Beth Tipton

Professor of Statistics and Data Science at @NorthwesternU, fellow at @IPRatNU. meta-analysis, causal generalization. @statstipton.bsky.social

[Institute for Policy Research at Northwestern @IPRatNU]

In a move to "maximize efficiency," the Trump administration canceled $1 billion in education research contracts. @stats_tipton and her colleagues argue that without this research, policymakers won’t know what works—ultimately reducing efficiency. https://t.co/3emUWTYK51

[Beth Tipton @stats_tipton]

@psybalazs Not a good idea. See here for a reference to help! https://t.co/yAGrYywh1j

[Beth Tipton @stats_tipton]

@JasonChow Hahahahahahaha - I mean, it is in Portland :)

[Beth Tipton @stats_tipton]

@DA_Houston Thank you!! This is a great chapter.

[Beth Tipton @stats_tipton]

@robvargas21 CONGRATULATIONS!

[Beth Tipton @stats_tipton]

@chrisweiss_175 Yes please!

[Beth Tipton @stats_tipton]

@jd_wilko Not recent, but we did a review in this paper that may be relevant. https://t.co/uPUe2R2oPo

[Beth Tipton @stats_tipton]

@karlrohe 100% agree. It's how we teach our intro stats - 5 weeks of data analysis + 5 weeks of statistics. Works WAY better and students leave actually being able to work with data AND better off than students who came out of AP Stats.

[Beth Tipton @stats_tipton]

@NeuroStats @AlexKale17 Oh this is great - thank you!

[Beth Tipton @stats_tipton]

@npparikh I love this paper!

[Beth Tipton @stats_tipton]

@NeuroStats @AlexKale17 e.g., is experimental design part of DS? regression? logistic regression? is it only DS if there is fancy computation involved?

[Beth Tipton @stats_tipton]

@NeuroStats @AlexKale17 100% agree. As institutions become interested in DS/AI, I've noticed that there is a desire to define/separate out the parts of statistics that are / are not DS. I'm genuinely confused though about where the demarcation is, other than "uses ML" or "big data".

[Beth Tipton @stats_tipton]

@karlrohe Let's meet up to chat while you're here!

[Beth Tipton @stats_tipton]

@karlrohe Ooh, this is interesting - what's the source?

[Beth Tipton @stats_tipton]

@noah_greifer Yeah, there is a part of stats that is closer to math - but ultimately the theory is still related to data / estimation / testing. But that theory is the foundation upon which applied methods (with data) are built.

[Beth Tipton @stats_tipton]

@matloff To be clear, I don't think DS = STAT. I hypothesize that its more like STAT \subsetof DS.

[Beth Tipton @stats_tipton]

The same problem occurs in medicine and public health - but the answer there is to have collaborative teams with biostatisticians on every paper.

[Beth Tipton @stats_tipton]

I've advised students struggling with this. The easy answer - videos, workshops - don't provide deep knowledge. The harder answer - courses - often require foundational courses that are not immediately applicable.

[Beth Tipton @stats_tipton]

@garjoh_canuck Clearly the foundation for all that I do :) 😜 (Also: Hi!!! So good to hear from you!)

[Beth Tipton @stats_tipton]

@SLKoole The paper conveniently doesn't estimate the degree of heterogeneity in effect sizes, though it shows there are significant Q-statistics (indicating there is heterogeneity). I'd be careful with interpretation - that effect size is the average of a lot of heterogeneous data.