Screening, Descriptives in R, and Asking Questions

This summer has been FULL of learning experiences and I'm guessing that there are plenty more to come.  But for now, here are a few key points from this summer:

1. Two-Stage Screening. Why don't we do this with all universal screening? The big argument FOR universal screening is that we're missing way too many kids who have speech, language, and/or reading problems in schools. They fall through the cracks until it's "too late" (it's never too late, but remediation is more difficult as kids get older). Opponents, however, point out that universal screening is a huge time investment that isn't reasonable in most schools - you can't screen for everything and it results in a bunch of kids being over-identified. Over-identification is bad because it takes time and resources away from the kids and service providers who really need them. 

• Potential solution = use a high-sensitivity screener with everyone that takes next-to-no-time. It could even be a parent/teacher report measure. The key is that you want to kick out as many high-achieving kids as possible. This gives you a higher prevalence rate for whatever you're trying to identify in the second sample... allowing you to use a second-stage screener that takes slightly more time but that really lets you tease apart the typically-developing kids from the ones who need more support. This can save time, money, and reduce the number of kids we're missing. 
• Unfortunately... our screeners aren't currently optimized to function in this framework. We need to work on that. Maybe there are some grant ideas within that...

2. I love Twitter. Recently, I ran across a Twitter conversation between some colleagues talking about how hard it is to get nicely-printed output from descriptives by group in R (at least compared to other programs, including Excel and SPSS). Conveniently, one of them shared some great code for easy, pretty-looking descriptives by group that can be shared with coauthors. One of these days I'll do an example and share it but for now here's a link to that thread.  Credit to Ashley Edwards for the R Code! Also - check out her github account for this and more resources!

ALSO - because of this Twitter conversation, I was referred to this awesome thread as well. Dave Braze on making pretty tables here. 

3. There are a lot of people out there who have way more knowledge and experience than me in conducting educational research. Many of them are more than happy to help the rest of us when we get stuck (me all the time). The challenge is really in connecting those people - the ones who want to be helpful to the ones who need the help. Twitter helps with this somewhat. There are also other options out there like the Meta-Analysis Training Institute (PI = Terri Pigott. Team includes Tasha Beretvas, Beth Tipton, Josh Polanin, and Ryan Williams), which can connect interested researchers with each other. Obviously also having amazing colleagues also helps, but that's not the case for everyone. I'm interested in figuring out more ways that we can connect people - everyone has unique strengths (and needs) and there is SO much work to be done that it seems silly not to work together.