Support #951

Deriving alcohol consumption units

Added by jason morgan almost 3 years ago. Updated over 2 years ago.

Derived variables
Start date:
% Done:



Dear USoc Team,

My research is examining the impact of alcohol consumption on individual earnings.

With regards to wave 7 of the Understanding Society dataset.
I was wondering if you could confirm for me whether the variable g_auditc4 (Drinks on a typical day) translates as units on a typical day or as any arbitrary definition of drink (e.g. could be a shot, could be a pint; with the definition of a standard drink often being 2.5 units). I am assuming it translates as units as the word drink appears to have implicitly been defined in the g_auditc5 (six or more drinks frequency) variable on the questionnaire where it says on page 557 “How often have you had 6 or more units on a single occasion in the last year? By a unit, I mean 1/2 pint of beer, a glass of wine or a single measure of spirit or liqueur.” but I’d be grateful if you could clarify this for me.

I'm therefore trying to derive a units consumed per week variable. My current approach has been to create a “Typical Units consumed per occasion” variable (TYPunits) which is the mean of the variable g_auditc4 (renamed to DrinkTypical) and then to create a “Weekly Drink Frequency” variable (DFx) which derives an approximation of the number of occasions the individual consumes alcohol per week using the variable g_auditc3 (renamed to DrinkFreq). I then multiply these two variables together to approximate a units consumed per week variable (i.e. multiplying the typical units an individual consume per occasion by their reported Weekly Drink Frequency). Of course depending on your answer to the first part of my question I will need to either multiply this number by 2.5, or redefine this variable to measure “drinks” consumed per week rather than units consumed per week. I have copied an excerpt from my STATA do-file related to the above explanation of how I’ve created this variable.
I’d be grateful for any thoughts on the above derivation or if there is a ‘standard’ or ‘official’ or more accurate method for deriving the number of units or drinks consumed by individuals in a given period.

Many thanks,

Jason Morgan

Do-File Excerpt - (Note: DrinkFreq = g_auditc3; and DrinkTypical = g_auditc4)
*//creating a TYPunits variable
gen TYPunits = 0
replace TYPunits =1.5 if DrinkTypical 1
replace TYPunits =3.5 if DrinkTypical 2
replace TYPunits =5.5 if DrinkTypical 3
replace TYPunits =8 if DrinkTypical 4
replace TYPunits =10 if DrinkTypical ==5
gen TYPunitsSQ = TYPunits^2
label variable TYPunits "approximation of units typically consumed on drinking day"
label variable TYPunitsSQ "square of (approximation of units typically consumed on drinking day)"

*//creating a “Weekly Drink Frequency” or DF multiplier
gen DFx = 0 if DrinkFreq 1
replace DFx = .23 if DrinkFreq 2
replace DFx = .69 if DrinkFreq 3
replace DFx = 2.5 if DrinkFreq 4
replace DFx = 5.5 if DrinkFreq ==5
label values DFx DFx
label variable DFx "Weekly Drink Frequency derived from individual's stated drink frequency"

*//creating a derived weekly units consumed variable
gen units = DFx*TYPunits
label values units units
label variable units "derivation of weekly units consumed (TypicalUnits*WeeklyDrinkFreq)"
gen unitsSQ = units^2
label variable unitsSQ "square of (derivation of weekly units consumed)"

So for example: if an individual typically consumes 3 drinks per occasion and 2-3 times per week, then
TYPunits = 3
DFx = 2.5
units = 3*2.5 = 7.5

Related issues

Copied from Support #119: Large values on alcohol consumption variablesClosedRedmine Admin01/31/2013

Copied to Support #965: Sample weightsResolvedjason morgan01/31/201304/21/2018


Also available in: Atom PDF