Announcement

Collapse
No announcement yet.
X
Collapse
  • Page of 1
  • Filter
  • Time
  • Show
Clear All
new posts
  • #1

    How to track changes in morbidity status over time in Panel Data

    Hi everyone,
    I am working with panel data that consists of children surveyed over three waves: 3, 4, and 5.I am interested in how the health status of a child (morbid), a categorical variable that takes a value of 1 if the child has a chronic disease and 0 otherwise changes over time. I want to know how this variable varies with time by the child's stunting status (stunted); a categorical variable that takes a value of 1 if the child is stunted and 0 otherwise. Specifically, I want to know the proportion of children who were stunted in wave 3 and went on to develop/report having a chronic disease in waves 4 and 5, conditional on not having a chronic disease in wave 3. The idea is to establish whether there is a correlation between a child being stunted in a previous period and going on to report/develop a chronic disease in subsequent periods. Is there a code I can run to achieve this?

    My example data is as below:

    Code:
    * Example generated by -dataex-. To install: ssc install dataex
clear
input long pid float(wave morbid stunted)
401011 3 0 0
401011 4 0 0
401011 5 0 0
401013 3 0 0
401013 4 0 0
401013 5 0 0
401014 3 0 0
401014 4 0 0
401014 5 0 0
401016 3 0 0
401016 4 0 0
401016 5 0 0
401017 3 0 1
401017 4 0 0
401017 5 0 0
401018 3 0 0
401018 4 0 0
401018 5 0 0
401019 3 0 1
401019 4 0 0
401019 5 0 0
401020 3 0 0
401020 4 0 0
401020 5 0 0
401021 3 0 1
401021 4 0 1
401021 5 0 1
401022 3 0 0
401022 4 0 0
401022 5 0 .
401023 3 0 .
401023 4 0 0
401023 5 0 0
401024 3 0 0
401024 4 0 0
401024 5 0 0
401026 3 0 1
401026 4 0 0
401026 5 0 0
401028 3 0 0
401028 4 0 0
401028 5 0 0
401030 3 0 1
401030 4 0 0
401030 5 0 0
401031 3 0 1
401031 4 0 1
401031 5 0 0
401032 3 0 1
401032 4 0 1
401032 5 0 1
401034 3 0 .
401034 4 0 .
401034 5 0 .
401035 3 0 .
401035 4 0 .
401035 5 0 0
401036 3 0 .
401036 4 0 0
401036 5 0 0
401037 3 0 .
401037 4 0 0
401037 5 0 0
401038 3 0 .
401038 4 0 0
401038 5 0 0
401039 3 0 1
401039 4 0 0
401039 5 0 0
401040 3 0 0
401040 4 0 0
401040 5 0 0
401041 3 0 1
401041 4 0 0
401041 5 0 0
401042 3 0 0
401042 4 0 0
401042 5 0 0
401043 3 0 0
401043 4 0 0
401043 5 0 0
401044 3 0 1
401044 4 0 0
401044 5 0 0
401045 3 0 0
401045 4 0 0
401045 5 0 0
401046 3 0 1
401046 4 0 1
401046 5 0 1
401047 3 0 0
401047 4 0 0
401047 5 0 0
401048 3 0 0
401048 4 0 0
401048 5 0 0
401050 3 0 0
401050 4 0 0
401050 5 0 0
401052 3 0 .
401052 4 0 0
401052 5 0 0
401056 3 0 0
401056 4 0 0
401056 5 0 0
401060 3 0 0
401060 4 0 0
401060 5 0 0
401064 3 0 0
401064 4 0 0
401064 5 0 0
401065 3 0 0
401065 4 0 .
401065 5 0 0
401066 3 0 1
401066 4 0 1
401066 5 0 1
401068 3 0 0
401068 4 0 0
401068 5 0 0
401069 3 0 0
401069 4 0 0
401069 5 0 .
401070 3 0 0
401070 4 1 0
401070 5 1 0
401073 3 0 0
401073 4 0 0
401073 5 0 0
401075 3 0 0
401075 4 1 1
401075 5 0 .
401076 3 0 .
401076 4 0 0
401076 5 0 0
401077 3 0 0
401077 4 0 0
401077 5 0 0
401079 3 0 0
401079 4 0 1
401079 5 0 0
401080 3 0 0
401080 4 0 0
401080 5 0 .
401083 3 0 0
401083 4 0 0
401083 5 0 0
401084 3 0 0
401084 4 0 0
401084 5 0 0
401085 3 0 0
401085 4 0 0
401085 5 0 .
401087 3 0 1
401087 4 0 1
401087 5 0 .
401088 3 0 0
401088 4 0 0
401088 5 0 .
401089 3 0 0
401089 4 0 0
401089 5 0 .
401091 3 0 1
401091 4 0 1
401091 5 0 1
401092 3 0 0
401092 4 0 0
401092 5 0 0
401095 3 0 1
401095 4 0 1
401095 5 0 1
401097 3 0 0
401097 4 0 0
401097 5 0 0
401101 3 0 0
401101 4 0 .
401101 5 0 .
401102 3 0 0
401102 4 0 0
401102 5 0 0
401104 3 0 0
401104 4 0 0
401104 5 0 0
401105 3 0 0
401105 4 0 0
401105 5 0 0
401106 3 0 0
401106 4 0 0
401106 5 0 0
401108 3 0 0
401108 4 0 .
401108 5 0 0
401111 3 0 0
401111 4 0 0
401111 5 0 0
401112 3 0 0
401112 4 0 0
401112 5 0 0
401114 3 0 .
401114 4 0 .
end
label var pid "Person identifier" 
label var morbid "RECODE of morbidity_index"
    Tags: None
  • #2
    In some ways this kind of question could be handled by a reshape to wide layout.

    Instead an old trick is to calculate a profile over waves for each variable as a string variable. See https://journals.sagepub.com/doi/pdf...36867X20909698 for a write-up.

    Code:
    * Example generated by -dataex-. To install: ssc install dataex
clear
input long pid float(wave morbid stunted)
401011 3 0 0
401011 4 0 0
401011 5 0 0
401013 3 0 0
401013 4 0 0
401013 5 0 0
401014 3 0 0
401014 4 0 0
401014 5 0 0
401016 3 0 0
401016 4 0 0
401016 5 0 0
401017 3 0 1
401017 4 0 0
401017 5 0 0
401018 3 0 0
401018 4 0 0
401018 5 0 0
401019 3 0 1
401019 4 0 0
401019 5 0 0
401020 3 0 0
401020 4 0 0
401020 5 0 0
401021 3 0 1
401021 4 0 1
401021 5 0 1
401022 3 0 0
401022 4 0 0
401022 5 0 .
401023 3 0 .
401023 4 0 0
401023 5 0 0
401024 3 0 0
401024 4 0 0
401024 5 0 0
401026 3 0 1
401026 4 0 0
401026 5 0 0
401028 3 0 0
401028 4 0 0
401028 5 0 0
401030 3 0 1
401030 4 0 0
401030 5 0 0
401031 3 0 1
401031 4 0 1
401031 5 0 0
401032 3 0 1
401032 4 0 1
401032 5 0 1
401034 3 0 .
401034 4 0 .
401034 5 0 .
401035 3 0 .
401035 4 0 .
401035 5 0 0
401036 3 0 .
401036 4 0 0
401036 5 0 0
401037 3 0 .
401037 4 0 0
401037 5 0 0
401038 3 0 .
401038 4 0 0
401038 5 0 0
401039 3 0 1
401039 4 0 0
401039 5 0 0
401040 3 0 0
401040 4 0 0
401040 5 0 0
401041 3 0 1
401041 4 0 0
401041 5 0 0
401042 3 0 0
401042 4 0 0
401042 5 0 0
401043 3 0 0
401043 4 0 0
401043 5 0 0
401044 3 0 1
401044 4 0 0
401044 5 0 0
401045 3 0 0
401045 4 0 0
401045 5 0 0
401046 3 0 1
401046 4 0 1
401046 5 0 1
401047 3 0 0
401047 4 0 0
401047 5 0 0
401048 3 0 0
401048 4 0 0
401048 5 0 0
401050 3 0 0
401050 4 0 0
401050 5 0 0
401052 3 0 .
401052 4 0 0
401052 5 0 0
401056 3 0 0
401056 4 0 0
401056 5 0 0
401060 3 0 0
401060 4 0 0
401060 5 0 0
401064 3 0 0
401064 4 0 0
401064 5 0 0
401065 3 0 0
401065 4 0 .
401065 5 0 0
401066 3 0 1
401066 4 0 1
401066 5 0 1
401068 3 0 0
401068 4 0 0
401068 5 0 0
401069 3 0 0
401069 4 0 0
401069 5 0 .
401070 3 0 0
401070 4 1 0
401070 5 1 0
401073 3 0 0
401073 4 0 0
401073 5 0 0
401075 3 0 0
401075 4 1 1
401075 5 0 .
401076 3 0 .
401076 4 0 0
401076 5 0 0
401077 3 0 0
401077 4 0 0
401077 5 0 0
401079 3 0 0
401079 4 0 1
401079 5 0 0
401080 3 0 0
401080 4 0 0
401080 5 0 .
401083 3 0 0
401083 4 0 0
401083 5 0 0
401084 3 0 0
401084 4 0 0
401084 5 0 0
401085 3 0 0
401085 4 0 0
401085 5 0 .
401087 3 0 1
401087 4 0 1
401087 5 0 .
401088 3 0 0
401088 4 0 0
401088 5 0 .
401089 3 0 0
401089 4 0 0
401089 5 0 .
401091 3 0 1
401091 4 0 1
401091 5 0 1
401092 3 0 0
401092 4 0 0
401092 5 0 0
401095 3 0 1
401095 4 0 1
401095 5 0 1
401097 3 0 0
401097 4 0 0
401097 5 0 0
401101 3 0 0
401101 4 0 .
401101 5 0 .
401102 3 0 0
401102 4 0 0
401102 5 0 0
401104 3 0 0
401104 4 0 0
401104 5 0 0
401105 3 0 0
401105 4 0 0
401105 5 0 0
401106 3 0 0
401106 4 0 0
401106 5 0 0
401108 3 0 0
401108 4 0 .
401108 5 0 0
401111 3 0 0
401111 4 0 0
401111 5 0 0
401112 3 0 0
401112 4 0 0
401112 5 0 0
401114 3 0 .
401114 4 0 .
end
label var pid "Person identifier" 
label var morbid "RECODE of morbidity_index"


bysort pid (wave) : gen waves = strofreal(wave) if _n == 1 
by pid : gen shistory = strofreal(stunted) if _n == 1 
by pid : gen mhistory = strofreal(morbid) if _n == 1 

by pid : replace waves = waves[_n-1] + strofreal(wave) if _n > 1
by pid : replace shistory = shistory[_n-1] + strofreal(stunted) if _n > 1 
by pid: replace mhistory = mhistory[_n-1] + strofreal(morbid) if _n > 1 


list pid shistory mhistory if waves == "345"
    Now you can ask questions.

    Code:
    . list pid shistory mhistory if waves == "345" & strpos(substr(shistory, 2, 2), "1") & substr(mhistory, 1, 1) == "0"

     +------------------------------+
     |    pid   shistory   mhistory |
     |------------------------------|
 27. | 401021        111        000 |
 48. | 401031        110        000 |
 51. | 401032        111        000 |
 90. | 401046        111        000 |
117. | 401066        111        000 |
     |------------------------------|
132. | 401075        01.        010 |
141. | 401079        010        000 |
156. | 401087        11.        000 |
165. | 401091        111        000 |
171. | 401095        111        000 |
     +------------------------------+

. count if waves == "345" & strpos(substr(shistory, 2, 2), "1") & substr(mhistory, 1 , 1) == "0"
  10
    • 2 likes

    Comment

    • #3
      To be fair, we can ask questions in terms of reshaped variables:

      Code:
      reshape wide morbid stunted, i(pid) j(wave)
      after which we have morbid3 ... morbid5 and stunted3 ... stunted5.

      Comment

      Previous Next
      Working...
      X