Data analysis
As with other stages in the evaluation process, the analysis of evaluation data, in quantitative or qualitative form, requires the cooperation of analysts, data generators and responsible authorities able to resolve queries and guarantee appropriate, high-quality results.
See the steps necessary for data analysis
Types of statistical tests
There are several types of statistical processes for data analysis, including:
- Analysis of qualitative variables (frequency analysis, percentages, cross tabulation).
- Analysis of quantitative variables (means, mode, median and comparison of means).
- Testing hypotheses.
- Defining composite indicators and parametric and non-parametric tests.
The choice of statistical test type will depend on the hypotheses established. The use of less complex statistics is recommended to demonstrate the claim in question, simply bearing in mind that the results are not always read by statistical experts, and it is consequently important to use clear, simple language.
Example of the analysis of proportions comparing groups at a single time
Example of the analysis of proportions comparing groups at a single time
The table shows an example of evaluation in Nicaragua where an analysis of proportions was carried out comparing groups at a single time (a single survey carried out), regarding the titling of the property. Considering that RCT activities are monopolistic state functions, ascribing the effects of the intervention directly is unambiguous.
COMPARISON GROUP ESTABLISHED | WITH VALID DOCUMENTS |
|---|---|
Group 1 | 1% |
Group 2 | 14% |
Group 3 | 85% |
This analysis compares three groups at once, reflecting three different stages in the project according to the intervention territories in which group 1 can be considered the control (no intervention as yet). The situation of group 1 can therefore be considered the ex ante stage, and groups 2 and 3 ex post intervention groups. It can be concluded from this that the programme has a high impact on the legal regulation of property, considering that only 1% of the population has the capability to carry out titling themselves. In territories where the RCT process has started, this figure is increased to 14% and in those where the state has had a widespread influence on titling, the result is as much as 85%. Transposing these results over time gives:
Example of the analysis of gender equity
Example of the analysis of gender equity
To test the gender equity hypothesis in access to legal certainty, an analysis of compared groups can be carried out, with surveys carried out at two separate times. In this case two indicators can be used that prove the same thing: the first is the proportion of women who believe they are the owner, and the second the relation between women and men who believe they are owners. The groups compared are on the one hand those who believe they are owners, even though they have no documents to prove it, taking as the initial condition the stage prior to the RCT process in LAP intervention territories, and on the other the control group of those who believe they are owners who are not taking part in the programme. In this case the assumption was that: “as proprietors, more women declare themselves to be owners but in tenure documents it is men who believe they are proprietors or owners”.
In the case of this example, this claim is true for the non-intervention group in the programme as in 40.6% of cases women assume they are owners while only 33.7% of them have legal certainty (a legal document in their name). This situation is reversed for the treatment group as, of the 42% of women who consider themselves to be owners, 45% now have legal documents in their name.
