In a paper published 25 years ago, Ross and Morrison ("Educ Technol Res Dev" 37(1):19-33, 1989) called for a "happy medium" in educational technology research, to be achieved by balancing high rigor of studies (internal validity) with relevance to real-world applications (external validity). In this paper, we argue that, although contemporary research orientations have made substantial strides in capturing these two features, success in combining them and achieving the happy medium envisioned remains limited and disappointing. Highly prevalent today are (a) "technology effects studies," which are strong in rigor but continue to view educational technology (ET) as a "treatment" rather than as a mode for delivering and potentially enhancing treatments; and (b) "surface learning studies," which examine processes and outcomes of realistic ET applications, but often without including meaningful measures of student learning. To promote studies that more successfully bridge research and practice, we present suggestions and positive examples for finally achieving a happy medium in this "Phase II" quest.