Project

General

Profile

Support #352

BHPS

Added by Elisa Macchi almost 5 years ago. Updated almost 5 years ago.

Status:
Closed
Priority:
High
Assignee:
Redmine Admin
Category:
Data documentation
Target version:
Start date:
03/23/2015
Due date:
% Done:

100%

Estimated time:

Description

I am working with the BHPS.
I got the data sorted among all the waves. Once I tried to put all the waves together to create a panel (I used PID as reference variable), I noticed that for some individuals some years are missing: not something that can be explained by attrition, indeed generally just one year is missing (as if the individual just disappeared for a year).

Is this ok or am I doing something wrong? How do you explain this?

Second, as I said, I am using the cross-wave person identifier as variable of reference across the waves. Once I looked at the household id, however, I noticed that to the same PID were related different HID in the different waves. Is this ok?

Thank you for your help.
Elisa

History

#1 Updated by Redmine Admin almost 5 years ago

Please provide more detail - ground rules can be found here: https://www.understandingsociety.ac.uk/support/projects/support/wiki
Jakob

#2 Updated by Elisa Macchi almost 5 years ago

So, details:
- I am using Stata
- I considered from wave "i" to wave "r".
- I renamed all the variables in the waves, so that the name was always the same.
- I generated a new variable, "wave", which tell associates each obs to a wave.
- I used command "append" to merge all the dataset from wave "i" to wave "r", keeping only the variables I need (including PID).

sorting by "PID" and wave, I expect to have vertically the first PID observed for wave i,j,k,l,m,n,o,p,q,r; then, the second PID for all the waves and so on. Because of attrition, it might be that after some time there are no more observation (something like i, j,k.) Instead, for some PID I have : i,j,l (k missing).

For instance for pid "10014578" I have all the waves,except k (2001).

#3 Updated by Redmine Admin almost 5 years ago

  • Category changed from Data inconsistency to Data documentation
  • % Done changed from 0 to 50

I have copied in a reference to the relevant sections of the user guide below. Occasional unavailability and some forms of refusal were possible. You may also find the file, XWAVEID, useful. It holds the outcome variables for all waves (in a wide format).

IV. Sampling and Survey Methods .................................................................................................... A4-1

IV.9. Refusal Conversion ..................................................................................................... A4-9
IV.9.1. Previous wave refusal ................................................................................. A4-9
IV.9.2. Current wave new refusals .......................................................................... A4-9
IV.10. Maintaining Contact with Respondents ................................................................... A4-12
IV.10.1. Panel maintenance .................................................................................. A4-13
IV.10.2. Procedures during fieldwork .................................................................... A4-13

On behalf of the team,
Jakob

#4 Updated by Redmine Admin almost 5 years ago

  • Status changed from New to Closed
  • % Done changed from 50 to 100

#5 Updated by Alita Nandi almost 5 years ago

1. In the BHPS respondents are eligible for interview even if they have not responded in one wave. As a result different response patters are observed including the ones you pointed out. Some people may respond for a few waves and then not respond every again, while others respond intermittently, yet others respond in all waves.

2. There is no concept of a longitudinal household in the BHPS and so household IDs are not unique across waves - these are unique within a wave across different data files. So, the same person living in a single person household may have two different hid in two waves.

Also available in: Atom PDF