Robust vs Bootstrap standard errors for logistic regression model

I want to run a logistic regression with clustering. Can someone provide input regarding the use of Robust vs Bootstrap standard errors? Depending on this choice, the p-values for certain variables are not statistically significant (significant with bootstrap but not with Robust). We want to make the right choice regardless of the statistical significance of the findings.

It will be useful to compare the results with those from a random effects (for clusters) logistic model. For a Bayesian approach to that see

Did you do cluster-based bootstrap?