Top Tweets for #StatWisdom
A link funtion in generalized linear models is like a transformation, but it is applied to the mean of Y, not each value of Y. #StatWisdom
Every time you reread your stat software manual, you'll learn something new. #StatWisdom
Being an outlier is not a good reason to delete a datum. First figure out what led to being an outlier.#StatWisdom
AIC and BIC are based on -2LL of the model. All are measures of model fit, but AIC and BIC penalize for model complexity. #StatWisdom
In a linear model, Y does not have to be normally distributed. The errors do. #StatWisdom
Linear model does not mean a linear relationship between Y and Xs. It means a linear form between Y and betas. #StatWisdom
When you treat an ordinal variable as numerical, you're making big, big assumptions that intervals between values are equal. #StatWisdom
A large sample size alone doesn't make a better sample. Representativeness matters. #StatWisdom
Some outliers are measurement errors. Some are sample errors. Some are genuine data points. Don't delete them by default. #StatWisdom
One indication of multicollinearity is large changes in coefficients when adding new predictors to the model. #StatWisdom
Repeated measures ANOVA drops cases with any missing outcomes. Mixed Models don't. #StatWisdom
Probability, odds, and log-odds all measure the propensity of one outcome to occur, but on different scales. #StatWisdom
Transformations of Y affect linearity, normality, and homoskedacticity. Solving one can cause problems in the others. #StatWisdom
Identification in SEM indicates there is enough data to estimate all model parameters. Underidentified models can't be fit. #StatWisdom
Proportional odds is often an assumption in ordinal logistic regression. Check it. #StatWisdom
The coefficient of multiple correlation measures the correlation of a set of variables with a single variable. #StatWisdom
GLM requires repeated measures data in the wide format. Mixed requires it in long, or stacked format. #StatWisdom
Negative binomial models are one way to account for overdispersion in Poisson models. #StatWisdom
Good power calculations are based not on the effect size you expect or someone else found, but the smallest meaningful effect. #StatWisdom
A link funtion in generalized linear models is like a transformation, but it is applied to the mean of Y, not each value of Y. #StatWisdom
Most Popular Users
Elon Musk 
@elonmusk
241.5M followers
Barack Obama
@barackobama
118.9M followers
Cristiano Ronaldo
@cristiano
114.6M followers
Donald J. Trump
@realdonaldtrump
111.8M followers
Narendra Modi
@narendramodi
107.1M followers
Rihanna
@rihanna
98.7M followers
NASA
@nasa
92.3M followers
Justin Bieber
@justinbieber
91.8M followers
KATY PERRY
@katyperry
90M followers
Taylor Swift
@taylorswift13
84M followers
Lady Gaga
@ladygaga
75.5M followers
Virat Kohli
@imvkohli
73.5M followers
Kim Kardashian
@kimkardashian
70.9M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
66.5M followers
Bill Gates
@billgates
65.2M followers
Selena Gomez
@selenagomez
63.1M followers
The Ellen Show
@theellenshow
62.2M followers
CNN
@cnn
61.8M followers
X
@x
60.7M followers