Tuesday, April 8, 2014

CCHS Mental Health PUMF


Question

I’ve been working on the PUMF released for the Canadian community Health Survey 2012 (Mental Health), and have some questions and comments about the file. For all of the interference means (DEPDINT, MIADINT, HYPDINT, BIPDINT1, BIPDINT2, BIPDINT, GADDINT, AUDDINT, SUDDINT): do lower scores indicate less interference? I’m assuming so, based on looking at the derivation of DEPFINT and MIAFINT, but want to be sure.

Incorrect SPSS variable labels:
MEDGOTHR 'Used - antidepressant - 2 D - (F)' should be “Used - other – 2D - (F)”
PN1_01D2 "…: not readily available" should be “…: job interfered”
EDUDR04 "Highest level/edu.: HH 4 levels (Derived)" should be “Highest level/edu.: RESPONDENT 4 levels (Derived)

Incorrect SPSS value labels:
SPSDATT, SPSDGUI, SPSDALL, SPSINT, SPSDWOR
- 6 is NOT “Not applicable”, but part of scale
- 7 is NOT “Don’t know”, but part of scale
- 8 is not “Refusal” by highest end of scale

WSTDJIN, WSTDPHY
- There is no value 5 as 4 is the top of the scale

DHHGHSZ
- 5 should be “5 persons or more in household”

EDUDH04 EDUDR04
- 3 should be “Some post-secondary”

FMI_05
99999.96 "NOT APPLICABLE"
99999.97 "DON'T KNOW"
99999.98 "REFUSAL"
99999.99 "NOT STATED"
- Variable has 2 decimal places; 99999.96 through 99999.99 should be declared as the missing values, not 99996 through 99999.

Answer

We’ve looked into the format errors for SPSS and we will be preparing an errata for most of the items listed below. We will be producing new format files that will be available on request as well.I was unable to replicate the issue for this variable: PN1_01D2. It seems that all is fine on the SPSS code we currently have.

In regards to the other question: Yes, lower scores indicate less interference.