In my opinion, I would say "list of publications" (for PhD students as you have stated) which is generally found in the CV. When I think of hiring a PhD student, I think of a student who was able to publish 1-3 journal articles (maybe same number for conference proceedings) during his MS. A student who knows how to design/conduct an experimental testing (or at least was a part of one) and have a good grasp on his field. keep in mind that different students have different background and each field is different (for publication purposes). But, a good work at the MS level can result (most of time) in publishing in 1-2 journal articles. Even if he or she have had some help (most likely they would) from their adviser in writing and publishing. At least I know that they know the process and "been there, done that". For me, a student at PhD level should somehow be able to think independently and has his/her own scientific charisma!
I'm not trying to disregard an industry experience (which is very valuable, depending on the field). However, it is my preference to see solid proof of academic achievement. I tend not to weigh academic awards such as "outstanding student of department X award" that much since a lot can goes into that. However, I would weigh other awards like "Best paper in a conference" more heavily. I would also like to consider students who I have interacted with or know their advisers.
Also, I would not weigh GPA that much, especially at undergraduate level. I'm an example of a student with an undergrad GPA of < 2.6! Been admitted under probation couple of times too! But, in grad school, I was a completely different person. I would like to work with a student that can "think, analyze and be passionate about research" than working with a student with 4.0 GPA that can not figure out "stuff" if they are to be different than a text book example. I would not dare to generalize this statement, I would rather agree to the fact that there will always be that "odd" student who breaks the general norm.
Still, I would read his/her research statement and consider it to some extent. As you can not be really sure if s/he wrote it on his/her own, had somebody to go over it etc. If you were to take your chances and only judge a 1-2 page research statement (that has been edited multiple times by couple of people to perfect it), rather than list of publication (if the student has any), awards, personal connection (that I trust and value a lot) and recommendation letters, you might not end up with a sound selection. I think one should evaluate the student as a "full package" rather than focusing on independent criteria.
I'm interested in seeing what others would post and share. Especially with different prescriptive and fields.