MARCH Outcome Evaluation
4.4. Data processing and analysis
Each questionnaire was edited at the site by the supervisors and the consultants. All questionnaires were also edited and post-coded for computerization. The data were computerized using EPI-Info software. Double data entry procedures and other programming techniques were employed to assure the quality of the data.
Data analysis, including Bivariate and multivariate methods, were done using STATA. All associations/correlations were tested for significance. Different multivariate analyses procedures were employed in accordance with the nature of the outcome variables. Whereas Logistic regression analyses were frequently used for the dichotomous outcome variables, Ordinary Least Square (OLS) method was employed for the continuous outcome variables. To examine responses on psychosocial (e.g. self-efficacy and outcome expectancy) and attitudinal statements, a Likert-type summative scale was used. Five categories were used, where 1 = Not at all sure 2= Only a little sure 3 = Somewhat sure 4 = Very sure and 5 = Extremely sure. (Scores were assigned to each of the responses to reflect the strength and direction of the attitude expressed in a particular statement, with 5 indicative of a strongly positive perception/attitude and 1 reflective of a strongly negative perception/attitude towards the particular statement. By averaging the responses to a given series of related statements, summary scores were developed. To evaluate the internal consistency of the items, Chronbach’s alpha coefficient was computed. Typically, social scientists consider an alpha coefficient of 0.60 or greater to constitute a strong measure of internal reliability.
The poor quality of the baseline as well as part of the outcome survey data (such as data on condom use and HIV testing self-efficacy, and outcome expectancy, and those related to stigma and discrimination) precluded the baseline-outcome comparison of results for West Hararghe. Also in Addis Ababa, data on psychosocial constructs concerning condom use and HIV testing (i.e. self-efficacy and outcome expectancy) and information on stigma and discrimination were measured differently at baseline, thus, not used in the present evaluation.
Addis Ababa and West Hararghe