Some terrific material in @MDRC_News's top 15 releases in 2022! https://t.co/5zDYN1OzJx The list includes 3 short pieces that reflect on the work we've done in our center; see links below 👇

#ICYMI #PredictiveAnalytics We continue to find that simpler methods are "good enough" for prediction problems https://t.co/GyYJnq1Kww

Learn more about using the Power Under Multiplicity (PUMP) R package and app from our new blog post! https://t.co/0uxsHWZ1jo - by @keporter_ca, Zarni Htet, Kristen Hunter, @LMiratrix

Researchers often test the effectiveness of an intervention on multiple outcomes, but this strategy can also substantially reduce the probability of detecting effects when they truly exist. Here are two open-source software tools that can help.

A new brief by Shawna Anderson and @donmarwhar describes how human-centered, operational data science supports continuous program improvement in our ongoing research-practice partnership with @PerScholas.

Grateful & excited for this @IESResearch award for @CdiPorter and our team to investigate the value and limitations of #MachineLearning to detect the presence of heterogeneous subgroup impacts in #RCTs https://t.co/xArw2pGLda

ICYMI! Join our team and make a difference using data science! Multiple positions open 👇 Research Assistant: https://t.co/cVRPJy1Be1 Technical Research Analyst: https://t.co/DcEVJG2W13 Technical Research Associate:

There is no silver bullet for reducing poverty, but “microloans give options. They are another tool in our anti-poverty tool box,” said Richard Hendra, director of @MDRC_News. “For this population that we studied, they were very effective.” Via @CityLab: https://t.co/h8M3V83Uwb

As a @ceoworks alumni, I was thrilled to reconnect with Ahmed Whitt at #fblsummit and hear the results of CEO's data analysis on participant feedback evaluated by @mdrc. Bottom line: feedback practices drive outcomes w. response rate over 50%. Kudos! @feedbacklabs @4SharedInsight

.@CDI_MDRC has collaborated w/ @PerScholas & @ceoworks to place program participants’ lived expertise at the center of #DataScience. This brief by MDRC's @edith_yang & @alissa_stover offers lessons learned from these partnerships.

It always feels like a triumph to get CRAN to post your package. And @stats_hunter, @CdiPorter and I just got PUMP (a package for calculating power, including multiple testing) out there! This is my most "this is real software" package yet.

In addition to our app, watch for our #Rstats package on CRAN coming soon & read about our methods on https://t.co/igcY3XbupL: https://t.co/sXA6RQByXs. Below is an example figure than can be produced with the #Rstats package.

Check out our #Rshiny app! Work by @CdiPorter, Zarni Htet at @MDRC_News @CDI_MDRC and @LMiratrix, @stats_hunter at Harvard.

PUMP also allows researchers to consider a variety of definitions of statistical power, allowing for more flexible definitions of success.

Why should you use PUMP? Adjustments for multiple testing reduce the likelihood of spurious findings in RCTs but also change power. This is frequently ignored in power analyses. Existing tools don’t easily accommodate the use of adjustment procedures.

PUMP estimates power, minimum detectable effect size & sample size when adjusting for multiple hypothesis tests in multi-level RCTs.

The #IESfunded Power Under Multiplicity Project (PUMP) introduces new software tools for estimating power. This project is a collaboration between researchers at @MDRC_News & Harvard and is funded by @IESResearch.

ICYMI - CDI director @hendr239 presented on the TANF Data Innovation project and using data to improve outcomes at @data_foundation's Data Symposium 2021. You can watch Henrda's presentation "Working Together to Use Data to Improve Outcomes" here https://t.co/yBAruwKN2c #datalive

CDI works across policy areas at @MDRC_News. Check out this #brief on Balancing Promise and Caution in Pretrial Risk Assessments by CDI deputy director @CdiPorter, @cindyredcross & @LMiratrix -- a product of CDI's work with @CTE_MDRC

Welcome to the MDRC (@MDRC_News) Center for Data Insights (CDI) Twitter! CDI helps our partners improve their programs and systems by harnessing the benefits of operational data science techniques to produce actionable insights. #FirstTweet #DataScience