Project

General

Profile

Support #369

missing paju maju

Added by Carolina Zuccotti about 5 years ago. Updated about 5 years ago.

Status:
Closed
Priority:
High
Assignee:
-
Category:
Data documentation
Target version:
Start date:
04/30/2015
Due date:
% Done:

100%

Estimated time:

Description

Hello,
From what I understood in one of the replies you gave, paju (and maju) were asked only in Wave 1. But then, what are b_paju c_paju and d_paju? From the question/universe it seems as if it was asked in all waves to all respondents:

Universe
if (ff_ivlolw = 2|3|MIS) & ff_everint <> 1 //proxy last wave, non-interviewed adult or new entrant never interviewed, excluding rising 16 year olds

However, the number of missing for Waves 2, 3 and 4 is much higher (some new entrants are actually responding to the question, while others aren't). Why is this the case?

Looking forward to your response.
Thanks in advance.
Best regards,
Carolina

History

#1 Updated by Redmine Admin about 5 years ago

  • % Done changed from 0 to 50

Only asked of adults not interviewed before, why the frequencies look sparse for the later waves. The cross-wave file, XWAVEDAT, contains for convenience a consolidated version of these and other stable characteristics;
https://www.understandingsociety.ac.uk/documentation/mainstage/dataset-documentation/wave/xwave/datafile/xwavedat/variable/maju
Jakob

#2 Updated by Carolina Zuccotti about 5 years ago

Hi Jacob,
Thanks for your message. I cannot find the file. Where should it be?
I think, however, that my question is a different one. I have created "paju" and "maju" variables myself, with information from all 4 waves (I know that this is background information and, hence, stable information, and that is why it is b_paju c_maju etc. are sparse in Waves 2, 3 and 4).
However, the number of missing "paju" and "maju" is much higher in waves 2, 3 and 4 (20% or more) than in wave one (less than 2%). The only way to explain this is that new entrants are not being asked the question. See also: https://www.understandingsociety.ac.uk/support/issues/159. Here I understand that parental information is collected in Wave 1; and then in Wave 8, for entrants between Wave 2 and Wave 8.
Am I missing something?
Thanks,
Carolina

This is the syntax:
I merge all 4 waves and then create the following variables:

gen a_flag=1 if a_psu!=.
gen b_flag=1 if b_psu!=.
gen c_flag=1 if c_psu!=.
gen d_flag=1 if d_psu!=.

. gen paju=.
(73119 missing values generated)

. replace paju=a_paju
(50994 real changes made)

. replace paju=b_paju if paju<0 | paju==.
(18599 real changes made, 1452 to missing)

. replace paju=c_paju if paju<0 | paju==.
(7181 real changes made, 2804 to missing)

. replace paju=d_paju if paju<0 | paju==.
(6224 real changes made, 2327 to missing)

.
. gen maju=.
(73119 missing values generated)

. replace maju=a_maju
(50994 real changes made)

. replace maju=b_maju if maju<0 | maju==.
(18523 real changes made, 1417 to missing)

. replace maju=c_maju if maju<0 | maju==.
(7150 real changes made, 2795 to missing)

. replace maju=d_maju if maju<0 | maju==.
(6187 real changes made, 2291 to missing)

.
. lab values paju a_paju

. lab values maju a_maju

.
end of do-file

. tab paju

paju |      Freq.     Percent        Cum.
----------------------------------------+-----------------------------------
missing | 5 0.01 0.01
inapplicable | 11,028 16.34 16.34
proxy respondent | 2,412 3.57 19.92
refused | 1 0.00 19.92
don't know | 15 0.02 19.94
father working | 45,322 67.14 87.08
father not working | 3,844 5.69 92.77
father deceased | 2,410 3.57 96.34
father not living with respondent so do | 2,469 3.66 100.00
----------------------------------------+-----------------------------------
Total | 67,506 100.00

. tab paju if a_flag==1

paju |      Freq.     Percent        Cum.
----------------------------------------+-----------------------------------
inapplicable | 143 0.29 0.29
proxy respondent | 557 1.13 1.42
refused | 1 0.00 1.42
don't know | 1 0.00 1.42
father working | 40,833 82.79 84.22
father not working | 3,378 6.85 91.07
father deceased | 2,237 4.54 95.60
father not living with respondent so do | 2,169 4.40 100.00
----------------------------------------+-----------------------------------
Total | 49,319 100.00

. tab paju if b_flag==1

paju |      Freq.     Percent        Cum.
----------------------------------------+-----------------------------------
inapplicable | 9,359 18.45 18.45
proxy respondent | 1,163 2.29 20.74
don't know | 1 0.00 20.74
father working | 33,991 67.00 87.74
father not working | 2,693 5.31 93.04
father deceased | 1,791 3.53 96.57
father not living with respondent so do | 1,738 3.43 100.00
----------------------------------------+-----------------------------------
Total | 50,736 100.00

. tab paju if c_flag==1

paju |      Freq.     Percent        Cum.
----------------------------------------+-----------------------------------
inapplicable | 9,938 20.96 20.96
proxy respondent | 1,421 3.00 23.96
refused | 1 0.00 23.96
don't know | 3 0.01 23.97
father working | 30,555 64.45 88.41
father not working | 2,367 4.99 93.40
father deceased | 1,564 3.30 96.70
father not living with respondent so do | 1,563 3.30 100.00
----------------------------------------+-----------------------------------
Total | 47,412 100.00

. tab paju if d_flag==1

paju |      Freq.     Percent        Cum.
----------------------------------------+-----------------------------------
missing | 5 0.01 0.01
inapplicable | 11,028 23.39 23.40
proxy respondent | 2,412 5.11 28.51
refused | 1 0.00 28.51
don't know | 15 0.03 28.55
father working | 28,645 60.74 89.29
father not working | 2,188 4.64 93.93
father deceased | 1,440 3.05 96.98
father not living with respondent so do | 1,423 3.02 100.00
----------------------------------------+-----------------------------------
Total | 47,157 100.00

#3 Updated by Redmine Admin about 5 years ago

The XWAVEDAT file will be in the same folder as the other Understanding Society data files downloaded from UKDS. The older issue refers to the BHPS study.
Jakob

#4 Updated by Carolina Zuccotti about 5 years ago

Hi Jakob,
Sorry but I am a bit confused. From what I see the file Xwavedat has more information on parental background (as compared to the paju/maju I created by adding a_paju + b_paju + c_paju + d_paju).
Why is that? Why I do get fewer cases with parental background if I do not include the Xwavedat? Does this mean that in order to work with background variables we always have to use the Xwavedat?
I would appreciate very much if you could explain this to me with some detail (or refer to a document), since I am really puzzled at the moment.
Thanks in advance,
Carolina

#5 Updated by Redmine Admin about 5 years ago

XWAVEDAT also consolidates data from the surviving BHPS sample in the study, see e.g. https://www.understandingsociety.ac.uk/documentation/mainstage/dataset-documentation/wave/xwave/datafile/xwavedat/variable/maju
Jakob

#6 Updated by Carolina Zuccotti about 5 years ago

Ok, got it. Thanks.

#7 Updated by Redmine Admin about 5 years ago

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

Also available in: Atom PDF