- Видео 8
- Просмотров 68 649
Jonathan Bartlett
Великобритания
Добавлен 9 окт 2011
Videos related to statistics, with a particular focus on R and Stata.
Reference based multiple imputation for trials - what's the right variance and how to estimate it?
This short talk is about referenced based multiple imputation methods for missing data in clinical trials.
Просмотров: 455
Видео
Bootstrap Inference for Multiple Imputation Under Uncongeniality and Misspeci fication
Просмотров 4714 года назад
This is a recording of a talk given at the International Society for Clinical Biostatistics conference 2020. It briefly discusses the content of this paper: doi.org/10.1177/0962280220932189
Missing category method for missing values
Просмотров 4084 года назад
This video is about the missing category method for handling missing values in categorical/factor variables.
Confidence interval interpretation - Statistics by Simulation in R
Просмотров 3,4 тыс.5 лет назад
Using simulation in R to explore the interpretation of frequentist confidence intervals for a mean.
The Central Limit Theorem - Statistics by Simulation in R
Просмотров 8 тыс.5 лет назад
In this video I use simulation in R to demonstrate the Central Limit Theorem, whose result is that the sample mean is normally distributed in large samples even if the underlying distribution is not normal.
Standard deviation versus standard error - Statistics by Simulation in R
Просмотров 2,3 тыс.5 лет назад
In this video I use a small simulation in R to illustrate the difference between standard deviation and standard error.
Using Stata's sem (structural equation modelling) builder to adjust for covariate measurement error
Просмотров 4,8 тыс.10 лет назад
In this video I look at how Stata's sem (structural equation modelling) builder can be used to adjust for covariate measurement error.
Hosmer-Lemeshow goodness of fit test in R
Просмотров 49 тыс.10 лет назад
A short video showing how the Hosmer-Lemeshow goodness of fit test for logistic regression can be performed in R. For more on this topic, see thestatsgeek.com/2014/02/16/the-hosmer-lemeshow-goodness-of-fit-test-for-logistic-regression/
This is extremely well-explained, thank you. If the missing category overwhelmingly consisted of females in your example, would this mitigate this effect on X? And can the impact of including a missing category be identified by examining the estimates from the missing category and seeing if they are similar to any of the other non-missing categories?
Legend
great explaination. Here is the code explained in the lecture: nSimulations = 100000 nSample = 100 Y = numeric(nSimulations) for(i in 1:nSimulations){ X = runif(n); Y[i] = mean(X) } sigma = sd(Y) mu = mean(Y) hist((Y-mu)/sigma, breaks = 50) Here, I have scaled and shifted the random variable to make it the standard normal. Also, you can replace runif() with any other distribution of your choice.
Hi, the Hosmer-lemeshow works in logistic regression, but what about poisson regression( binary dependant variable in cross-sectional studies) what options for goodness-of-fit do we have?
You do not need to multiply logicals by 1 in R to make them integers.
Excellent explanation. Thanks!
Simple script, clear explanations!
You cleared many doubts, thank you!
Merci beau gosse
Great video, thank you. Simple, easy to follow and straight to the point!
Thank you sir.
good morning.sir. I have a question:what is y1 and y0 presentation?
I love you thanks for doing this!!!
This is pure gold. You sir are a diamond in the rough
hello, how do you move around the lines? I was keep pressing shift+enter to go below the line without producing something..
cntrl+enter or cmd+enter
if i have a smaller p value than alpha score 0.05. what should i do if my model are not fit well?
how to do this test by eviews 8?
Hey, in the line > hl <- hoslem.test(mod$y, fitted(mode), g=10) it should not be "mydata$y" instead of "mod$y"??
I can only get it to work using this solutuion so I think you could be right?
Hello! Thank you for this video. But I have a question. The X-squared and p-value of Hosmer-Lemeshow test in R and SPSS is very different. Why?
can be by choosing the cut point
hey i tried lot but following errors occure > hoslem.test(result,phat,g=10) Error in hoslem.test(result, phat, g = 10) : object 'phat' not found > hoslem.test(result,fit,g=10) Error in hoslem.test(result, fit, g = 10) : object 'result' not found > hoslem.test(diagnosis,fit,g=10) Error in rank(x, ties.method = "min", na.last = "keep") : unimplemented type 'list' in 'greater' > hoslem.test(diagnosis,fit,g=8) Error in rank(x, ties.method = "min", na.last = "keep") : unimplemented type 'list' in 'greater' > hoslem.test(data,fit,g=8) Error in rank(x, ties.method = "min", na.last = "keep") : unimplemented type 'list' in 'greater'
Beautifully explained! Though i listened to it like 20 times. Thank you very much!
thank you for this video ... this has been really helpful
when 'm trying find Hosmer-Lemeshow tests this appear am using mutivariate logistic Error in model.frame.default(formula = cbind(y0 = 1 - y, y1 = y) ~ cutyhat) : variable lengths differ (found for 'cutyhat') can u help me if possible thanks in advance
+Balaji leon use hoslem.test(dataname$y,fitted(model))
+CH D doesn't help really :( it returns chi-square = NaN and p-value: NaN
+Balaji leon Try to decrease the number of groups (g) . Instead of using 10 use 6.
Super useful. Thanks!
thanks for you