Home » Topics » Wealth Index » Error in SPSS code for Honduras 2011  2012 Wealth Index
Error in SPSS code for Honduras 2011  2012 Wealth Index [message #10206] 
Fri, 08 July 2016 11:42 

I wanted to alert you to an error in the SPSS code for the Honduras 2011 12 that's published on the wealth index construction page.
The error is near lines 496 508, in the Drinking water supply section. From the DHS, the code is:
compute h2oires=0.
if (qh121=11) h2oires=1.
variable labels h2oires "Public Piped into dwelling".
compute h2oyrd=0.
if (qh121=12) h2oyrd=1.
variable labels h2oyrd "Public Piped into yard/plot".
compute h2opires=0.
if (qh121=11) h2opires=1.
variable labels h2opires "Private Piped into dwelling".
compute h2opyrd=0.
if (qh121=12) h2opyrd=1.
variable labels h2opyrd "Private Piped into yard/plot".
You'll notice that when QH121 = 11 and 12, it's transformed twice into different dummy variables. The code should instead be:
compute h2oires=0.
if (qh121=11) h2oires=1.
variable labels h2oires "Public Piped into dwelling".
compute h2oyrd=0.
if (qh121=12) h2oyrd=1.
variable labels h2oyrd "Public Piped into yard/plot".
compute h2opires=0.
if (qh121=13) h2opires=1.
variable labels h2opires "Private Piped into dwelling".
compute h2opyrd=0.
if (qh121=14) h2opyrd=1.
variable labels h2opyrd "Private Piped into yard/plot".
This corresponds with the DHS survey question where 'Private Piped into dwelling' is coded as 13 and 'Private piped into yard/plot' is coded as 14.
Additionally, this error seems to have been incorporated into the factor analysis itself. The mean, standard deviation and component score coefficient for h2oyrd is identical to those for h2opyrd in the provided excel sheet. The same applies to h2oires and h2opires. This, of course, affects the values for the entire model.
Jeffrey W. Rozelle





Re: Error in SPSS code for Honduras 2011  2012 Wealth Index [message #10214 is a reply to message #10206] 
Sat, 09 July 2016 17:26 
TrevorDHS
Messages: 802 Registered: January 2013

Senior Member 


Thanks  this does look like an error in the code for the wealth index, but I suspect that this will actually have only a very small effect on the calculated wealth index, and very little practical effect for any analysis in terms of meaningful results. We will review this and let you know what the effect is.



Goto Forum:
Current Time: Mon Oct 7 13:43:54 Coordinated Universal Time 2024
