The forgotten statistics – Explore

 In Statistics & methods

I have a feeling sometimes that many user miss the informative command: Analyze – Descriptive statistics – Explore. This command, I have put some focus on during my statistical trainings as it tells you a lot about your continous (scale) variables.

Here is the dialog box:

And here is the result:

I can see that the mean salary is 20 798 but the median is 18383, so I suspect a not normal distribution so then it’s better to present the median.

The spread is also showing a big variation from 14721 to 85182,  and the range between first and third quartile is 5197 for example.

As many statistical test require test for scale variables that follow a normal distribution, so that’s something that we study on the statistics training. So how can you figure out that by looking on some “strange” numbers at skewness?

Why don’t have a look on the help? Just double click the table, and then right click the Word (here “Skewness”):

So as the skewness value 4.107 is MORE than twice its standard error (0.077), we say that this distribution is very skewed. How is the skewness formed, here we have a long right tail as the value 4.107 is positive.

So because of this skewness this means that we have to be careful to use mean values in tests like t-test, but rather nonparametric tests if we want to compare salaries between 2 groups like male and female. But to be more specific, you should study the skewness for EACH group, so one for male and one for female. 

Would you like more how to think when working with statistics, and follow a red thread during your studies? Contact me on “Linked in” and I will give you more information.

Greetings Gunilla Rudander

 

error: Content is protected !!