CHAPTER TWO: METHODOLOGY

The overall design of the study was a cross-sectional survey of patterns of alcohol abuse in terms of frequency, risk and protective factors and impacts. This necessitates examination of self-reported use and abuse in the general population and among users, as well as examination of their views and those of key stakeholders including users significant others, government officials, civil society, faith based organization, business community among others. The study was conducted through the following methods and tools:

2.1. HOUSEHOLD SURVEY

2.1.1. Survey coverage
The target population for the study was all adults aged 15 – 64 years. The survey covered all the 7 districts (as of 1999 Census) in Central province ; Kiambu, Kirinyaga, Muranga, Nyandarua, Nyeri, Thika and Maragua (as at now Central province is divided into several other districts but there is no complete information, since 2009 census results have not been released)3. The survey disaggregated for rural and urban areas of the province.
2.1.2. Sampling methodology
A two stage cluster sample design was adopted. The first stage involved selection of EAs using Probability Proportional to Size (PPS) and the second was random selection of households.
2.1.3. Sampling Frame
The sampling frame for the study was the EAs from 1999 census and stratified according to district and urban/rural classification.
3 In addition, the new constitution of 2010 has abolished and replaced the districts with counties. The province has now five counties namely Kiambu, Muranga, Nyeri, Kirinyaga, and Nyandarua. However, the specific boundaries have not yet been specified. This implies that although the counties would be suitable context for presentation of the current survey, it is not possible to do so given that the survey was designed in the context of the seven districts existing as of 1999.

2.1.4. Sample Size and Allocation
In computation of the sample size, it was estimated that 30 % of adults in Central province consume alcohol, a coefficient of variation (CV) of 10% is targeted, design effect of 2 and a non-response adjustment of 5%. Based on above assumptions a uniform sample of 500 households per district was expected resulting into an overall sample of 3,500. Given that the Kenyan population distribution is 80% rural and 20% urban, the number of households to be selected from the various districts as in Table 2.
Table 1: Sample Allocation Dist Code District 1999 Census HH Estimated Sample (HH) Estimated EAs Rural EAs Urban EAs Total EAs Final Adjusted Sample (HH)

Table 2: Selection of Households Dist Code District Rural Number of households Urban number of households Total

2.1.5. Sampling of Eligible Respondents
Upon entry into a household the head of that household or in event of his/her absence any responsible member of that household (i.e., aged between 18 – 64 years and mentally aware) was selected to provide general household information. Such information included the identity of household members in terms of age, gender, occupation etc. and the general socioeconomic indicators such as property/assets owned, types of income generating activities, expenditure etc.
Out of all the members of the household, one of them aged 15 - 64 years was selected randomly to provide information on their own alcohol use or abstinence and related risk and/or preventive and impacts. The only exclusion criteria were mental or physical illness that would render such respondent unreliable or distressed from the interview.

2.1.6. Data Collection Methods
Face to face interviews with the aid of questionnaires was conducted among the household level respondents to identify the prevalence of alcohol use and abuse and quantifiable impacts (e.g., deaths) and identify perceptions on risk and protective factors.

2.1.7. Methods of Data Analysis
Quantitative data were summarized, organized and presented through graphics including pie-charts, histograms and frequency tables through SPSS.

2.1.8. Expected Outputs, Outcomes and Impact
The major output of the study is a report detailing:
1. Key evidence in terms of findings, conclusions, recommendations and other relevant information emerging,
2. A review on how alcohol abuse has been mainstreamed in the public policy and make proposals for further mainstreaming,
3. Specific recommendations feasible interventions relevant for internal and external mainstreaming of alcohol abuse which may be translated into programmes.

The major expected outcome is improved policy formulation, planning and programming towards evidence-based alcohol abuse interventions in Central Province and in Kenya. A related outcome is improved networking among various actors involved.
The major expected impact is ensuring that alcohol abuse does not compromise developmental objectives. This will be achieved by ensuring that abuse is mitigated among all but particularly among the youth.