Hello عاطف يوسف ,

In SPSS, goodness-of-fit tests for the normal distribution may be found in Analyze / Descriptive Statistics / Explore subprogram. Be sure to click the "Plots" button, and check the "Normality plots with tests" check box in the subsequent dialog box.

Remember, if you intend to run some sort of analysis like anova or regression, the normality issue applies to the behavior of model residuals, not the original set of scores.

Good luck with your work.

You don't. The sample data you collect cannot show that the population distribution is exactly normally distributed. If you are trying to estimate a parameter or test if the population value of a particularly X, these procedure can make different assumptions. If you have a more specific interest, you should share it with readers. If you have a continuous variable and want to make it appropriately normal, you can rank the data then turn them into z scores. Other transformations would require knowing the sample distribution.

@my dear friends Daniel and David I was surprised that neither of you said that this has nothing to do with parametric statistics. One of my favorite papers makes this point indirectly. I have attached this here. Best wishes, David Booth

David Eugene Booth , I did say different procedures have different assumptions. I was going to give the example of estimating the median, but the logistic regression in the paper works. Thanks for the paper!

You to check the outlier test first, then you should delete the outliers and retest the normality. Otherwise how many of your samplings do you have? What kind of tests you are intending to do?

deleting outliers is a bad idea. First, check how your DGP generated those outliers. They can be legitimate data points. by all means refrain from deleting outliers. Instead, use models that accomodate outliers.

Very easy and good video to learn about the data collected follows approximately normal distribution or not through SPSS

https://www.youtube.com/watch?v=IiedOyglLn0

(Normality test using SPSS: How to check whether data are normally distributed)

For testing the normality of data, you can follows Shapiro-Wilk Test of Normality or Normal Q-Q Plot. By locating and removing outliers from the data set, you can easily convert the data to a normally distributed one.

Apart from using a Goodness-of-fit as mentioned by David Morse, testing for normality can be done through normal Q-Q plots, histograms, skewness, kurtosis, Kolmogorov-Smirnov or Shapiro-Wilk (wanted to add to the list by Sangeetha K L). I use Kolmogorov-Smirnov (population above 100) and Shapiro-Wilk (population below 100) – other authors may suggest different population threshold. The p-values for Kolmogorov-Smirnov and Shapiro-Wilk should be greater than 0.05 for normal distribution.

Kolmogorov-Smirnov for more than 30 sample size and Shapiro-wilk for less than 30 sample size.

The data exploration procedure helps determine whether the statistical analysis methods you are considering for SpSS analysis are appropriate

Including Exploring normal distributions

You can use Kolmogorov-Smirnov (with Lilliefors significance correction) for more than 30 in sample size and Shapiro-Wilk for less than 30 in sample size.

if we want to do the test 'tow way ANOVA in SPSS but our data don't follow a normal distribution, What is the alternative before us?