Timeline for Alternative methods for logistic regression
Current License: CC BY-SA 3.0
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 2, 2016 at 17:11 | vote | accept | Happy Cretine | ||
| Jun 2, 2016 at 17:10 | answer | added | Happy Cretine | timeline score: 1 | |
| Jun 2, 2016 at 17:03 | history | edited | Happy Cretine | CC BY-SA 3.0 |
deleted 48 characters in body
|
| Apr 17, 2016 at 11:41 | history | edited | Happy Cretine | CC BY-SA 3.0 |
added 70 characters in body; edited title
|
| Apr 15, 2016 at 12:40 | comment | added | Björn | What do you mean by using Poisson regression? Is your outcome not just a yes/no, but rather a 0,1,2,3,4...? Or do you mean approximating the binomial distribution for extremely low rates and large risk sets by a Poisson distribution? Or do you mean using a Poisson regression likelihood with a follow-up offset to fit an exponential time-to-event model? | |
| Apr 15, 2016 at 12:08 | history | edited | Happy Cretine | CC BY-SA 3.0 |
added 120 characters in body
|
| Apr 15, 2016 at 11:42 | history | edited | Happy Cretine | CC BY-SA 3.0 |
added 120 characters in body
|
| Apr 15, 2016 at 9:27 | comment | added | Björn | My point was the selection bias of looking at it that way, when some may have becone unhealthy but you do not know (losses to follow-up) and some may yet do so (i.e. after the end of follow-up). Why only look at those alive? Presumably all will die eventually, what is the effect on your inference of looking at only those alive? | |
| Apr 15, 2016 at 9:18 | comment | added | Happy Cretine | @Björn Our sample consisted of people with no misssing values for baseline data. For example BMI at baseline had no missing values but after 6 year of follow up, 9% of information was missing for this variable. Anyway, we are mainly interested in analysing the variation of BMI over healthy aging and consider a selected sample (about 3000 observations) of people who survived at the end of the follow-up. | |
| Apr 15, 2016 at 6:08 | comment | added | Björn | Did all the subjects die already or do you have aome right- and/or left-censoring here? | |
| Apr 14, 2016 at 12:24 | comment | added | Frank Harrell | Poisson models are for count data for $Y$. Your $Y$ dichotomized several continuous variables. Although it's challenging to put together multiple outcomes into one variable, it's worth doing because binary $Y$ has minimal information/minimum power. | |
| Apr 14, 2016 at 12:11 | comment | added | Happy Cretine | Thank you Dr. Harrell. In this context the variable for healthy aging was created using several criteria on healthy aging components, such as absence of cancer-diabetes-cardiovascular disease ; overall cognitive functioning; depressive symptoms; quality of life, etc. As far as i know there was no dichotomization. I was told to use a poisson regression as an alternative to the logistic model, due to the high event rate. For me, it's not really necessary to use a Poisson model here. | |
| Apr 14, 2016 at 11:49 | comment | added | Frank Harrell | I often find that 20 events per predictor are required. But this is oversimplified because you need at least 96 observations just to estimate the intercept. I don't understand why you think a proportion of 0.37 for $Y=1$ could be a problem. And isn't healthy aging a continuous concept? If you used any dichotomization to arrive at $Y$ you'll get an inappropriate analysis. | |
| Apr 14, 2016 at 11:38 | history | asked | Happy Cretine | CC BY-SA 3.0 |