MICS4 Survey Design Workshop Multiple Indicator Cluster Surveys Survey Design Workshop Data Analysis and Reporting
MICS4 Survey Design Workshop Provide an overview of the MICS4 process in analyzing data Provide an overview of the preparation of the Preliminary Findings and the Final Report Purpose
MICS4 Survey Design Workshop Data entry and processing programs will be covered in second workshop Standard set of MICS4 tabulations and guidelines will be provided Standard guidelines and templates for preliminary findings and final report provided Process for report preparation
MICS4 Survey Design Workshop Process for report preparation If needed, technical support for data analysis and report writing will be through UNICEF regional consultants Once a draft final report is produced, arrangements will be made to review reports among different experts with the MICS4 Country teams
MICS4 Survey Design Workshop Carry out basic checks of data quality (non-sampling errors) Calculate response rates Response rates lower than 90% suggest results may be biased; include response rates in the reports Check for variables with large numbers of missing values Make sure it is not a problem with data entry or editing; any variable with 10% or more missing values, use with caution; if very high, do not use at all Check for variables with large numbers of “don’t know” or “other” answers. Large numbers of DKs can indicate difficulty understanding the questions; High proportions of “other” indicates questionnaire did not account for most common responses Before producing tabulations and writing reports, the following tasks should be completed:
MICS4 Survey Design Workshop Check for expected patterns in the data. – Unexpected patterns may result from faulty sample design, improper implementation of the sample, interviewer errors, or respondents’ inaccurate answers. Calculate the wealth index Decide on basic background variables and their groupings Geographic areas, urban/rural residence, level of education, male/female, wealth quintiles, religion of household head, etc. (continued)
Check unweighted number of observations for displaying results –Cells based on very small numbers of cases are not reliable and should not be shown; results based on fewer than 25 unweighted cases should not be shown, cases should be shown within parentheses (caution). MICS4 Survey Design Workshop (continued)
MICS4 Survey Design Workshop Executive summary Introduction and background Data quality and sample characteristics Results (with sections by topic) Sample design (does not have to wait to the end) Survey instruments Outline of Final Report
MICS4 Survey Design Workshop Dissemination of Findings Separate session Third regional workshop: –Dissemination and Archiving National seminar, distribution of findings, other dissemination activities
MICS4 Survey Design Workshop MICS data can be used for statistical analysis beyond that contained in the survey report The data collected should be exploited as fully as possible Potential topics for further analysis: –Analysis of data quality –Analysis of trends –Identification of “most vulnerable” groups –Multivariate analyses of determinants of child health or schooling outcomes Further Analysis of the Data