A scatter plot tracks a collection of data on axes to determine a correlation between variables. The correlation between the variables equals the slope that the points collectively make on the graph. Though the points don't make a solid line with a distinct slope, an artificial line through the points will approximate this gradient. Computers and graphing calculators draw these lines and compute the gradient automatically. You can, however, do this calculation manually.

Draw a line on your graph around every point on the graph, other than anomalous points that appear far from the others. The line should create an oblong shape.

Draw a line through the shape, dividing it into two oblong halves that equal each other in area. This line is the scatter plot's line of best fit.

Choose any two points on this line. These points may or may be actual scatter points. In fact, the line may or may not go through any actual original scatter points.

Subtract the first point's ycoordinate from the second point's. If the points coordinates are, for instance, (2, 9) and (4, 15): 9  15 = 6.

Subtract the first point's xcoordinate from the second point's: 2  4 = 2.

Divide the difference in ycoordinates from the difference in xcoordinates: 6 / 2 = 3. This is the line's slope and the points' evident correlation.
References
 Photo Credit Jupiterimages/Goodshoot/Getty Images