Aspects of quantitative analysis include:
- Sampling and selection (click link for further discussion)
- Tools for collecting data
- Data management
- Analysis methods
In addition, this page also touches on how some teams have engaged participatory numbers to aggregate or count information gathered through qualitative and participatory approaches.
Sources of Data
For quantitative analysis, sources of information can come from a spectrum of methods, which include:
Data Management: to ensure quality of information gathered, a data management plan must be in place. The plan should outline objectives of the study, the hypothesis being explored, variables to be explored and plans for analyzing the study objective.
For quantitative analysis, data management should have
A tested and reliable codebook: that labels each question, and values for each response
A system to treat missing data and to clean data: Where responses are missing, teams may return to respondents for follow-up, or provide a code (i.e. 99) for blank answers. It may be helpful to analyze how frequently answers are missing for each question. For variables where levels of missing information are significant, it may be worth exploring patterns across respondents to identify patterns around who do or do not respond to these questions. Cleaning data also includes an analysis of data to identify missing or potential issues with outliers, deciding how to treat missing data, etc.
Clear and detailed documentation on data management: cleaning, treatment of missing data, and how indices are developed (i.e. a variable on consumption may = food expenditures + housing + education costs)
Quantitative methods to analyze data include:
Descriptive Measures: mean, median, spread (range, standard deviation, variance).
Comparing data across groups: differences in specific variable values (outcomes or processes /behaviors) between different groups. Within CARE, use of comparison has been used to compare between groups of people, as well as to compare how certain variables have changed over time within a community:
- For evaluation or impact measurement, studies may engage communities or groups which CARE has engaged at varying levels. This may include:
- Women and their households who have had varying levels of engagement with CARE projects
- Time (if information from baseline studies could be mined to identify change in women’s empowerment)
- Control groups through ‘desk controls’ or evidence from similar zones where CARE does not work through reports, interviews with informants from other organizations, etc.
- In situational analysis, studies use comparison to understand the varying realities across richer, poorer or excluded village communities.
Estimating the correlations between variables:
- 2 variables: cross-tabulation, chi-square, scattergrams, correlation coefficients, analysis of covariance
- 3+ variables: three-variable contingency tables, partial correlations, multiple regression
Given the myriad of variables that exist, studies can only assert associations between variables, and not causality.
To adapt methods more closely with the context of specific communities, as well as engage more empowering methods, participatory numbers have also been gathered through a spectrum of participatory approaches (see Chambers, 2005):
Other methods for collecting participatory numbers may be through trend and change analysis, seasonal diagramming, scoring, mapping, well-being analysis, etc. Information from these exercises can also be counted and aggregated. However, this approach makes it more difficult to be able to compare across sites, or understand changes in a more generalized fashion.