Twitter | Search | |
Mike Kearney📊
Assistant Professor at and MU Informatics Institute. Interested in new media, partisanship, data journalism, data[["big"]] science, and .
5,232
Tweets
1,433
Following
4,320
Followers
Tweets
Mike Kearney📊 12m
Replying to @LucyStats
It's interesting to see how people deal with the ambiguity. I think asking about *p-values* on *Twitter* may have created a perfect [and very entertaining] storm.
Reply Retweet Like
Mike Kearney📊 1h
Replying to @kearneymw
(Because as sample sizes increase, p values decrease)
Reply Retweet Like
Mike Kearney📊 1h
Replying to @kearneymw
So if you're reporting an "insignificant" p value but still believe in the theory, then you can always say "it would have been significant with a larger sample size" (which is why p values are a little silly)
Reply Retweet Like
Mike Kearney📊 2h
If the only difference you know of is sample size, and if p values are a function of sample size, then the p value associated with the smaller sample size would be more 'impressive'
Reply Retweet Like
Mike Kearney📊 7h
Replying to @LucyStats
🤦🏻‍♂️🤦🏻‍♂️🤦🏻‍♂️
Reply Retweet Like
Mike Kearney📊 retweeted
Pew Research Methods 22h
A topic model is a type of algorithm that “reads” a set of documents, examines how words and phrases co-occur, and automatically finds clusters of words - aka topics. Learn more:
Reply Retweet Like
Mike Kearney📊 retweeted
Scott Came 10h
Another look at the Russian IRA tweet dataset: Cumulative Tweets vs Followers by account type each day, animated with .
Reply Retweet Like
Mike Kearney📊 retweeted
mikefc Aug 13
Reply Retweet Like
Mike Kearney📊 retweeted
Garrick Aden-Buie 11h
Getting ready to teach dplyr joins to new users tomorrow, so naturally I productively procrastinated by getting to know the new gganimate. It is the coolest!
Reply Retweet Like
Mike Kearney📊 retweeted
Luis D. Verde Aug 13
code to animate common data munging approaches
Reply Retweet Like
Mike Kearney📊 retweeted
Patrick Miller 10h
I hate to deflate , but Colyer voters aren’t moderates. Nor is Colyer. They’re largely conservatives. They have more in common with Kobach than either Orman or Kelly. Why would they not heavily vote for Kobach? Most mod Rs were likely not voting for Kobach anyway.
Reply Retweet Like
Mike Kearney📊 13h
Replying to @petemohanty
👍🎉💥🎊👍
Reply Retweet Like
Mike Kearney📊 retweeted
Pete Mohanty 17h
now on CRAN: kerasformula 1.5.0. Build neural nets in with as little as one line of code now with greatly improved support for advanced hyperparameters to prevent overfitting, embed text, and more!
Reply Retweet Like
Mike Kearney📊 14h
Replying to @MaartenvSmeden
You can explain your logic...e.g., if MCAR then listwise, if MAR then multiple imputation, and if MNAR then we will hope to retrofit this as a methods/learning paper.
Reply Retweet Like
Mike Kearney📊 retweeted
Pablo Barberá 22h
Back home after two intense weeks in Budapest teaching at the ECPR Summer School in Methods and Techniques! All my teaching materials are now publicly available on my website. Here's a quick overview: 👇
Reply Retweet Like
Mike Kearney📊 20h
When you get captcha'd by google for looking up whether "never-mind" is one or two words
Reply Retweet Like
Mike Kearney📊 Aug 14
I have a feeling Donald Trump is not very nice to dogs...
Reply Retweet Like
Mike Kearney📊 Aug 13
Yep. You'd have to find an emoji/Unicode library you like but rtweet returns that data!
Reply Retweet Like
Mike Kearney📊 retweeted
Thomas Mock Aug 13
The welcomes you to week 20 of ! We're exploring Russian Troll data on their GitHub, please see thread for more details! Data: Article:
Reply Retweet Like
Mike Kearney📊 retweeted
David Robinson Aug 13
Great blog from , scraping and analyzing words in song lyrics over time. Looks like "baby" peaked in the 90s and "love" is back in style 🎵 
Reply Retweet Like