Presentation on theme: "Multiple Indicator Cluster Surveys Data Interpretation, Further Analysis and Dissemination Workshop Basic Concepts of Further Analysis."— Presentation transcript:
Multiple Indicator Cluster Surveys Data Interpretation, Further Analysis and Dissemination Workshop Basic Concepts of Further Analysis
Further Analysis: The Concept Any finding from the survey not covered in the final report Further = Beyond = Additional = Secondary May range from very simple (and unused) descriptive analysis to sophisticated statistical analyses, comparative analyses, trend analyses… Further analysis should not be seen as sophisticated/complex statistical analysis only!
Why Perform Further Analysis The final report is a first descriptive presentation of survey findings, describing Levels Patterns Associations/correlates Disparities, vulnerable groups (mostly mono- dimensionally) There is (almost) an infinite number of further analyses that can be performed, for various purposes
Purpose: Why Perform Further Analysis Better understand relationships, correlations, explanatory factors, causality Design more effective policies, interventions Understand change (or lack of change) Put results in a context of trends, comparisons Generate more descriptive results, unused data Inputs to future surveys Understand data quality New concepts A N D M O R E……
Every further analysis should start with a well defined (research) question and curiosity to understand more, and better…. Important! And ideas on possible uses of the results
Types, methods, approaches Descriptive analysis of unused data Simple descriptive analysis of indicator associations New analytical constructs Multivariate/complex/sophisticated analysis Trend analysis Comparative analysis Enhance data with other data sources Data quality analysis
Descriptive analysis of unused data MICS surveys/questionnaires and the final report are designed to be cost-effective since we do not collect data on topics for which at least a descriptive analysis plan does not exist use almost every question for the tabulations, indicators However, unused data still exists
Descriptive analysis of unused data More information on levels, patterns, associations, disparities Clues for future surveys Interview durations Household size, composition Educational attendance beyond secondary Consumer items Dwelling characteristics, ownership Vaccinations by age cohorts A full report on adolescents… Re-packaging?
Descriptive analyses of associations Final report includes tabulations of “indicators” by background characteristics, and patterns In a few cases shows trends – early marriage, childbearing, mortality Indicators can be easily cross-analyzed with other indicators – you already have them! Attitudes towards violent discipline of children with attitudes toward domestic violence Diarrhoea and use of improved water
New Analytical Constructs New “handles” to better understand Final report produces findings with pre-defined categories Create new categories to better define population groups experiencing elevated risks – for example, the urban poor Combine background characteristics: An ethnic group living in one zone Index construction
Multivariate/complex/sophisticated analysis Difficult and needs expert knowledge All such analyses are only as good as the research question, theoretical model….. But can produce better understanding of relationships, correlates, even determinants/causality (if supported with good theory) Socio-economic determinants of child mortality
Trend Analysis Very useful (and popular) to understand progress… (or lack of it) Challenges: Comparability Statistical significance
Trend analysis One survey can produce information on trends – early marriage, smoking, childbearing or mortality Compare results with results of previous surveys, data sources Always useful to support with external data – e.g. changes in use of bed nets, together with process indicators on distribution
Comparative Analysis Compare with other data sources collected in (more or less) the same time period Comparability? Cross-country or intra-country comparisons Useful for putting results in a context, as more can be understood through comparisons MICS has an advantage for such analyses as survey tools (indicators) are consistent with international definitions, harmonized with other data (e.g. DHS)
Enhance Data With Other Data Sources If linked with other data, new “findings” can be presented, to shed new light on survey findings Use GPS readings from MICS surveys with those from the health system to understand accessibility issues Use census information to convert results into absolute numbers - “magnitude” of necessary interventions, effective dissemination of results, sizes of groups of special interest, magnitudes of events or characteristics
Data Quality Analysis Some analysis presented in the final report The more this can be done early on, the better – but it is never possible to do all early on For better understanding of the non-sampling errors What worked, what did not, what needs to change Further approach “reality”
Each survey, each setting, each topic is somewhat unique and no prescription exists – in terms of specific topics However, a certain number of topics are of common interest and analysis of these are performed across the world Equity measures Youth and adolescents Data quality
Strategies to Encourage Further Analysis How do we ensure that data are further analyzed? Who should perform further analysis? “We need to get academics, research institutions to perform further analysis” Or do we?