The end of your question makes it sound like you are confused about something: arXiv and conference/journal submissions are not alternatives. The arXiv is a preprint server, and posting there is like posting to on a personal webpage in that arXiv postings are not counted as publications and do not preclude you from submitting to a peer-reviewed venue.
As for how a committee would evaluate published verus unpublished papers differently, one gets a preliminary sense of the quality/significance of a published (or accepted) paper from the venue it was accepted to. For unpublished papers, there is no immediate indicator of this (it could be great, mediocre, or seriously flawed), but they certainly signal you're research active, and since you already have a couple of published papers, assuming they are in reasonable journals, chances are your unpublished papers will be assumed to be of reasonable quality.
But, for both published and unpublished papers, an assessment of the quality should come from your letter writers, who should be able to give more insight than just the reputation of the journals you have published in.
In addition, there is no reason to delay your submission. On your CV, just list the title of the paper, with the word "submitted." (If you want, you can make a separate preprint section, but this seems unnecessary with only 4 papers.) If it is on the arXiv, you can give an arXiv link. Even for papers not submitted, you can still list them as preprints (though submitted or at least being posted online is better).
Note: I disagree with the other answer about saying what journal you submit to--saying where you submit to gives no indication of the quality of the paper (maybe just your biased perception of its quality), and it leave me personally with a slightly desperate impression, though this may depend on the field. (In mine, some people do this, but most do not.)