Anyone good with SPSS?

Ekancone

Hey, i am working with some grouped data and i want to calculate the mean using the midpoint. How can i do this using SPSS?

For example

1-99 Freq : 5
100-199 Freq : 6
200-299 Freq : 4

Any help appreciated.

2Scoops

You'll only be able to get an estimate of the mean with grouped data, but probably the easiest way is simply to get all the data into a single column and get descriptives or draw a histogram.

Ekancone

2Scoops wrote: »

You'll only be able to get an estimate of the mean with grouped data, but probably the easiest way is simply to get all the data into a single column and get descriptives or draw a histogram.

So just get the midpoint of each group and the calculate the mean? Do i have to do it manually, i was hoping there would be a feature of SPSS that would do it for you?

2Scoops

I'm having difficulty understanding your situation. If you have the raw data, SPSS can pretty much do anything you like with it. If you only have the frequency of values in certain categories, which is my understanding, you have to work within that limitation.

If you want SPSS to give you results, you'll have to manually put in the data. There are different ways of doing this but the easiest, considering the information you want to get, would be to put all the values in one column. So, if there are 5 values between 0-99, call it group 1 and put 5 instances in the column, setting it up as ordinal data. And so on with 100-199 etc.

Is this what you're trying to do?

Ekancone

2Scoops wrote: »

I'm having difficulty understanding your situation. If you have the raw data, SPSS can pretty much do anything you like with it. If you only have the frequency of values in certain categories, which is my understanding, you have to work within that limitation.

If you want SPSS to give you results, you'll have to manually put in the data. There are different ways of doing this but the easiest, considering the information you want to get, would be to put all the values in one column. So, if there are 5 values between 0-99, call it group 1 and put 5 instances in the column, setting it up as ordinal data. And so on with 100-199 etc.

Is this what you're trying to do?

No not at all, i conducted a survey about expenditure and i used a range of values as the selection choices.

eg. How much rent do you pay?

0-99 {}
100-199{}
200-299{}
etc,etc

I inputted the data in code form so that 0-99 = 1, 100-199 = 2, etc etc. But when i calculate the frequencies all i can use is the mode to get any measure of central tendency. When i attempt to use mean or median all i get is the mean of the coding system!

I know that the mean and median can be calculated by using the midpoint of the groups. So is there a way i can get SPSS to do this for me instead of me having to create new variables?

2Scoops

daithifleming wrote: »

When i attempt to use mean or median all i get is the mean of the coding system!

Ok, I get it now. It's easiest to get the median just by counting down - and it's also probably the best measure of central tendency in this case. I would go with that.

You will not be able to get a midpoint-based estimate of the mean without creating new variables, sorry! Actually, it's a job better suited to excel - just a few minutes work, if absolutely necessary!

Ekancone

I figured this one out anyway. You simply have to recode the variables to represent the midpoint values, which can be calculated outside of SPSS. It was actually really easy once i found the command to do it.

Df - you want to be an economist right?

Learn how to use R.

I've just gotten into it and its fantastic - a lot better then eviews.

Ekancone

Thats great, i will check it out!