Interesting question, re: the level of the math! That's not something we've explicitly studied. In this study, we worked with teachers teaching math in grades 3-8 (treatment and control teachers were in the same grade range), but it wasn't part of our experiment to try to compare across grade levels. In terms of some general differences between more and less coaching processes, the literature suggests that effective coaching programs have a few key features that set them apart from less effective ones:  

1. the coaching is focused on classroom observation and feedback over time,
2. the interactions between teachers and coaches are of a high quality, and
3. the coaches are highly trained, monitored, and supported.

In practice, this is not characteristic of all coaching programs. In fact, we hear from many coaches in the field who were great math teachers who were promoted to a coaching role, then were give no training, structure, or support to facilitate high-quality work with other teachers. (It was hearing from many coaches in this position that drove our initial transition from research to practice; there was a demand for training and support for coaches in the field.)   In terms of your second question, district funding and teacher time are certainly always a challenge!  Schools and districts spend an enormous amount of time and money on teacher professional development every year, and the literature suggests that much of it does not change teacher practice or student achievement for the better. In general, coaching appears to be a bright spot in the larger world of professional development, when the key features above are present. Of course, just about everywhere we've worked, teachers and coaches have reported that one of the most challenging aspects of implementing a rigorous coaching program is finding all the time. Still, in many places, folks are making it work. Anecdotally, it seems like the places that make it work are schools and districts with good alignment and support across all levels: administrators, teachers, and other instructional leaders (such as coaches, specialists, etc.). When everyone is on the same page and working toward a common goal and focused on the same priorities, finding time becomes far more possible. However, when there are competing initiatives or inconsistent priorities, it's really, really hard.     Thanks for the thoughtful questions, and let us know if we can clarify anything else about our work!   -The MQI Coaching team

Folks, I really appreciated the summary of what works in effective coaching programs. It makes sense, but it is great to have some firm guidelines. The importance of buy-in across the levels of a school or district is I suspect highly relevant across all educational initiatives - it does take a village with consistent and agreed upon goals. Thanks for all your work on this.

Great questions, Dave. In this study, teachers and coaches started working together in the mid/late fall, and the suggested/ideal timeline was to meet with their coach every other week. Of course, various conflicts get in the way, and very few folks were able to stick to the suggested biweekly schedule. Most teacher-coach pairs met 8-12 times over the course of the school year (with a max of 17). You can see a participation graph (and other research results) here: https://mqicoaching.cepr.harvard.edu/mqi-research-study.

In this study and in our work since, we have had many teachers express interest in continuing to work with their coach for a second year, but so far this is not something we’ve studied. We were very happy to see lasting effects on instruction after just one year of coaching, but most teachers don’t work on all the possible MQI codes in one year, so there’s plenty more potential if folks were to continue. It’s certainly something we’re looking into studying in the future (likely with our “train the trainer” model this time). As to your question about Errors & Imprecision, we have a couple of thoughts.

First, the overall rate of errors (which we count when they are by or endorsed by the teacher and uncorrected) across all lessons in this study were fairly low in both treatment and control groups. Second, our intervention doesn’t directly teach teachers more math content- we work on generalizable changes to instruction (e.g., students providing explanations, or teacher or students explicitly linking between different representations).

We only work on the Errors dimension on an as-needed basis if something comes up, so most teachers didn’t focus any of their time there. So, it’s not surprising that we didn’t see growth. This is also something that we’re looking into studying further, though- we’d love to add some more work on content knowledge alongside the MQI-focused instructional coaching to see if we can also address the content knowledge piece in a way that’s still guided by the MQI and aligned with the existing coaching program. Stay tuned!