Which analysis: comparing age and several variables

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

Vics
Posts: 4
Joined: Sun Nov 16, 2014 8:28 pm

Which analysis: comparing age and several variables

Postby Vics » Sun Nov 16, 2014 8:37 pm

I have a data set (observational data) where I have to compare age with several other variables. My hypothesis is that the variables' values will increase with increasing age, but which analysis should I run? Is it enough with a simple bivariate correlation? I'm going to look at the potential correlation between the variables, and gender differences as well. I have to write a report on this and I want the results part to be as good as possible, but my SPSS abilities is terrible.
GerineL
Moderator
Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

Re: Which analysis: comparing age and several variables

Postby GerineL » Sun Nov 16, 2014 9:38 pm

please provide more info on the type of data you have (i.e., # variables, type of measurement (ordinal? scale? etc.) etc).
Vics
Posts: 4
Joined: Sun Nov 16, 2014 8:28 pm

Re: Which analysis: comparing age and several variables

Postby Vics » Sun Nov 16, 2014 10:42 pm

We observed children for 30 minutes, and scored their behaviour every minute. Variables: three behaviour variables (ie 4 out of 30), age and gender. I want to see if these behaviours are related to age.
GerineL
Moderator
Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

Re: Which analysis: comparing age and several variables

Postby GerineL » Mon Nov 17, 2014 10:35 am

Thanks for that addtion, it is a little more clear now.
However, it is still not very clear what your behaviour variables look like (you say: 4 out of 30, but is is unclear what you mean by that).

So how is your data organized? what does your dataset look like? I don't need to know what you measure (i.e., aggression, withdrawn behaviour etc), but I do need to know how these behaviours are represented in your dataset.
For instance:

We have 40 cases, each case is on a row.
Variables are:
CaseID
behaviour1_minute1 (behaviour variable, scale of 1 to 5)
Behaviour1_minute2 (idem).
...
behaviour 1_minute40
We have these behaviour variables for 40 different behaviours, I am interested in 4 of these.
Vics
Posts: 4
Joined: Sun Nov 16, 2014 8:28 pm

Re: Which analysis: comparing age and several variables

Postby Vics » Mon Nov 17, 2014 1:01 pm

The behavior variables are mutually exclusive, so we registered only one behaviour each minute. As an example, case 1 was observed doing behaviour1 four times (minutes) during the 30 minute observation, behaviour2 three times and behaviour3 one time. We also had an "other" category (22 minutes where case 1 didn't do any of the three behaviours of interest). :)

We have almost 100 observations.
GerineL
Moderator
Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

Re: Which analysis: comparing age and several variables

Postby GerineL » Mon Nov 17, 2014 1:15 pm

I am sorry but it is still not clear.
You give information about the way you collected data.
I need information on how your data-set is set up.
So, can you indicate what variables you have, and which values they can hold?
Vics
Posts: 4
Joined: Sun Nov 16, 2014 8:28 pm

Re: Which analysis: comparing age and several variables

Postby Vics » Mon Nov 17, 2014 7:56 pm

Oh, ok, sorry. :)

We observed each child for 30 minutes, and registered what they did each minute on a schema. For each minute, we could only make one registration, so each of the variables has a value between 0-30. Each child/case has a total score of 30.

As an example:
Case x:
Minute___Behaviour1___Behaviour2___Behaviour3___BehaviourOther
1___________X
2______________________ X
3___________X
...

I want to see whether behaviour 1, 2 and/or 3 is related to each other, and whether they are related to the child's age.
GerineL
Moderator
Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

Re: Which analysis: comparing age and several variables

Postby GerineL » Tue Nov 18, 2014 9:24 am

OK! And I assume you have another variable called "child" or something, so it is actually.

Child - minute - behaviour...
1 ------1---------....
1 ------2---------....
1 ------3---------....
etc

right?

So, what are you interested in exactly? You say "increase" and "related to each other".
But this can be interpreted in sevaral ways.

For instance, you could just compute how often a child displays each of the behaviours, and see whether there is a relation between age and each of these vars, and see whether the behaviours are correlated to each other.
But maybe you mean how likely it is for a certain behaviour to occur, if another behaviour occurs.
Maybe you mean something else entirely...

The first one (how often each child displays each of the behaviours, and do analyses with this) is relatively easy, the others require much more.

What is it you want?

Who is online

Users browsing this forum: No registered users and 2 guests

cron