Linear Functions, Slope, and Rate of Change

A relationship is linear when equal steps in the input always produce equal steps in the output. Walk one more mile and the trip costs the same extra amount every time. Wait one more month and the balance drops by the same number of dollars every time. That single idea, constant change per unit, is the whole subject; everything else on this page is bookkeeping about how to write it down.

Linear functions are the single most heavily tested topic in SAT math. They dominate the Algebra domain, and they reappear inside problems that look like they are about something else: systems of equations, unit conversion, geometry with coordinates, and data interpretation questions that ask you to describe a line of best fit. The ACT tests the same ideas with a stronger emphasis on speed and on coordinate geometry, so you will see more questions about parallel and perpendicular lines, midpoints, and distance.

What separates a student who scores well here from one who does not is usually interpretation rather than arithmetic. Almost everyone can compute (y2 - y1) / (x2 - x1). Far fewer can look at C = 250 + 18h and say immediately that 250 is a one-time charge, 18 is dollars per hour, and the number 18 would be unchanged if the setup fee doubled. Both tests are built to reward the second skill, so this lesson spends as much time on meaning as on procedure.

Slope is defined as the change in the output divided by the change in the input: m = (y2 - y1) / (x2 - x1) for any two distinct points on the line. The reason this ratio is safe to compute from any two points you like is that a line has the same steepness everywhere. Pick two points close together and two points far apart, and the right triangles formed by their rise and run are similar triangles, so the ratio of rise to run is identical. That is why a linear function has one slope rather than a slope that depends on where you look, and it is why a table that is genuinely linear will show the same output change for every equal step in the input.

Slope carries units, and paying attention to them prevents most interpretation errors. If y is measured in dollars and x in hours, then m is dollars per hour, because you divided dollars by hours. This makes slope a rate of change. A positive slope means the output grows as the input grows; a negative slope means it shrinks; a slope of zero means the output never moves, giving a horizontal line. A vertical line has no slope at all, because the run is zero and division by zero is undefined, which is also why vertical lines are not functions.

Slope-intercept form, y = mx + b, is the form to reach for when you want to graph quickly or read off the starting value. Here b is the output when x = 0, which is why it is called the y-intercept: it is where the line crosses the vertical axis. In a modeling problem b is the fixed part, the amount that exists before anything happens, and m is the repeating part, the amount added for each additional unit. Note that b is only a meaningful starting value when x = 0 is meaningful in the context; if x counts years since 2010, then b is the value in 2010, not the value now.

Point-slope form, y - y1 = m(x - x1), is the form to reach for when a question hands you a slope and a point that is not the intercept. It works because the slope formula, rearranged, says that for any point (x, y) on the line, (y - y1) / (x - x1) = m, and multiplying both sides by (x - x1) clears the fraction. Writing the line this way takes one step instead of the two steps it takes to substitute into y = mx + b and solve for b, and it is far less error prone under time pressure. You can always expand it into slope-intercept form afterward.

Standard form, Ax + By = C, is the form that shows up in constraint problems: A dollars per unit of one thing plus B dollars per unit of another equals a fixed budget C. Its two intercepts fall out instantly, since setting y = 0 gives x = C / A and setting x = 0 gives y = C / B. Its slope is -A / B, which you can confirm by solving for y: By = -Ax + C, so y = (-A / B)x + C / B. Recognizing standard form matters because the test likes to ask for the slope of a line that is written this way, hoping you will misread A as the slope.

Two distinct lines are parallel exactly when they have the same slope and different intercepts, since equal steepness means they never converge. Two lines are perpendicular exactly when their slopes multiply to -1, that is, when one slope is the negative reciprocal of the other. The reason is geometric: rotating a direction arrow with run p and rise q by ninety degrees produces an arrow with run -q and rise p, so the slope changes from q / p to p / -q = -p / q. Flip and negate. The same slope comparison decides how many solutions a system of two linear equations has, because a solution is a point that lies on both lines: different slopes cross exactly once, equal slopes with different intercepts never meet, and equal slopes with equal intercepts describe one line with infinitely many shared points. Finally, to recognize these question types quickly, watch for the phrases constant rate, per, each additional, and the request to interpret a coefficient; each is a signal that the underlying model is linear and that the tested skill is reading the slope or the intercept in context.

In the interactive graph below, drag the slope slider and watch the line tilt around its y-intercept while the x-intercept marker slides along the axis; its position is always -b / m, which is why it races away as the slope approaches zero. Drag the intercept slider and watch the whole line slide up and down without changing its steepness. Set the slope to 0 to get a horizontal line and notice the x-intercept readout disappears, then try to find pairs of slopes that would meet at a right angle, such as 2 and -0.5.

Inverting the slope formula. Students who are rushing sometimes compute run over rise, (x2 - x1) / (y2 - y1), which produces the reciprocal of the correct answer. The fix is to say the words out loud as you write: rise over run, change in output over change in input. If your answer is the reciprocal of a listed choice, you probably made this error, and the test writers will have included that reciprocal as a distractor.

Mismatching the order of subtraction. If you compute the numerator as y2 - y1 you must compute the denominator as x2 - x1, using the same point as the starting point in both. Subtracting in opposite orders flips the sign of the slope, turning an uphill line into a downhill one. Circle which point you are calling point one before you compute anything.

Treating the first row of a table as the y-intercept. The intercept b is the output when x = 0, not the first output listed. If a table starts at x = 3, the value there is not b, and you must either extend backward or solve for b algebraically. This mistake is especially costly in modeling questions that ask for the fixed cost or initial amount.

Using the reciprocal without negating it for perpendicular lines. A line with slope 3/4 is perpendicular to a line with slope -4/3, not 4/3. Both the flip and the sign change are required. A quick sanity check is that perpendicular slopes must have opposite signs, since one line goes uphill and the other goes downhill, unless one is horizontal and the other vertical.

Assuming a table is linear because the outputs change by a constant amount, without checking that the inputs step by a constant amount. Outputs of 5, 8, 11 look linear, but if the inputs are 1, 2, 4, the rates are 3 per unit and then 1.5 per unit, so the relationship is not linear. Always confirm equal input steps before you read off first differences.

Stopping at equal slopes when a system question asks about solutions. Equal slopes tell you the lines are parallel or identical, but not which. You must also compare the intercepts: different intercepts means no solution, equal intercepts means infinitely many. A question that asks for the constant that gives infinitely many solutions requires both conditions, and choosing the value that only matches the slopes is the most common wrong answer.