continuous independent variable

What’s the best way to use a continuous independent variable to analyze task-based fMRI data within AFNI?

ie IV=continuous 1-10 scale of how much you like chocolate to analyze DV=neural activity during a decision-making task?

(In contrast to the usual dichotomous IV between-group choco-lovers vs choco-haters).

Tnx!

Are you interested in the effect of the quantitative variable? Or are you adjusting for its effect? What does your model look like?

Yes–I want to know the effect of the quantitative/continuous variable.

To see if the higher your chocolate-love score, the greater your accumbens area activation is when making a decision.

(example is embellished to make the point–not really doing a study of chocolate love)

-Dan

Dan,

In addition to the quantitative variable, what other explanatory variables do you have? Task types, groups of subjects, etc?

Task types/events–several.

Was trying to go “full RDoC”-and keep the sample dimensional/continuous–without dichotomous/bifurcated/categorical groups.

Is that possible?

If not–then–I guess–I have to define cut points–for a standard analysis (high vs low chocolate love–rather than full dimensional).

Tnx Gang and hope you’re ok!

-Dan

Dan,

It’s difficult to provide suggestions without knowing the whole picture and details: what exact variables are involved? What are you research hypotheses/questions?

My hypothesis is the greater levels of chocolate love on the Godiva scale are associated with increased striatal activity during a food preference task when picking desserts vs. asparagus.

I thought it would be interesting to keep the independent variable continuous/scalar (score on Godiva scale of chocolate love).

But–maybe that’s not possible within AFNI?

And I have to go the standard/traditional route–turn the continuous scale into a dichotomous “high chocolate love” vs. “low chocolate love”–and have a traditional 2 group IV?

Tnx for your help Gang!

turn the continuous scale into a dichotomous “high chocolate love” vs. “low chocolate love”-

That’s usually a bad idea, and I would not even consider that option.

If I understand your description correctly, you have two explanatory variables: (1) a quantitative variable (chocolate likeness), and (2) a categorical variable (food preference). This is a very basic and simple scenario from the modeling perspective. One clarification is still needed though: Does chocolate likeness change between the two food preferences?

We do not know if the continuous independent variable (chocolate loving on the Godiva scale) changes between the two preferences.

That’s what we want to find out via AFNI/imaging.

Put in more realistic terms, we have kids from across the range of irritability–with ratings of irritability on a continuous scale.

This is not a standard dichotomous group AFNI analysis.

Meaning–these participants were not gathered from any particular DSM categorical diagnoses (ie I did not pick them based on depression, bipolar disorder, ADHD etc.

And this is not a dichotomous/categorical independent variable “Patients vs controls” analysis.

I want to test the relationship between:
a) independent variable–continuous–parent rating on irritability scale

b) dependent variable–fMRI activation during a rigged feedback task–at least 2 main behavioral events–brain activity during rigged feedback vs. brain activity during accurate feedback.

How do I do this in AFNI?

Tnx!

-Dan

We do not know if the continuous independent variable (chocolate loving on the Godiva scale) changes between the two preferences.

Is “chocolate loving” some behavior data you collected from the subjects? If so, does each subject have one “chocolate loving” score shared between the two preferences? Or each preference is associated with a separate “chocolate loving” score?