Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Multiple Imputation with Plausible Values in the Data

    Hello everyone,

    I'm working with PISA 2012 dataset and the student performance data (my dependent variable) were intentionally set as 5 plausible values in the dataset. My independent variables, however, also contain missing values so I wanted to use -mi imputed chained- on these variables. I wonder if there is a way to impute the missing values of my independent variables while taking advantage of the -mi- function in analyzing the plausible values of my dependent variable?

    Any thoughts are appreciated!

    Best,
    Mindy

  • #2
    Hi All,

    Just a quick follow up as I'm still trying to figure out how to analyze the dataset with plausible values and missing data by using -mi-command. Here is how I imputed the plausible values of the math performance.

    Suppose the five plausible values are PV1MATH PV2MATH PV3MATH PV4MATH PV5MATH (already existed in the dataset), I created a new variable, pv0math, and imputed these five values as pv0math
    -mi import wide, imputed(pv0math = PV1MATH PV2MATH PV3MATH PV4MATH PV5MATH) clear-

    When I tried to utilize
    -mi imputed chained- on the missing values of other covariates, Stata indicated the imputation started from m=6. Is there a way to conduct multiple imputation on the variable of math scores and other covariates simultaneously, so that I'll have five imputed files in total?

    Any thoughts are appreciated!
    Mindy

    Comment


    • #3
      I can't easily think of a way to do this, although someone else may be able to help if you post an example of the data as specified in the FAQ. Use the -dataex- command, it is really useful!

      That said, the general recommendations for MI are to ensure that all the variables in whatever final statistical analysis you plan to run should be in the imputation model as well. And 5 imputations is too few. It may be worth re-doing the entire imputation sequence.
      Be aware that it can be very hard to answer a question without sample data. You can use the dataex command for this. Type help dataex at the command line.

      When presenting code or results, please use the code delimiters format them. Use the # button on the formatting toolbar, between the " (double quote) and <> buttons.

      Comment

      Working...
      X