Mentions
Extended reach
Total likes
Video views
AI synopsis
The discussions reflect a broad recognition that the article clarifies misconceptions about the normality assumption, exemplified by references to the robustness of the t-test and linear regression in large datasets. Participants use technical terminology and cite relevant literature, illustrating a deep understanding of statistical principles and emphasizing the article's potential to impact research methodology standards.
Most discussions acknowledge that the article corrects common misconceptions about the normality assumption, supporting agreement with its message.
Participants show moderate interest by referencing related methods and statistical principles, indicating engagement but not deep focus on the publication's specifics.
Several comments delve into statistical details, citing specific papers and theoretical frameworks, demonstrating a strong level of engagement.
The discussions highlight the importance of understanding large-sample properties, suggesting the publication can influence statistical practices and education in public health research.
And if you are looking for a paper to cite if / when a reviewer says you need to use the shapiro-wilk test then look no further than here. https://t.co/BDJsGZttOP
@SolomonKurz I think other people gave good answers (exact inference with small n; asymptotic efficiency as n goes to infinity), so I'll just add that @tslumley et al have a helpful paper on the (un)importance of normality (in case no one's mentioned it) https://t.co/UNTVzkKY2x
@ronbun_factory 「正規分布してないからt検定だめ、線形回帰だめとかうっせぇ！」 https://t.co/ywCCHXGmFz
We consciously wrote pubmed.ncbi.nlm.nih.gov/11910059/ to be used as a reply to reviewers
I always thought data needs to be normally distributed (or close) for t-test and linear regression to work. I am wrong: https://t.co/eMuQ9MXcOW Per the central limit theorem, normality is not required for large samples. Any idea about the effect of the extremely large values?
Also a good opportunity to re-up this excellent article that should be required reading on many more syllabi than it is https://t.co/tA8yK0gcyp
The normality of residuals is a key assumption in regression analysis. This video explains how to test for normality using histogram, QQPlot, PPPlot, and statistical tests like Kolmogorov-Smirnov and Shapiro-Wilk. Large…
How are linear models actually fit? Assumptions and diagnostics are discussed, emphasizing that in large datasets, methods like t-tests and regression are robust despite non-normal data, guiding appropriate statistical…
We consciously wrote pubmed.ncbi.nlm.nih.gov/11910059/ to be used as a reply to reviewers
RT @linstonwin: @SolomonKurz I think other people gave good answers (exact inference with small n; asymptotic efficiency as n goes to infin…
RT @linstonwin: @SolomonKurz I think other people gave good answers (exact inference with small n; asymptotic efficiency as n goes to infin…
@FloriSoPoliSci My main concern with OLS for binary outcomes in surveys is power - for the same effect size, you need a larger N than logistic But if you're using a large federal survey, that's usually not an issue. Good discussion here: https://t.co/tA8yK0gcyp
RT @linstonwin: @SolomonKurz I think other people gave good answers (exact inference with small n; asymptotic efficiency as n goes to infin…
RT @linstonwin: @SolomonKurz I think other people gave good answers (exact inference with small n; asymptotic efficiency as n goes to infin…
@SolomonKurz I think other people gave good answers (exact inference with small n; asymptotic efficiency as n goes to infinity), so I'll just add that @tslumley et al have a helpful paper on the (un)importance of normality (in case no one's mentioned it) https://t.co/UNTVzkKY2x
Struggeling with non-parametric statistics because your data do not follow a normal distribution? If you work with large data sets (N>30), this 2002-paper might interest you: https://t.co/Cwl5zBVeOc
RT @genkuroki: https://t.co/itY7274yO2 【It is widely but incorrectly believed that the t-test and linear regression are valid only for Norm…
https://t.co/itY7274yO2 【It is widely but incorrectly believed that the t-test and linear regression are valid only for Normally distributed outcomes. 略】 t検定や線形回帰が正規分布にのみ使えるという間違っている考え方が広く信じられていることは問題である。
RT @masonfidino: And if you are looking for a paper to cite if / when a reviewer says you need to use the shapiro-wilk test then look no fu…
RT @masonfidino: And if you are looking for a paper to cite if / when a reviewer says you need to use the shapiro-wilk test then look no fu…
RT @masonfidino: And if you are looking for a paper to cite if / when a reviewer says you need to use the shapiro-wilk test then look no fu…
RT @masonfidino: And if you are looking for a paper to cite if / when a reviewer says you need to use the shapiro-wilk test then look no fu…
RT @masonfidino: And if you are looking for a paper to cite if / when a reviewer says you need to use the shapiro-wilk test then look no fu…
RT @masonfidino: And if you are looking for a paper to cite if / when a reviewer says you need to use the shapiro-wilk test then look no fu…
RT @masonfidino: And if you are looking for a paper to cite if / when a reviewer says you need to use the shapiro-wilk test then look no fu…
And if you are looking for a paper to cite if / when a reviewer says you need to use the shapiro-wilk test then look no further than here. https://t.co/BDJsGZttOP
今更「指数分布めいてる時ですら500あればt検定でいい」的な文献見つけた。 https://t.co/FE1KOMIHJc
RT @AssumeNormality: @YFeyman @XuanziQ @michalhorny Gotta invoke Lumley next time https://t.co/tA8yK0gcyp
@YFeyman @XuanziQ @michalhorny Gotta invoke Lumley next time https://t.co/tA8yK0gcyp
RT @AssumeNormality: Also a good opportunity to re-up this excellent article that should be required reading on many more syllabi than it i…
Also a good opportunity to re-up this excellent article that should be required reading on many more syllabi than it is https://t.co/tA8yK0gcyp
@ClosedCurves @SolomonKurz @ChelseaParlett More/better discussion of this sort of thing: https://t.co/KQZp1xOfoJ
@evans1_d @michelnivard Mostly that's what they have been taught, I have seen protocols like :rntrasnform trait -> regress out covariate effects -> rntrasnform residuals -> regress out kinship -> rntrasnform residuals ->GWAS I do not think many understand how small "large" is. https://t.co/U4IkVS0jjf
Citation: https://t.co/12FoASSHmU.
Reference here: https://t.co/NS92P2LDYW
RT @koro485: @ronbun_factory 「正規分布してないからt検定だめ、線形回帰だめとかうっせぇ！」 https://t.co/ywCCHXGmFz
RT @koro485: @ronbun_factory 「正規分布してないからt検定だめ、線形回帰だめとかうっせぇ！」 https://t.co/ywCCHXGmFz
RT @koro485: @ronbun_factory 「正規分布してないからt検定だめ、線形回帰だめとかうっせぇ！」 https://t.co/ywCCHXGmFz
RT @koro485: @ronbun_factory 「正規分布してないからt検定だめ、線形回帰だめとかうっせぇ！」 https://t.co/ywCCHXGmFz
@ronbun_factory 「正規分布してないからt検定だめ、線形回帰だめとかうっせぇ！」 https://t.co/ywCCHXGmFz
RT @davidandacat: @AedinCulhane @dfcidatascience I'm not familiar with what estimands are typically of interest in scRNAseq experiments, bu…
@AedinCulhane @dfcidatascience I'm not familiar with what estimands are typically of interest in scRNAseq experiments, but it's worth noting that normality is not actually necessary for linear regression to be valid: https://t.co/KQZp1xOfoJ
RT @StevenXGe: I always thought data needs to be normally distributed (or close) for t-test and linear regression to work. I am wrong: htt…
I always thought data needs to be normally distributed (or close) for t-test and linear regression to work. I am wrong: https://t.co/eMuQ9MXcOW Per the central limit theorem, normality is not required for large samples. Any idea about the effect of the extremely large values?
MP
WL
K
SG
KG
M
SW
JC
MH
DB
黒K
MS
RR
JJ
L
DA
ⱰY
KŨ
平尚
XQ
KG
DC
NP
KI
コ
M
นม
MT