Tobacco
Tobacco
Number of cigarettes consumed is an indicator of several mental illnesses including anxiety (Lawrence et al. 2010).
Data
What variables are included? Why is this output chosen. What explanatory variables are used and why are they chosen
counts_density("data/transitions/ncigs/zip/ncigs_2018_2019.rds", "y")
plot of chunk tobacco_data
Methods
The number of zero inflated values is higher than expected for a count distribution such as a poisson distribution. This inflation occurs naturally as a large proportion (over 50%) of the population do not smoke. There are two sources of cigarette consumption that can be modelled using zero inflated models. In this case a zero-inflated poisson (ZIP) is used. Two models are fitted simulatenously. One is a logistic regression that estimates whether a person smokes cigarettes or not. This provides a simple probability of smoking or not. The second is a poisson counts model estimating the number of cigarettes consumed.
Data
Two set of variables are needed for the logistic and poisson parts of the ZIP model respectively.
Variables that predict how much a person smokes.
age. persons age. generally older people and very young smoke. SF_12. wellbeing estimates number of cigarettes smoked. labour_state. whether a person is employed or not. ethnicity. certain ethnicities more likely to smoke cigarettes. education_state. highest qualification. job_sec job quality hh_income household income ncigs previous number consumed.
Variables that predict whether a person smokes
ethnicity. certain ethnicities more likely to smoke cigarettes. labour_state. whether a person is employed or not. age SF_12. wellbeing estimates number of cigarettes smoked. ncigs previous number consumed.
Results
Almost all coefficients significant. Particularly prevous consumption of cigarettes. Good estimation of the number of non-smokers in the population at around 55%. Counts of smoking are underdispersed and fail to estimate consumption over 20 cigarettes.
##
## Call:
## zeroinfl(formula = formula, data = dat.subset, weights = weight, dist = "pois", link = "logit")
##
## Pearson residuals:
## Min 1Q Median 3Q Max
## -0.22590 -0.02780 -0.01848 0.00000 3.90660
##
## Count model coefficients (poisson with log link):
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.826e+00 NA NA NA
## age 1.364e-02 NA NA NA
## factor(sex)Male 5.071e-02 NA NA NA
## relevel(factor(education_state), ref = "3")0 3.819e-02 NA NA NA
## relevel(factor(education_state), ref = "3")1 3.870e-01 NA NA NA
## relevel(factor(education_state), ref = "3")2 9.921e-02 NA NA NA
## relevel(factor(education_state), ref = "3")5 -3.817e-02 NA NA NA
## relevel(factor(education_state), ref = "3")6 -2.838e-01 NA NA NA
## relevel(factor(education_state), ref = "3")7 2.962e-01 NA NA NA
## SF_12 -3.491e-03 NA NA NA
## relevel(factor(job_sec), ref = "3")1 8.265e-02 NA NA NA
## relevel(factor(job_sec), ref = "3")2 7.372e-01 NA NA NA
## relevel(factor(job_sec), ref = "3")4 1.603e-01 NA NA NA
## relevel(factor(job_sec), ref = "3")5 1.904e-01 NA NA NA
## relevel(factor(job_sec), ref = "3")6 1.211e-01 NA NA NA
## relevel(factor(job_sec), ref = "3")7 1.474e-01 NA NA NA
## relevel(factor(job_sec), ref = "3")8 2.507e-01 NA NA NA
## hh_income 4.793e-05 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")BAN -5.872e-01 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")BLA -5.784e-01 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")BLC -5.287e-01 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")CHI -4.005e-01 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")IND -5.132e-01 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")MIX -2.855e-01 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")OAS -4.219e-01 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")OBL -7.849e-01 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")OTH 7.849e-02 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")PAK -6.763e-01 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")WHO -3.410e-02 NA NA NA
##
## Zero-inflation model coefficients (binomial with logit link):
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.3968641 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")BAN -0.0665143 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")BLA 0.7077274 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")BLC -0.8377246 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")CHI 0.2628890 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")IND 1.4577262 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")MIX -0.2970005 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")OAS 1.0415864 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")OBL 0.3374910 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")OTH -0.8345472 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")PAK -0.0972652 NA NA NA
## relevel(factor(ethnicity), ref = "WBI")WHO -0.2608373 NA NA NA
## relevel(factor(job_sec), ref = "3")1 0.4389168 NA NA NA
## relevel(factor(job_sec), ref = "3")2 0.5864644 NA NA NA
## relevel(factor(job_sec), ref = "3")4 -0.1661200 NA NA NA
## relevel(factor(job_sec), ref = "3")5 -0.2828293 NA NA NA
## relevel(factor(job_sec), ref = "3")6 -0.8209764 NA NA NA
## relevel(factor(job_sec), ref = "3")7 -0.7934972 NA NA NA
## relevel(factor(job_sec), ref = "3")8 -0.7788828 NA NA NA
## hh_income 0.0002134 NA NA NA
## SF_12 0.0179618 NA NA NA
##
## Number of iterations in BFGS optimization: 90
## Log-likelihood: -84.87 on 50 Df
plot of chunk tobacco_output
References
Lawrence, David, Julie Considine, Francis Mitrou, and Stephen R Zubrick. 2010. “Anxiety Disorders and Cigarette Smoking: Results from the Australian Survey of Mental Health and Wellbeing.” Australian & New Zealand Journal of Psychiatry 44 (6): 520–27.