Monday, November 24, 2008

Notes about SHS 2006

Question

1) The Data Dictionary (shs2006cbk.pdf) presents conflicting information for two variables (NMVEHONP and VEHLEASP, page 36).

The Readme file indicates that the effective date for household equipment counts was the interview date. Would you confirm that the "Long name" information is inaccurate? If it is inaccurate, the variable labels used in SPSS and SAS are also in error.

NMVEHONP
Long name: Number of vehicles owned on December 31
Description: This variable gives the number of vehicles (car, van/mini-van, truck/sport utility vehicle) owned by members of the household at time of interview completely or partially for private use, excluding those leased.

VEHLEASP
Long name: Vehicles leased on December 31
Description: This variable gives the number of vehicles (car, van/mini-van, truck/sport utility vehicle) leased by members of the household at time of interview completely or partially for private use.

I'd also point out that since the responses to VEHLEASP are "yes" and "no", the "Description inaccurately describes the variable: it's not the "number of vehicles" leased but instead is a code to reflect the presence of leased vehicles (as indicated in the "Unit of Measure").

2) I was looking at variable NETCONEP, and saw that there was no "missing" category for the variable. Instead, people who weren't asked the question (those who did not use the internet from home, from variable INTERNET) are collapsed into the NETCONEP category of "0 -- No internet connection". I think that it would be preferable to explicitly separate those who weren't asked the question from those who responded that they didn't have an internet connection. It's unlikely, but possible, that respondents have a home internet connection, but do not use it: this possibility is eliminated by the current coding, which doesn't accurately reflect the questionnaire.

Answer
I have discussed your comments and questions about the SHS 2006 with the author division. Here are the responses they have provided.

NMVEHONP and VEHLEASP
1. "The correct date is December 31 of the reference year for vehicles. And the vehleasp does indicate the presence of at least one leased vehicle on Dec 31, not the number. The definitions should be modified."

NETCONEP
2. "If the respondant says no to internet use from home we do not ask any further questions on internet access and by default assign "no access". This may not be strictly true in some rare cases, but we have to minimise questions that the respondants consider a waste of time when they have already told us "no"."
The author division will be correcting the documentation to reflect the correct reference date for vehicles. I have asked them to provide us with the corrected versions as soon as they are available.

For the NETCONEP variable, I will ask our SPSS coder to change the status of the "0 -- No internet connection" code to "missing".

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.