2.5. Measurement of Potential Confounding Variables

Potential confounding variables were chosen on the basis of having a statistically significant relationship with both the independent and dependent variables used in the logistic regressions.

All variables in this study were tested for this relationship in addition to other suitable variables from the SHIP. Variables that had a statistically significant relationship with independent and dependent variables include: income per fortnight, smoking, homelessness and accessibility to public transport.

Income per fortnight was the net income after tax and included non-taxable allowances (see demographics section for more information).

Smoking status was measured using a question from the Fagerstrom Test for Nicotine Dependence [41]. Participants were asked if they had smoked in the previous 4 weeks and if their answer was yes, participants were asked to quantify the number of cigarettes smoked per day. Responses ranged from 0 to 60. Number of cigarettes smoked per day is a valid measure for nicotine dependence [41].

‘Homeless’ in the SHIP was defined as living on the streets, parks, deserted buildings or living in temporary shelters. Homelessness was measured by asking participants how many days in the previous 12 months they had been homeless. Participants were asked to quantify the number of days they were homeless and responses ranged from 0 to 365.

The accessibility to public transport was measured as a categorical variable. Participants were asked if they had access to public transport close to where they were living and possible responses were either yes or no.