Formula sheet: Statistics
Statistics14 min readLesson 1 of 6
1.1 Measures of dispersion
A measure of central tendency tells us where the data are centred, but two data sets can have the same mean and still be very different. The runs of two batsmen may average fifty, yet one may score steadily near fifty while the other alternates between nought and a century. The amount of scatter is measured by a measure of dispersion, and the standard measures are the range, the mean deviation about a central value, and the variance with its square root, the standard deviation.
1.2 Grouped data
For a frequency distribution each observation is weighted by its frequency. In a continuous distribution each class is represented by its mid-point, and the mean is computed from those mid-points. The mean deviation is then the weighted average of the absolute deviations, and N stands for the total frequency.
1.3 Reading the formulas correctly
Three symbols need care. The letter n counts observations in ungrouped data, while N counts the total frequency in a frequency distribution; mixing them is the commonest error. The mean x bar is the central value about which deviations are measured, and for a frequency distribution it is the weighted mean, not the simple average of the distinct values. The variance carries squared units, so the standard deviation is usually quoted alongside it. When the data are grouped, the class mark stands in for every observation in the class, and this is an approximation. These conventions decide which formula applies and keep the arithmetic consistent across the chapter.
1.4 Computing formulas
The variance is also written in a form that avoids calculating every deviation. The mean of the squares minus the square of the mean gives the same number, and for grouped data the frequencies weight each term. When the numbers are large, step deviations shrink them: choose an assumed mean a and a common factor h, write d = (x - a)/h, and use the step-deviation formula for the variance.
Exam
The variance changes by the square of a scale factor: if every observation is multiplied by k, the variance becomes k squared times the old variance. Adding the same constant to every observation shifts the mean but leaves the variance unchanged.
