Crappy earbuds are more likely to expose flaws in lossy encodes than high-end gear.
The psychoacoustic model assumes that certain sounds will mask other sounds so you can leave them out or replace them with something that doesn't sound quite right. If like most cheap headphones you don't have a flat response then the masking sound can be too low to work as intended.
Having said that, with modern encoders it's mostly not an issue, even if you're using 90kps Vorbis created specifically for portable use. What is an issue is that you'll probably want to use the same file for portable and home listening, or you'll have got the file from someone who wanted to be on the safe side, since some songs are trickier to encode than others, so you'll end up with a 256kbps AAC file from iTunes, or 320kbps MP3. At that point the jump to FLAC compression sizes isn't particularly great and the management becomes a bigger hassle plus it allows you to transcode to various other formats and sizes as required, saving even greater space for listening in noisy environments like cars, trains, buses etc.
I like my classical music, jazz, and noise music to be of the highest quality possible. The sound quality of most MP3s is distracting.