Multiple imputation in Stata®: Linear regression

Поділитися
Вставка
  • Опубліковано 1 лис 2024
  • Learn how to use Stata's mi suite of commands to handle missing data. This tutorial covers how to impute a single continuous variable with predictive mean matching using the mi impute regress command.
    www.stata.com
    Copyright 2011-2019 StataCorp LLC. All rights reserved.

КОМЕНТАРІ • 37

  • @statacorp
    @statacorp  10 років тому +4

    The command shown in the title screen is "mi regress". Note that the actual command is "mi impute regress".

  • @mirjanagrkovska6748
    @mirjanagrkovska6748 6 років тому +2

    Thanks for this comprehensive video. The last regression (after the imputation was done) gives results based on the exact (full) number of observations. But once I save my dataset and then reopen to run regressions, it gives me results where the number of the observations now exceed the regular size of the sample. How is this possible? Where did I make mistake? Many thanks

  • @kayak0000
    @kayak0000 6 років тому +1

    If attack is the dependent variable, I don't it's the appropriate variable to be used for imputing missing values of the predictor variable bmi. If you run a logistic regression with a dichotomous attack and continuous bmi, wouldn't the association between the 2 variables be overestimated? I would probably select a covariate/s that is/are highly correlated with the variable being imputed but then I'll have collinearity problem. Any guidelines on choosing the variables that should be used for imputation?

  • @shelleyyan4112
    @shelleyyan4112 4 роки тому

    Very helpful, even for advanced users of Stata. Thanks!!!

  • @iGuRuch
    @iGuRuch 11 місяців тому

    Can i impute a dependent variable? Thank you!

  • @sarwatmasud8600
    @sarwatmasud8600 2 роки тому

    Hi Thanks for this video. In the beginning of the video you refer to a manual. Can you please provide a link for that manual ?

    • @statacorp
      @statacorp  2 роки тому

      www.stata.com/bookstore/multiple-imputation-reference-manual/

  • @vitaltopics316
    @vitaltopics316 3 роки тому +2

    How to do MI for panel data with multiple variables (N=30 countries, T = 21 years), some countries have missing values for some variables, please someone guide me, thank you in advance

  • @florentbesnier6946
    @florentbesnier6946 Рік тому

    thank you ! 1/ how to do it with 2 (or more) variables where we observe missing data ? (in your exameple it is bmi only. how to do it if it would be 'attack' and 'bmi' that you observed missing data ?) . 2/ how to report the statistical analysis part in an article ? thank you for your response

    • @shareemrasyidi9948
      @shareemrasyidi9948 11 місяців тому

      For more than one variables just include the other variable you need to impute when using the impute command, make sure you also use register command on that variable

  • @yeneworkmulu9584
    @yeneworkmulu9584 6 років тому +1

    good tutorial,how can i impute categorical variables to set for survival analysis?

  • @Silverstein1883
    @Silverstein1883 9 років тому

    A very helpful video! Thank you so much!

  • @noorakazanji7423
    @noorakazanji7423 Рік тому

    Very helpful.

  • @nitinrajput6831
    @nitinrajput6831 8 років тому

    Chuck. thanks very much for the information. How can I impute more than 1 variable at the same time?

    • @lonjezosithole6285
      @lonjezosithole6285 7 років тому +1

      Nitin Rajput you can use MICE (multiple imputation using chained equations)

  • @SyedShah-bn5wc
    @SyedShah-bn5wc 3 роки тому

    If i wanted to impute data for a PCA, how would i do this? Simply impute without specifying a regression model?

  • @lesiba5382
    @lesiba5382 6 місяців тому

    What about imputing categorical variables, the video can be just as brief?

  • @mariahlm9748
    @mariahlm9748 11 років тому

    Hi thanks for the video, another one about multivariate imputation would be useful. Thank you.

  • @lucagiraldi372
    @lucagiraldi372 8 років тому

    Why you performed the logistic regression model from the multiple imputation panel?
    It would be the same if you had launched the logistic regression model from the logistic regression command?

  • @sainaveenbali
    @sainaveenbali 8 років тому

    est how to check if i should go for fixed or random model post-imputation, hausam dont work on weighted data

  • @sainaveenbali
    @sainaveenbali 10 років тому

    okie now i imputed and wanna try panel model, but hausman wont work , help plz

  • @kalebnbrown
    @kalebnbrown 3 роки тому

    Thanks!

  • @sovsen92
    @sovsen92 9 років тому

    I'm sorry, what manual is it that you are referring to?

    • @statacorp
      @statacorp  9 років тому

      Stata's Multiple Imputation Reference Manual.

  • @OrtegaMiss
    @OrtegaMiss 9 років тому

    How do I check the imputed data?

    • @statacorp
      @statacorp  9 років тому +3

      +Gemari Ortega - mi set - creates a new variable named "_mi_m". Each imputed dataset is numbered 0 through M where M is the number of imputed datasets. So if you created 5 imputed datasets, the variable "_mi_m" would contain 0-5. The observations that correspond to _mi_m=0 are the original data. You can use _mi_m to check each imputed dataset separately. If you would like more information, please contact Technical Services.

  • @sonosoomi
    @sonosoomi 9 років тому

    The screen is hard to see.

    • @statacorp
      @statacorp  9 років тому +1

      +Soomi Lee HI! Be sure to set your video playback to the highest quality. The HD setting is preferable.

  • @alexelefelt
    @alexelefelt 4 роки тому +1

    stop being so boring

  • @RPDBY
    @RPDBY 7 років тому

    useless tutorial. how to do it with panel dataset and multiple variables?

    • @lonjezosithole6285
      @lonjezosithole6285 7 років тому

      RPDBY you should read the manual

    • @RPDBY
      @RPDBY 7 років тому

      i have about thirty variables, 700 units of analysis and 10 years worth of data and have not been able to find anything there that would help me to do multiple imputation with this dataset, maybe you could direct me to a specific page if you know there might be something useful?

    • @vitaltopics316
      @vitaltopics316 3 роки тому

      @@RPDBY Hope you are well and expert in STATA. Do you mind sharing your experience with me please? After 4 years, still there is no helpful guidance on how to do it for unbalanced panel data of multiple variables.

    • @RPDBY
      @RPDBY 3 роки тому +2

      @@vitaltopics316 still no idea, but after doing some research I concluded it is not worth doing it because it does not help much to improve the analysis quality (imputations will always be questioned by any serious academic) but is a lot hassle. either get more data or work with what you have

    • @vitaltopics316
      @vitaltopics316 3 роки тому

      @@RPDBY Thank you so much for the fastest reply. I will work with the data I have. You saved me a lot of time. THANK YOU SO MUCH