SAT Statistics & Regression on Desmos (Straight From a Table)

Desmos on the Digital SAT — Part 7. One specific calculator skill, shown on the real Desmos screen. Part of the full series; each part builds on the last.

Turn a Data Table Into Answers

Some of the most time-consuming questions on the Digital SAT hand you a small table of numbers and ask for a mean, a median, a prediction, or a rate of change. You can grind those out by hand — but the built-in Desmos calculator (available on both adaptive Math modules in Bluebook, on every one of the 44 Math questions) will do the arithmetic the instant the data is typed in. Here is the exact screen you are aiming for:

Those five points — (1, 1), (2, 2), (3, 4), (4, 5), (5, 6) — are living in a Desmos table, and the straight line running through them is a best-fit line Desmos found on its own. Once you know the two commands below, reproducing this is a 30-second move that turns a wordy statistics question into a reading exercise.

Step 1: Get the Data Into a Table

Open the calculator, click the + in the top-left corner, and choose table. Desmos hands you two columns already named x_1 and y_1 — the subscripts are part of the column names, so there is nothing to set up.

Type the x-values straight down the first column and the y-values down the second:

x_1 | y_1
 1  |  1
 2  |  2
 3  |  4
 4  |  5
 5  |  6

The five points plot themselves as you type. That is the whole setup. Now every statistic the SAT can ask for is one line away.

Step 2: Read the Statistics Directly

On a fresh expression line (not inside the table), type a list function and point it at the column name. To type the subscript, use the underscore key — y, then _, then 1. Desmos evaluates the moment you finish:

• mean(y_1) returns 3.6 — the average of 1, 2, 4, 5, 6.
• median(y_1) returns 4 — the middle value once the list is sorted.
• stdev(y_1) returns about 2.07 — the standard deviation.

Two things worth knowing here. First, the SAT loves the difference between mean and median, so read the question's last line carefully — asking for one when you computed the other is the classic careless miss. Second, you almost never have to calculate a standard deviation on the Digital SAT; those questions ask you to compare the spread of two data sets conceptually. But if a question ever does want the number, stdev(y_1) hands it to you (and stdevp(y_1) gives the population version, about 1.85, if the wording calls for that instead).

You can also type total(y_1) for the sum or count(y_1) for how many values there are — handy when a question buries "how many students" inside a longer setup.

Step 3: The Line of Best Fit in One Line

This is the move that saves the most time. On a new expression line, type:

y_1 ~ a*x_1 + b

That squiggle is a tilde (~), which shares a key with the backtick, and it tells Desmos "find the values of a and b that best fit this data." Desmos immediately reports a parameters box:

• a ≈ 1.3
• b ≈ -0.3
• plus an R² value telling you how tight the fit is.

So the line of best fit is about y = 1.3x - 0.3, and Desmos draws it right through your points — exactly the line in the screenshot above. You never touched a formula for slope or intercept.

The two numbers map onto two things the SAT constantly asks about:

Slope a is the rate of change. Here a = 1.3 means every 1-unit increase in x predicts a 1.3 increase in y. If the table were "years since 2010" versus "members," you would say the model predicts about 1.3 more members per year.

Intercept b is the predicted value when x = 0. Here that is -0.3 — often a starting amount or a baseline, depending on the story.

Predicting Values From the Model

Once you have y = 1.3x - 0.3, prediction is arithmetic. Say a question asks, "Based on the line of best fit, what is the predicted value of y when x = 6?" Type 1.3*6 - 0.3 right into Desmos and read 7.5. Going the other way — "For what value of x does the line predict y = 10?" — type 1.3*x - 0.3 = 10 on its own line, or graph y = 10 alongside the fit and click the intersection: about x ≈ 7.92.

There is one more phrasing the SAT genuinely likes, and it trips people up: the residual — the gap between the real data and the line. "At x = 3, how does the actual value compare to the value predicted by the line of best fit?" The model predicts 1.3(3) - 0.3 = 3.6, but the table's actual value at x = 3 is 4. So the actual data point sits 0.4 above the line. Keeping "predicted" and "actual" straight is the entire question, and having both numbers on screen makes it trivial.

What the SAT Actually Asks

Watch for these stems — every one of them is now a table-plus-one-command problem:

• "The table shows... What is the mean (or median) of the values?" → mean(y_1) or median(y_1).
• "A line of best fit is drawn. Which value is closest to the slope of that line?" → the a from y_1 ~ a*x_1 + b, here 1.3.
• "Based on the line of best fit, what value does the model predict when x = ...?" → plug into y = 1.3x - 0.3.
• "Which best describes the rate of change?" → read the slope: about 1.3 per unit.

A few guardrails so the tool never bites you: name the column exactly (y_1, not y1 or y), use the tilde ~ for regression and the equals sign = for a plain equation, and re-check which statistic the question wants before you bubble anything in. The calculator is only as accurate as the data you type, so glance back at the table for a transposed number before trusting the fit.

Bring This to Real Questions

The fastest way to make this automatic is to run it on real, Bluebook-style data questions until the keystrokes are muscle memory. New here? Create a free account on UnlimitedTests and drill statistics questions with worked, expert-verified solutions that show the Desmos path. Already signed in? Head to your dashboard and pull up a Data Analysis set to practice reading a, b, and the mean straight off the screen under a timer.

Master this one skill and an entire family of Digital SAT statistics questions collapses into: type the table, type one command, read the answer.

Next in the series: Part 8 — sliders for "find the value of k" problems.