Methodology Before Measurement
We set the method before we measure. There are no off-the-shelf designs here: for every study we weigh the research purpose, the target group, the geographic scope, the decision-maker’s needs, the available data sources, the time and budget constraints and the nature of the variables to be measured — together.
What comes out of that assessment is the research model best suited to the question. It may be quantitative, qualitative or mixed.
Survey Design
We design questionnaires and implementation models for household, enterprise, employee, student, youth, health, public service and private sector studies.
Sampling Design
We develop probability, stratified, cluster, multi-stage, systematic and mixed sampling structures.
Sample design is never only about sample size: representation, design effect, subgroup analysis, weighting and field feasibility are decided together.
Sample Size & Power
Sample size follows from the research question and the effect we expect to find.
Statistical power analysis is used particularly in impact evaluations, experimental and quasi-experimental studies, health research and group comparisons.
Questionnaire Development
Writing a questionnaire is not only about which questions to ask. It is about the order they are asked in, the wording, the scale and the group they are put to.
Quantitative, Qualitative & Mixed Methods
A research question does not have to be answered with a single method. Three approaches are in regular use at ID.
- Quantitative Research
- Representative surveys, indices, scales, experimental designs and statistical analysis.
- Qualitative Research
- Focus groups, in-depth interviews, expert and stakeholder interviews, case analysis and thematic analysis.
- Mixed Methods
- Explaining quantitative results through qualitative findings, or letting qualitative work shape the quantitative design. Particularly strong in complex social policy and development research.
Experimental & Quasi-Experimental Design
In impact evaluation we do not rely on a before-and-after comparison alone. Depending on the structure of the study we may use randomised controlled trials, control group designs, matching methods, difference-in-differences, propensity score approaches and other quasi-experimental models.
The aim is not to show that a change occurred, but to understand whether the change was actually caused by the intervention.
Measurement & Scale Development
Measuring something that cannot be observed directly — an attitude, a perception, a behaviour, satisfaction — needs particular methodological care.
Scale development, reliability analysis, validity analysis, EFA, CFA, SEM, index construction and scoring systems are used as the study requires. The point is not to produce a score, but to show that what is being scored is genuinely the concept intended.
Weighting & Representativeness
When fieldwork is complete, not every observation carries the same representational weight. Design weights, non-response adjustment, calibration, post-stratification and raking are applied where needed.
In national and regional studies, weighting is what allows results to be generalised correctly to the target population.
Methodology in the Field
Methodology is tested in the field. A good design does not only have to be theoretically correct; it has to be applicable.
At ID, research methodology and field operations are not separate worlds. The sample design is tested against field access, interview duration, target group behaviour, regional differences, cost and time. Where a pilot shows it is necessary, the design is revised.
Academic Rigor, Applied Research
Our approach rests on the meeting of two things: academic knowledge of method, and real field experience.
We ground our models in the methods of the statistical, epidemiological, sociological, economic and public policy literature — and then make them workable in the field and usable for a decision-maker.
Methodological Advisory
We can join a project as a methodological adviser rather than as the field implementer.
That covers research framework development, sample design review, questionnaire review, indicator development, power analysis, weighting strategy, analysis plans and impact evaluation design — either as standalone advice or as part of a full study.
