Hi all, i have a problem on how to get the combination of standard deviation for a number of ítem.examples im testing attitude. There are 5 question (item)to measure the attitude.the question using likert scale. Which analysis should i use to get only 1 standard deviation for the whole group instead of SD by each ítem. I have see that being done in a few papers but i only manage to get the SD by ítem.please help .

hi ShahSeville,
I'm not a Stats guru but hope this helps--

if I'm not mistaken, you can add up the 5 questions (assuming it is 5 questions to ONE item), and then divide by 5. Then just run descriptives again you will get one mean and one SD.

maybe someone else can help as well. Anybody? :-)

love satchi

Hi ShahSeville,
I hope I have correctly understood what you are asking!

In most cases, I would *not* recommend using means and SDs to describe raw data from Likert scales. This is because the data is not continuous (it is either interval or ordinal, depending on how you interpret it). More appropriate descriptive statistics could be the median or modal ratings across the five questions to report an average rating and the range, interquartile range etc. to report the dispersion of ratings.

But if you did want to get the SD ... Maybe when you are running the descriptive statistics, you are getting SDs by item because of the way your data is set out in SPSS. Are the ratings from your five questions set out in five separate columns? If that is the case and you want to get an SD of ratings from all five questions collapsed, then you need to put the ratings into a single column. You can do this by restructuring your data by clicking on "Data" and then "Restructure...". Then you need to select "Restructure selected variables into cases" and select "One" for "How many variable groups do you want to restructure". Next move the five questions in the box "Target Variable" and keep any other variables you want by moving them to the "Fixed Variables" box.

Hope that helps!