Preparation and initial analysis of WHS tobacco data

1
Preparation and initial analysis of WHS tobacco
data

• St. Olaf College
• World Health Survey Group
• January 28, 2005

2
Project Objective

• To clean and perform an initial analysis on
  tobacco-related data provided by the World Health
  Survey

3
World Health Survey

• Designed to provide a comprehensive database from
  which to describe the health of many of the
  worlds developing countries.

4
World Health Survey

• Health risk factors
• Physical activity
• Tobacco
• Alcohol
• Nutrition
• Environmental factors

5
World Health Survey

• Provides
• First-time standardized data
• Coverage for developing countries.

6
World Health Survey

• Applications
• Monitor health systems
• Inform policy-makers
• Institute programs aimed at reducing health risks

7
WHS Tobacco Questionnaire

• Q4000 Do you currently smoke any tobacco
  products such as cigarettes, cigars, or pipes?
• No, not at all non-smoker
• Yes, but not daily occasional smoker
• Daily daily smoker

8
WHS Tobacco Questionnaire

• For daily smokers
• Q4001 For how many years are you smoking daily?

9
WHS Tobacco Questionnaire

• For daily smokers
• Q4002-4005 On average, how many of the
  following products are you smoking daily?
• Manufactured cigarettes
• Hand-rolled cigarettes
• Pipefuls of tobacco
• Other

10
Data description

• 32 countries
• 26 countries with blank records
• 10,168 total blank records
• 184,878 respondents

11
Variables used

• Sex
• Age/Age range
• Education level
• Years of education
• Current job
• Setting

• Smoking status
• Years of smoking
• Type of product
• Manufactured cigarettes
• Hand-rolled cigarettes
• Pipefuls of tobacco
• Other tobacco products

12
Data description

• Complete responses pre-cleaning
• 128,631 complete responses
• 69.6 of responses complete

13
The cleaning process do-file

• Contains data cleaning documentation
• Executable for future cleaning

14
Highlights of do-file

• Purposeful missings .a
• Illogical/invalid responses

15
Highlights of do-File

• Generate age_bar and agerange_2

16
Highlights of do-file

• Standardization of education levels
• Based on U.S. educational system
• 1 no formal education (0 years)
• 2 less than primary school (1-5 years)
• 3 primary school completed (6 years)
• 4 secondary school completed (7-8 years)
• 5 high school completed (9-12 years)
• 6 college/pre-university/university completed
  (13-16 years)
• 7 post-graduate degree completed (17 years)

17
Post-cleaning data description

• Complete responses post-cleaning
• 132,406 complete responses
• 71.6 (vs. 69.6 pre-cleaning) of responses
  complete

18
Data analysis objectives

• Determine proportions by country for
• Current daily smoker
• Occasional smoker
• Non-smoker
• Compare these proportions by
• Sex
• Age group
• Standardized education level
• Current occupation
• Urban/rural status

19
Data analysis objectives

• Determine the mean number of tobacco products
  smoked for each of the following
• Manufactured cigarettes
• Hand-rolled cigarettes
• Pipes
• Other tobacco products

20
Methods for data analysis

• Generate analysis do-file
• Construct confidence intervals
• Future methods
• pweights

21
Data analysis methods

• Itemized non-response

22
Selected results Two global trends

• Countries with higher proportion of daily smokers
  were typically eastern European and SE Asian
• Countries with the lowest proportion were
  typically African or Latin American

23
(No Transcript)
24
(No Transcript)
25
Selected results Age

• Highest proportion of daily smokers seen in
  respondents 40-59 years of age.

26
Selected results Age

• Eastern European countries higher proportions of
  younger daily smokers
• Slovakia, Estonia, Turkey, Hungary, Croatia
• SE Asian countries higher proportions of older
  daily smokers
• Bangladesh, Nepal, Laos, India

27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
Limitations of cleaning analysis

• Enhance comparability by
• Use of pweights
• Standardizing age
• Lack of background information
• Sampling techniques
• Sample size
• Non-respondents/blank records

32
Limitations of cleaning analysis

• Questionnaire difficulties
• Education level definition
• Wording of tobacco question
• Overly broad definition of occasional
• Cultural limitations

33
Interpretation of results

• Eastern Europe SE Asia higher proportions of
  daily smokers. Why?

• Higher incomes
• Lifestyle differences
• Economic proximity/increased trade with
  producers
• Increased Western cultural influence
• Recent political upheaval

34
Interpretation of results

• Gender split in daily smoker proportions. Why?
• Social acceptability bias
• Different stigmas for women men
• Economic access for women
• Gender symbolism
• Religious expectations

35
Suggestions for future research analysis

• Explore interactions
• Age distribution by sex
• Sex by urban/rural status
• Applying demographics
• GDP comparisons
• Life expectancy
• Literacy rates
• Infant mortality rates

36
Further studies

• For pre-18 year-old smokers age of smoking
  initiation
• Advertising/Marketing techniques employed
  regionally
• More economically inclusive regional surveys
• Countries with Tobacco-Free Initiative