←
Steve Haroz 📊👁️🧠19,916 posts
Steve Haroz 📊👁️🧠 avatar

Steve Haroz 📊👁️🧠

@sharoz

Visual perception, cognition, and data vis scientist (he) 🦋 @steveharoz.com R guide: r-guide.steveharoz.com StatCheck: statcheck.steveharoz.com

◷ Joined March 2008▧ 2,242 Photos and videos
Steve Haroz 📊👁️🧠 avatar
⟳ Steve Haroz 📊👁️🧠 retweeted
Steve Haroz 📊👁️🧠@sharoz·7 Jun 2024

BlueSky June! Many people who have a BlueSky account still prioritize posting, reading, and asking questions on Twitter. So for the rest of June, I ask you to commit to using bsky.app before the mess called X. Please retweet and tag people! 1/

Mattan S. Ben-Shachar avatar
⟳ Steve Haroz 📊👁️🧠 retweeted
Mattan S. Ben-Shachar@mattansb·7 Jun 2024

Let's do it!

BlueSky June! Many people who have a BlueSky account still prioritize posting, reading, and asking questions on Twitter. So for the rest of June, I ask you to commit to using bsky.app before the mess called X. Please retweet and tag people! 1/

Steve Haroz 📊👁️🧠 avatar
Steve Haroz 📊👁️🧠@sharoz·7 Jun 2024

If you want to crosspost to get a broader reach, go for it! But I'm asking you to post to BlueSky first to break the habit of thinking of twitter as your primary or first place to go. For June, start with BlueSky. Here's a starter guide: blueskyscience.steveharoz.co… 3/3

Steve Haroz 📊👁️🧠 avatar
⟳ Steve Haroz 📊👁️🧠 retweeted
Steve Haroz 📊👁️🧠@sharoz·14 Mar 2024

BLOG POST: Invalid Conclusions Built on Statistical errors With a small p-value or a large separation in confidence intervals, you may assume that an effect is reliable. But overly simple reporting can hide serious conclusion-altering errors. steveharoz.com/blog/2024/wro… 1/🧵

Post attachment
ᴅʀ ᴍɪʀᴄᴇᴀ ᴢʟᴏᴛᴇᴀɴᴜ 🍂 🍃 avatar
⟳ Steve Haroz 📊👁️🧠 retweeted
ᴅʀ ᴍɪʀᴄᴇᴀ ᴢʟᴏᴛᴇᴀɴᴜ 🍂 🍃@mzloteanu·15 Mar 2024

#statstab #50 (!) Simulating how replicate trial count impacts Cohen’s d effect size Thoughts: Not ready to give up std effects, but the caveats list for them keeps growing. P.S. 50 stats tabs down! 🎉 My RAM is recovering. #effectsize #misinformation steveharoz.com/blog/2023/sim…

Steve Haroz 📊👁️🧠 avatar
Steve Haroz 📊👁️🧠@sharoz·14 Mar 2024

Long-term improvements: * More stats education in applied fields. * Reporting standards that make errors easier to spot. * Mandatory open data+code (or an explicit reason why it can't be shared). Publication venues that don't mandate open practices don't warrant credibility. 7/7

Steve Haroz 📊👁️🧠 avatar
Steve Haroz 📊👁️🧠@sharoz·14 Mar 2024

Some red flags for spotting sketchy results: 🚩 Incomplete reporting. P-values aren't enough. Look for a full APA report F(_,_)=_, p=_, 95%CI ES=_. 🚩 Unreported model. 🚩 Hidden data or code. 🚩 Claims about an analysis approach being immune from common statistical concerns. 6/

Steve Haroz 📊👁️🧠 avatar
Steve Haroz 📊👁️🧠@sharoz·14 Mar 2024

False negative: Experiment design lacks sensitivity Experiment design can go wrong too. When baselines vary, a between-subject experiment may not find a reliable effect. Here, an experiment with 200 subjects fails where a more sensitive experiment with 5 subjects succeeds. 5/

Post attachment