p-values are a minimum standard. They demonstrate that certain models are too simple to explain the data. By all means having ruled these simpler models out go Bayesian with your model of choice, but there's no good reason not to first demonstrate that your model's complexity is necessary... and it's rather suspicious if you don't.