Comparing two linear regression gradients?

#1
Hi,
I am completely new to linear regression so this may sound daft, but I'm wondering if its possible to test for a significant difference between the gradients of two linear regression lines, and if so how can I do this in SPSS?

Basically I'm comparing the number of foreign words learnt in a week by adults and children and have two linear regression lines which show that the number of words decreases very rapidly between the ages of 1-10 and much less rapidly between the ages of 20-55. I want to know how I can perform a statistical test to show this difference.

Thank you!
 

rogojel

TS Contributor
#2
Hi, from what you told us it is not clear what the regression model is, but I assume you want to show that there is a significant difference in the slopes of the regression lines for the two age groups adult and child.

You can analyze all the data together and add a dummy variable for Age ( e.g 1 if Adult 0 if child) . The difference in slopes translates to a significant interaction between this Age variable and whatever X you might have.

I hope this helps a hit.
 

Karabiner

TS Contributor
#3
have two linear regression lines which show that the number of words decreases very rapidly between the ages of 1-10 and much less rapidly between the ages of 20-55.
What did you regress on what? "Number of words learnt" on "age of the learner"?
"Number of words" on something else?

With kind regards

K.
 
#4
What did you regress on what? "Number of words learnt" on "age of the learner"?
"Number of words" on something else?

With kind regards

K.

Yes, "Number of words learnt" on "age of the learner". I want to compare the linear regression gradients between children and adults
 

hlsmith

Omega Contributor
#5
rogojel's post is what you would want to do, see post #2. You should also use the word slope not gradients. Lastly, it will be beneficial for you, while learing, to graph these data and see what the results look like in action. Some times this can help you better understand and at times find errors in your processes.
 

Karabiner

TS Contributor
#6
rogojel's post is what you would want to do, see post #2. You should also use the word slope not gradients. Lastly, it will be beneficial for you, while learing, to graph these data and see what the results look like in action. Some times this can help you better understand and at times find errors in your processes.
He seemingly has only 2 variables, age and number of words.
Will the introduction of "age category" work here? It is confounded
with age, i.e. all subjects with age < 20 are "young", all other subjects
are "old" (and the distinction is data-driven, as it seems, so testing
could be pointless).

With kind regards

K.