I'm trying to understand how people decide to go to CS grad school in the US. For instance, what proportion of PhD applicants are coming straight from undergrad, versus another graduate degree (such as an MS in math), versus industry? Do they generally decide first on what they want to research, whom they want to work with, or that they want to go to grad school at all? I presume there is a survey of this somewhere?

I've found the Computing Research Association's Center for Evaluating the Research Pipeline's report on why grad students choose computing, and I've also looked at the National Center for Science and Engineering Statistics website. I have found a little data on what proportion of doctoral recipients were previously in a baccalaureate program, a master's program, or industry, but not about what proportion of applicants, or accepted applicants, come from those categories.

Any pointers?

(Maybe there's paywalled research on this within the ACM's special interest group on CS education? If so, lemme know and I will try looking for that?)

The reason I am asking this is to help professors and guidance counselors I know (maybe you've heard that Outreachy has a new career advisor). They want to improve their abilities to help students and programmers consider research careers, and better target their outreach consider applying to specific graduate programs. How does the engagement funnel currently work? And thus, where are the gaps? I presume there are a bunch of people who would do well in grad school, and find it fulfilling, and use their research and their degrees in ways that would benefit the world, but no one ever says to them "hey have you thought about going to grad school," so they don't think of it as a possible thing for them.