I checked the normality of my data on SPSS and one of the variables is not normally distributed. I have the mean, standard deviation, skewness, kurtosis , min and max values of my distribution. But I do not know the name of my distribution. How can I learn which type of distribution (Bernoulli, Beta, Binomial, Categorical, Exponential, Fixed, Gamma, Lognormal, Negative Binomial (Failures, Trials), Poisson, Range, Triangular, Uniform, Weibull) I have? If I can find it, I will simulate a data based on my distributions on SPSS
1 Answer
$\begingroup$
$\endgroup$
1
Neither. Those distributions are used only to approximate the distribution of your data. First of all, normality checking is pretty useless procedure. Use the distribution that is useful as an approximation of the distribution of your data, that shares the characteristics you find important. E.g. if the data is continuous and the distribution is roughly symmetric, Gaussian might be a reasonable choice, if you have roughly i.i.d. binary events, binomial might be a good choice, etc. Neither of those is the distribution of your data, they are just used to approximate it.
-
$\begingroup$ Thanks for the answer! Let's put it that way. I need to select a distribution type for simulation. My variable is continuous Mean= 55, Stddev= 20. My range is from 0 to 120. Most frequency is at between 40-60, I got 0 frequency between 80-100, but 2 frequencies at 20-40, 60-80 and 100-120. Distribution does not seem normal. But I cant identify how does it look... $\endgroup$– user310572Feb 6, 2021 at 23:51