I am not an astronomer, but I imagine intergalactic dust leaves a distinct fingerprint on star spectra that you can measure.

Also, these numbers always have error bars. They spend a lot of time coming up with sources of error and characterizing them in order to understand the data fully.