Це відео не доступне.
Перепрошуємо.
Multinomial logistic regression using Stata: Covid-19 threat perceptions from Pew data (revised)
Вставка
- Опубліковано 4 кві 2020
- This video provides a demonstration of how to perform multinomial logistic regression using Stata. The data comes from the Pew Research Center (www.pewresearch.org/) and reflects survey responses concerning participants' perceptions of the Covid-19 crisis. In the demonstration, we focus on testing predictors on people's perceptions of Covid-19 as a threat to personal health.
You can obtain a copy of the Stata data file at this link: drive.google.com/file/d/11s1a...
You can download a copy of the Powerpoint referenced in the video here: drive.google.com/file/d/15kGp...
Hi, Dr. Mike. I always look forward to your Stata videos. Great work!
Thank you, Professor. It was a very constructive lecture, I am glad to find these type of contents on the Internet.
Big fan of you. The lacture video is really great. Very clear and concise. It shows your efforts.
Thank you, I've been looking for a demo of multinomial logit in Stata!
This is the explanation I needed to finish my research! Thanks heaps!!
Thank you for writing out the interpretations!
Thanks for the video professor, such a simplistic and lucid way of explanation.Loved it.
This was a really thorough video and a great PowerPoint. Thank you. And oh my goodness, thank you for citing my 2009 article. That was very kind of you.
Thank YOU for that article. It was very helpful to me when I was learning about the topic. Cheers!
@@mikecrowson2462 And now your work is helping me further my work into multilevel multinomial logistic regression, so thank you. I'm a big fan of many, many of your videos.
Thanks Mr. Crowson! This is a really good detail video that explains each concept really good!
Great work, nice practical example and interpretations.
Your explanation is very clear and concise. Thank you for posting this video! This video helps me a lot!
I will just say excellent. Marvellous explanation. Thank you
I deeply grateful to you as a PhD candidate since you explain the statistical processes very clearly. Your contribution is huge for my professional development. Thank you very much. Greetings from Turkey!
Hi there, thank you for your kind words! Good luck to you as you finish out your program! Cheers!
Thank you for this video and sharing your slides.
This was extremely helpful. Thank you prof.
Thank you. As a PhD student, this was very helpful.
Glad it was helpful! Thank you for visiting!
Thank you very much for this video!
This was very helpful! Very clear and concise!
Glad it was helpful, Isaac! Best wishes!
Thank you for making this video, very helpful!
You are very welcome! Best wishes!
Thanks for the great video. Do you have any experience with gologit2? When would you use that over mlogit in cases where the outcome is ordinal but the model does not fit the proportional odds assumption?
Thanks for the video, it's excellent. Can robust variances be calculated for prevalence ratios or odds ratios in multinomial regression models for cross-sectional studies? Please, I await your help. Thank you!
This video helped me so much! 💜Thanks Professor! Greetings from Germany!
Thanks. Clear. How would you use the table to write the coefficient for the categorical age variable in the models?
Outstanding Video
very clear! thanks!
thanks Dr Mike..
This was very useful. Thank you.
I'll now go and check your channel; I am also looking for a video on interpreting marginal effects following a multinomial logistic regression
using STATA.
nice video. thank you very much.
May I ask how were you able to code the dependent variable into different categories? It's something I'm very curious about since I'm working on a logistic regression that has 3 different categories for the dependent variable.
so incredibly helpful thanks a lot
thank you, is there a criteria for choosing a base outcome?
Thank you so much for your explanation. I learned a lot. God bless
thank you sir for the explanation
this is really helpful, can you also guide me on few points:
1) is it better to import files with normal data and then recoding it into state? or even a coded data can be taken care off in state.
2) What if all variables are either categorical (having more than 2 categories) or binary (having two categories) then how to onvert them all together before running the model
3) if one needs to partition the data into training and testing split?
PS: I am new state user and your video helped a lot, please take out some time to reply to these doubts, it will help me finish my own model.
awesome explaination... thanks a lot
You are most welcome!
Thank you!
Thanks a lot!
Thank you so much !! Well explained
You are very welcome!
Please mike could you show me how to do multinomial regression for manel data ?
thank you very much 😍😍😍
Sir, It was indeed extremely helpful.. Sir, can you please guide that using the findings from the table, for example, females have a higher risk of belonging to the minor threat group, can we interpret something about males? looking forward to your reply.
Your guidance will be helpful to me to interpret my result as the category of interest for the Independent variable is selected by STATA as the base outcome
Hi Professor Crowson. Me again. Your videos are much better than Netflix! I want to perform a multinomial logistic regression in order to know the effect of age and education over the psychological impact of lockdown due to COVID-19. The thing is age and education are correlated. What do you recommend? May I have to control one of these two variables? How can I control it using Stata? Thank you for the enlightenment.
Do you have an example of probing an interaction in a multinomial logistic regression model?
This is very helpful, please explain how to deal with endogenous independent variable in logit model.
Thank You sir
Very nice presentation!
Thank you very much, Joy! Best wishes!
Many thanks
what do you mean treat the agecat as a factor in your analysis? Thank you great work !
Would you please explain the difference between RRR and odds ratio in multinomial logistics? why did many papers report the estimate association in Odds Ratio? Thank you for the explanation
Dear Dr. Crowson, I have not thought of the proportional odds assumption. Assuming my dependent variable type of job (based on skill level) low skilled, trades, and professional., I am using multinomial logit. I wonder if I will be challenged if I do not report or do any test. These are clustered categories different types of jobs are included in each category and skill levels are classified for professional A, for Trades B, and others C and D I merged. Please help me what to do.
really it s good explanation
Hello,
Is it possible to add level 2 variable in this model? For example, we introduce a "State" column in which participants are nested and see if that state's "number of hospitals" variable has some impact on individuals' perception of Covid threat.
Or can you suggest any another model that fits multilevel multinomial/mixed effect regression?
what is the code to make the dependent variable into categories ? so, how did you make the variable covidthreat_ph
Thanks a lot. U just save my life
You are most welcome!
Why did you use the "i." function only with age if the education and covid made up were alsocategorical variables? It has been my doubt for months whether to use the 'i.' and when not to
AWESOME
hello, I tried adding the "rrr" comand for my multinomial model but it did not work when it was after multiple imputations. That is, rather than the simple "mlogit" command, I was using the "mi estimate: mlogit .... rrr" and it did not work. It only worked for mlogit. Do you have any suggestion what I could do?
how about mnl from stated preference (SP) survey with several task scenarios?
For this particular dependent variable, a ordered logistic regression is more appropriate as the categories are in an increasing/decresing order :)
Yes, I thought so too, but see p. 86 in Heck, Thomas, & Tabata's (2012) Multilevel Modeling of Categorical Outcomes Using IBM SPSS in which they explain that when the ordinal outcome variable does not meet the proportional odds assumption (test of parallel lines), then the multinomial model can be used with the appropriate family and link function.
Agree with Carrie, this data doesn't meet the proportional odds assumption, as the presenter noted in the video, so Multinomial Logistic Regression is good choice.
I am trying to perform multivariate logistic regression in Stata 1 and I would also want to get adjusted odd ratio. May can you kindly assist.
About RRR, why are you interpreting statistical no significant coefficients (30-64 ages)? And what it does mean if the 95% Conf Interval includes RRR=1 but the pvalue is < 0.05?? Thanks very much.
How can I produce one p-value in factor variables such as age, when I want to use one age category as reference in a multivariate analysis
Hey, I’m doing multivariate logistical regression and having an issue. Let’s say my IV is categorical (age), and I have it in brackets 18-30 30-40 etc, and the p-value is below 0.05 for all ages apart from 2 categories. How do I interpret this?
How to check the GOF of MNL regression?
great lecture! But one question:
at 12:22 (or slide 16), why the result indicates that persons aged 65+ are predicted to be at greater risk of falling into the ‘minor threat’ group, and at lower risk of falling into the ‘not a threat’ group, as compared to persons aged 18-29? I think it should be 'major threat group' vs. 'not a threat group'. Did I miss something here?
Unwanted mistake.
what is the purpose of putting a base outcome?
@12.20 should it be "greater risk of falling into major threat group" rather than "minor threat group"?
yup
In this data tobit model is applied
in the last slide, it should be "major threat" instead of "minor threat"
personal note: 10:24 interpretation for categorical variable
So amazing work. Can I get the data
Hi Wanja, I have a couple of links underneath the video description. One of them is a link to the data file so you can download the data. Best wishes!