[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

From |
Steven Samuels <sjhsamuels@earthlink.net> |

To |
statalist@hsphsun2.harvard.edu |

Subject |
Re: st: event history analysis with years clustered in individuals |

Date |
Sun, 15 Feb 2009 11:41:09 -0500 |

On Feb 15, 2009, at 9:22 AM, Austin Nichols wrote:

Hilde Karlsen <Hilde.Karlsen@hio.no>: If you have to use a mixed model as an exercise, and you have no compelling reason to choose a particular research question, you should ask a different research question where a mixed model is a more appropriate model, not apply it blindly to data you know is better suited to a survival model. Why not use the attrition dummy you have made as the explanatory variable in a mixed model instead--what other variables do you have on the data?On Sun, Feb 15, 2009 at 8:26 AM, Hilde Karlsen<Hilde.Karlsen@hio.no> wrote:Thank you both for the advice. However, I don't think I can do as yousuggest because I have to use a multilevel approach for this essay(it is anessay for a multilevel course I followed a while ago). I shouldprobablyhave been more clear on this issue, and on what my problem reallyis. What Iam wondering is not which method/command I should use, but how Iam going tointerprete the sigma_u estimate when my level 1 variable is yearsand mylevel 2 variable is individuals. As mentioned, I find it more intuitive to grasp the point of separatevariance estimates when the levels are schools, classes etc, butfor somereason I have a hard time understanding how I should interpreate thevariance estimate sigma_u when the years are clustered inindividuals. Howshould I interpreate sigma_u when years are clustered in individuals.I asked the professor who was leading the course which command Ishould use,and he told me I should use xtmelogit (my advicor told me the samething).As he is the one who is going to judge wheter I pass or not onthis essay,it is probably best to follow his advice.I agree that it is a survival model, and I have designed my datafor thistype of analysis (i.e. all individuals in the file start out with0 on thedependent variable, and when/if they drop out of the nursingoccupation,they receive 1 on the dependent variable. I have no info on whichdate/monthpeople drop out; I only have information on which year they dropout).Regards, Hilde Quoting Steven Samuels <sjhsamuels@earthlink.net>:Hilde, I agree with Austin's approach. Even if you have onlymonths, notdays, of starting and quitting, use that time unit in a survivalor discretesurvival model. I recommend Stephen Jenkins's -hshaz- (get itfrom SSC);his "model 1" (the "Prentice-Gloeckler model" is the same as thatfit by-cloglog-. His model 2 adds unobserved heterogeneity and so maybe morerealistic (Heckman and Singer, 1984).I would not be surprised if prediction equations for of early andlaterquitting differed. If so, time-dependent covariates or separatemodels forearly and later quitting, would be informative. -Steve Prentice, R. and Gloeckler L. (1978). Regression analysis of groupedsurvival data with application to breast cancer data. Biometrics34 (1):57-67.Heckman, J.J. and Singer, B. (1984). A Method for minimizing theimpact ofdistributional assumptions in econometric models forduration data,Econometrica, 52 (2): 271-320.Hilde Karlsen <Hilde.Karlsen@hio.no>: Attrition from nursing sounds like a survival model, probably in discrete time, using -logit- or -cloglog- with time dummies, not -xtmelogit- (seehttp://www.iser.essex.ac.uk/iser/teaching/module-ec968 for atextbookand self-guided course on discrete time survival models). Ifyou haveT years of data on each individual, all of whom are first-yearnursesin period 1, and some of whom quit nursing in each of thesubsequentyears, with a variable nurse==1 when a nurse (and zerootherwise), anindividual identifier id, a year variable year, and a bunch of explanatory variables x*, you can just: tsset id year bys id (year): g quit=(l.nurse==1 & nurse==0) by id: replace quit=. if l.quit==1 | (mi(l.quit)&_n>1) tab year, gen(_t) drop _t1 logit quit _t* x* and then work up to more complicated models with heterogeneousfrailty, etc. The main issues are that someone who quit nursinglastyear cannot quit nursing again this year, and people who never quit nursing might at some future point that you don't observe, which is why you use survival models...If you know the day they started work and the day they quit, youmightprefer a continuous-time model (help st). I've been assuming you had data on people working as nurses, but rereading your email, maybe you have data on breastfeeding mothers,though I suppose the same considerations apply (though withmultipleyears of data on breastfeeding mothers, there is probably no censoring).On Fri, Feb 13, 2009 at 9:19 AM, Hilde Karlsen<Hilde.Karlsen@hio.no>wrote:Dear statalisters,This is probably a stupid question, but I've been searchingaround thenetsand in books and articles, and I've still not grasped theconcept: WhenI'm performing a multilevel analysis of attrition from nursing using xtmelogit,and time (year) is the level 1 variable and individuals (id) isthelevel 2 variable (i.e. years are clustered within individuals; I have a person-yearfile), how do I formulate the expectation related to thismodel? Why isit important to separate between these two levels? I find it more intuitive to grasp the fact that individuals are clusteredwithin schools, and that variables on the school level - aswell asvariables on the individual level - may influence e.g. whichgrades astudent gets.I understand (at least I hope I understand) the point that whenthe sameindividuals are followed over a period of time, the individual's responsesare probably highly correlated, and that this implies aviolation totheassumption about the heteroskedastic error-terms. As I see it,I couldhaveused the cluster() - command (cluster(id))to 'avoid' thisviolation;however, I have to write an essay using multilevel analysis, sothis isnot an option.I don't know if I'm being clear enough about what my problemis, but anyinformation regarding this topic (how to grasp the concept ofyearsclustered in individuals) will be greatly appreciated.I'm really sorry for having to ask you such an infantilequestion.. Mycolleagues and friends are not familiar with multilevelanalyses, so Idon't know who to turn to. Best regards, Hilde* * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

* * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

**Follow-Ups**:**Re: st: event history analysis with years clustered in individuals***From:*Hilde Karlsen <Hilde.Karlsen@hio.no>

**References**:**st: event history analysis with years clustered in individuals***From:*Hilde Karlsen <Hilde.Karlsen@hio.no>

**Re: st: event history analysis with years clustered in individuals***From:*Austin Nichols <austinnichols@gmail.com>

**Re: st: event history analysis with years clustered in individuals***From:*Steven Samuels <sjhsamuels@earthlink.net>

**Re: st: event history analysis with years clustered in individuals***From:*Hilde Karlsen <Hilde.Karlsen@hio.no>

**Re: st: event history analysis with years clustered in individuals***From:*Austin Nichols <austinnichols@gmail.com>

- Prev by Date:
**Re: st: event history analysis with years clustered in individuals** - Next by Date:
**Re: st: Mata version control** - Previous by thread:
**Re: st: event history analysis with years clustered in individuals** - Next by thread:
**Re: st: event history analysis with years clustered in individuals** - Index(es):

© Copyright 1996–2016 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |