In math, the word "curve" refers to the graph of any equation, even if the graph is linear rather than curved. If the data points approximate a straight line, linear regression is the most appropriate type of curve analysis to perform. On most graphing calculators, this is called "LinReg." The researchers input the data into the calculator's statistical plot (usually called "STAT PLOT"), then select "LinReg." The calculator then calculates the linear equation that most closely matches the data.
The next order of curves is the quadratic category. Whereas linear curves have a variable raised to the first power, quadratic curves have a variable raised to the second power (for instance, "x^2"). If the data points form a parabola, quadratic regression is the most appropriate type of curve analysis to perform. On most graphing calculators, this is called "QuadReg." If the vertex of the parabola faces downward, the coefficient of the x^2 term should be positive. If the vertex of the parabola faces upward, the coefficient of the x^2 term should be negative.
Cubic curves have a variable raised to the third power ("x^3"). Cubic regression is appropriate if the data points are shaped like the graph of y = x^3, rather than forming a line or a parabola. On most graphing calculators, this is called "CubReg" or "CubicReg." If the slope of the curve points upward, the coefficient of the x^3 term should be positive. If the slope of the curve points downward, the coefficient of the x^3 term should be negative.
Periodic curves can have variables raised to various powers. Periodic curves are otherwise known as sinusoidal curves, because the epitome of the periodic functions is sine (x). On most graphing calculators, periodic regression is called "SinReg." Periodic regression is appropriate if the data points are shaped like a wave or other repeating pattern.