Hello everyone,

I'm using SPSS version 24 and I have a question regarding the kurtosis values provided under Analyze > Descriptive Statistics > Explore. Specifically, I would like to know:

Is the kurtosis value shown in the output the raw (direct) kurtosis, or is it the excess kurtosis (i.e., K - 3)?

Understanding this is important for correctly interpreting my data distribution, and I want to be sure whether I need to subtract 3 myself to get the excess kurtosis or if SPSS already does that automatically.

Thanks in advance for your help!

Best regards,

Hi @JOSE ACEVEDO
This is the kind of question most often answered by the IBM SPSS Statistics Algorithms guide.
In this case, see printed pages 291-292. (PDF viewer pages 365-366)

Hello Jose.  As @David Dwyer has said, you can consult the algorithms documentation.  But sometimes, the equations used there are a bit hard to decipher.  So here is another approach you could try.  1) Generate a (pseudo-)random sample from the standard normal distribution; 2) Compute kurtosis using whichever SPSS command you would use; 3) See if the value comes out closer to 0 or 3.  Here's an example I just cobbled together.  Perhaps you'll find it helpful.

NEW FILE.
DATASET CLOSE ALL.
SET RNG MT.
SET MTINDEX 250521.
INPUT PROGRAM.
- LOOP #i = 1 to 10000. 
-  COMPUTE Case = $casenum.
-  COMPUTE Y = RV.NORMAL(0,1).
-  END CASE.
- END LOOP.
END FILE.
END INPUT PROGRAM.
EXECUTE.
DATASET NAME raw.

DESCRIPTIVES VARIABLES=Y
  /STATISTICS=MEAN STDDEV MIN MAX KURTOSIS.

FREQUENCIES VARIABLES=Y
  /FORMAT=NOTABLE
  /STATISTICS=KURTOSIS SEKURT
  /ORDER=ANALYSIS.

EXAMINE VARIABLES=Y
  /PLOT NONE
  /STATISTICS DESCRIPTIVES
  /CINTERVAL 95
  /MISSING LISTWISE
  /NOTOTAL.

* The kurtosis value is very close to 0.
* This implies that SPSS reports the so-called 
* "excess" kurtosis.  I.e., it subtracts 3. 

The kurtosis value SPSS computes is .02370 for this example. 

I hope this helps.  ;-)