I personally believe that mixing to "manipulate the listener" in a manner like you describe is bad technique, but that is admittedly a subjective judgement that hinges upon one's definition of "bad". But here's my reasoning on it:
First, who came up with the idea that getting a listener to turn up the music is necessarily a good thing? Many people would find that having to turn up a song to hear it "properly', just to have to turn it down again or be blasted when the next song comes on would be irritating.
Second, turning it up is not always an option. Examples: listening on earbuds where you're already listening to the rest of your music at near-damaging levels, listening in an office, home, or car with passengers, where loud volumes my be socially unaccepted. In such cases, any supposed "marketing value" to such a manipulative mix would have a negative value, not a positive one.
Third - and most important, IMHO - is that one should mix for the demands of the song itself, and not for any external motivation like marketing or manipulation of the listener. If that particular song called for lower-level vocals, then they did the right thing. If the vocals "should" have been forward, but they kept them back as a tease, then I think they did the wrong thing by doing injustice to the song.
If the main purpose of the vocal in the composition and arrangement is basically to provide sound as if it were simply an instrument playing a melody or harmony or solo, and the actual lyrics were somewhat secondary, then it could very well call for a vocal that sits inside the mix. Examples could include anything from R.E.M. with their nonsensical lyrics that carry more melody hooks than verbal information, to Louis Armstrong who sings and scats many of his vocals as if he were playing his horn instead of singing, to the Beach Boys whose vocal arrangements in many songs carried far more impact than the lyrics.
There are even times, for the right music, where low vocals are adding to the "spacey" feel of the song. Imagine much Hawkwind or even some Floyd if they didn't ride some of their vocals roughshod with heavy verb and low gain levels to match the ethereal feel of the rest of the production. It wouldn;t work right.
If, OTOH, the lyrics are one of the main driving forces of the song, it would be, at best, unnecessary to make the listener turn up the sound to get the ipact of the song and, at worst, suicidal to bury them in the mix when they are a major strength of the song. Would you have to (or want to) hold back on the lyrics of your typical Lyle Lovett, Tom Waits or Bob Dylan tune, so that the listener has to turn it up? What's the point?
Bottom line in my personal book: If the producer feels they need to resort to some parlor trick in production to trick the listener into getting interested in a song, that song probably needs to be sent to the Recycle Bin instead of to the duplicator.
G.
) a bad technique. 