k-tree
E-learning book

Normality of the distribution

In the examples in this article, data is generated every time the page loads. If you want to see an example with different values - reload the page.

Some statistical tools assume that the distribution is normal. The algorithm for checking the normality of the distribution will be given below, and also an example in excel.

Distribution law

Checking for compliance with the normal distribution is a special case of solving the problem on finding among the known distribution functions one that describes as accurately as possible this distribution.

First of all, it is necessary to structure the available values, in the article properties distributions it describes how the distribution series is constructed, so here I will omit the details and give source data and processed values:

151 150 154 141 159 163 140 165 137 166
125 150 148 133 161 155 154 139 158 170
158 157 131 143 148 152 155 153 164 141
167 156 142 146 149 142 144 145 141 149
152 157 140 146 147 152 155 160 146 144
148 150 136 163 148 144 161 146 139 140
159 141 155 172 156 135 164 139 161 144
138 160 138 142 136 157 153 137 143 161
143 129 145 151 147 146 159 159 159 145
141 161 156 143 137 163 155 167 154 133
Table 1. Initial data for checking the normality of the distribution
# 12345678910
x2311151812141491
pi0.020.030.110.150.180.120.140.140.090.01
Table 2. Number of elements in each interval
Graph 1. Distribution range

Regardless of what we see on the graph, we need to check whether whether the distribution is normal.

The characteristics of a normal distribution are the mean and standard deviation. Let's calculate these values for our distribution:

μ = 149.6
σ = 9.92
The calculation of the mean and standard deviation is described in the article distribution parameters

Normal distribution

The normal distribution curve for μ=149.6 and σ=
Warning: Undefined variable $variation in /var/www/content/ktree/t9n/en/articles/statistics_check_is_normal.php on line 148
:

P(x) = e^[-0.5((x-149.6)/9.92)2] / [9.92√2π] Normal distribution formula
Graph 2. Distribution series and normal distribution, μ = 149.6, σ = 9.92

First approximation

Let's try to invent a criterion of normality, the simplest, what comes to mind is to determine the percentage of compliance the normal curve and the existing distribution.

To do this, add up the absolute values of the differences across all points of the graph, find the area under the normal distribution graph and calculate the deviation of interest, I will call such a criterion "criterion of normality" and I will decide that if the deviation more, let's say 30%, then the distribution is not normal.

diff = Σ|D(X) - P(X)|
S = ΣP(X)
Δ = diff / S
diff = 37.72
S = 63.41
Δ = 59%

The deviation is 59%, therefore, I draw the following conclusion: the distribution is not normal according to the normality criterion.

Download article in PDF format.

Do you find this article curious? /

Seen: 4 937


Read the following
Analysis of variance