into a fraction of its row or column — the fair way to compare groups of different sizes.
Extrapolation
Predicting far outside the data's range — where even a beautiful model may quietly stop being true.