Reporting evidence in favot of excluding a predictor

Modeling
,
1

Hi Everyone,

A small basic question. I have a logistic regression (sampled with brms) where y is predicted by x1, x2 and their paired interaction. I want to be able to report evidence in favor of no interaction effect.
I sampled these two models and used loo to compare them:

model_h0= brm(y ~ 0 + Intercept+x1+x2+(1+x1*x2| subject), 
                         data = df
                         family = bernoulli(link = "logit"),
                         warmup = 1000,
                         iter = 2000,    
                         cores =4,
                         chains=4,
                         backend = "cmdstan",
                        prior=mypriors)

model_h1= brm(y ~ 0 + Intercept+x1*x2+(1+x1*x2| subject), 
                         data = df
                         family = bernoulli(link = "logit"),
                         warmup = 1000,
                         iter = 2000,    
                         cores =4,
                         chains=4,
                         backend = "cmdstan",
                        prior=mypriors)

model_h1=loo(model_h1)
model_h0=loo(model_h0)
loo_compare(model_h0,model_h1)

What I really want is to be able to say that “A model without the interaction population level effect was X times more likely given the data compared to a model that included the x1:x2 interaction”

How can I get there? Or maybe other ideas on how to tackle this basic report?

Many thanks,
Nitzan