Predictors of supra-threshold speech-in-noise intelligibility by hearing-impaired listeners