Announcement

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

  • Reliability Index of panel data set.

    Hi,

    I need to find the reliability (consistency of data) across groups (defined by an ID) in a panel data set. The data set is huge and I want to make sure that the observation at across time mutually agree with each other for each ID. Is there a test/or a method to which furnish an agreement index on a scale of 0-100% for the whole data set and each ID (group).

    Any help would be appreciated!
    Thanks

  • #2
    You might like this help-page. Especially the part on the 'msschk' command might be what you're looking for.

    Comment


    • #3
      "msschk" command is irrelevant to my posted issue.

      Thanks

      Comment


      • #4
        I'm not sure I understand the structure of your data and what exactly you mean by "reliability" or "consistency". Perhaps a sample of your data would help (please see FAQ 12.2 for how to post data https://www.statalist.org/forums/help#stata.

        Depending on what you want, you could construct a number of -assert- commands to ensure the data look like you want (var1 is constant within groups, or var2 is constant within years). Without knowing more, I don't think I can help.
        Stata/MP 14.1 (64-bit x86-64)
        Revision 19 May 2016
        Win 8.1

        Comment


        • #5
          A sample of my dataset in given below.
          Its a panel data set on school enrollment across months.

          clear
          input long schoolid byte financialmonth long enrolment
          10000 2 21
          10000 3 12
          10000 4 12
          10000 5 12
          10001 2 115
          10001 2 5
          10001 3 2000
          10001 4 0
          10002 2 42
          10002 2 63
          10002 2 109
          10002 2 122
          10003 2 376
          10003 2 398
          10003 2 409
          10003 2 409
          10004 2 115
          10004 2 122
          10004 2 123
          10004 2 131
          end

          What I exactly meant by reliability or consistency!

          Schoolid 1000 has an enrollment of 21,12,12,12 in four months. Though month 3,4,5 data is consistent but month 2 data is 21, which could be a typo of 12. Similarly, schoolid 10001 has 115,5, 2000,0, which are stark variations and which raises questions on the reliability of this panel data set.
          So, I want to statistically assess if these variations are acceptable with the help of a reliability index in Stata.


          Thanks

          Comment


          • #6
            I'm sorry, I don't know of any such measures. You could look for significant changes in enrollments or variation in ranges:
            Code:
            bysort schoolid: egen min_enroll=min(enrolment)
            bysort schoolid: egen max_enroll=max(enrolment)
            gen range=max_enroll-min_enroll
            Looking through help egen and the user contributed suite -egenmore- available on SSC (ssc install egenmore) might give you some ideas for statistics that you can calculate within schoolids.
            Stata/MP 14.1 (64-bit x86-64)
            Revision 19 May 2016
            Win 8.1

            Comment

            Working...
            X