Chat with our AI personalities
normal distribution
normal distribution
There are many different bases for standardisation - even if you only consider the Gaussian (Normal) distribution. If a variable X has a Gaussian distribution, then the corresponding Standard Normal deviate, Z, is obtained from X by subtracting the mean of X and then dividing the result by the standard deviation of X. The variable, Z, more commonly known as the Z-score, has a Gaussian distribution with mean 0 and standard deviation 1. But, if X is an IQ score, for example, different measures of the X variable are used so that the resulting variable has mean 100 and standard deviation 15.
There is 1) standard deviation, 2) mean deviation and 3) mean absolute deviation. The standard deviation is calculated most of the time. If our objective is to estimate the variance of the overall population from a representative random sample, then it has been shown theoretically that the standard deviation is the best estimate (most efficient). The mean deviation is calculated by first calculating the mean of the data and then calculating the deviation (value - mean) for each value. If we then sum these deviations, we calculate the mean deviation which will always be zero. So this statistic has little value. The individual deviations may however be of interest. See related link. To obtain the means absolute deviation (MAD), we sum the absolute value of the individual deviations. We will obtain a value that is similar to the standard deviation, a measure of dispersal of the data values. The MAD may be transformed to a standard deviation, if the distribution is known. The MAD has been shown to be less efficient in estimating the standard deviation, but a more robust estimator (not as influenced by erroneous data) as the standard deviation. See related link. Most of the time we use the standard deviation to provide the best estimate of the variance of the population.
It shows primarily that the measurement unit used for recording the data is very large. For example, the standard deviation of the heights of individuals, when recorded in metres, will be one hundredth of the standard deviation of their heights when recorded in centimetres. The process is known as coding.