Our research process

The research process comprises five major stages, with each phase providing the foundations for the phase that follows. Design choices at particular phases impact on the phases that follow and the potential research questions that might be answered. The phases include:  research problem formulation; research design formulation; identifying data and data collection; data coding and analysis; and drawing inferences and recommendations.

Phase 1: Problem formulation

Good design in the problem formulation phase is critical to the success of a research project. The care taken in defining a research problem reflects in the quality of answers the research is able to produce. Good problem formulation successfully transforms a problem or topic area into real world measures and determines the entire research process. 

This stage comprises a scoping phase and literature search to ensure all constructs central to the research enquiry are effectively captured and measured. Close collaboration with clients during this critical phase establishes the foundations for the entire research process. 

Scoping – During this phase we collaborate with clients to establish the context of the research enquiry and the outcomes they wish to achieve. 

Literature review – positions the research problem within what is currently understood about the field from relevant scientific or industry publications as well as existing internal reports and data assets. This enables development of the research problem into an appropriate research question and ensures concepts fundamental to addressing the research question are fully conceptualised.  

Time investment at this critical stage ensures the research is:

• Relevant – conveying the research benefit and importance and alignment to strategic and policy objectives

• Comprehensive – identifying all concepts and terms required to address the research question

• Effective – ensuring research outputs can be turned into interpretable, actionable information.

• Feasible (researchable) – based on whether data can be acquired

• Appropriately establishes the boundaries of the project in what is answered and what is not.

Phase 2: Formulating the research design 

Based on the problem formulation stage, we will advise on the most appropriate research design. The design may comprise one or a mixture of qualitative or quantitative methodologies and incorporate, where possible, existing data assets. This stage will also establish the most appropriate target populations for the analysis as well as sampling methods. 

For qualitative approaches, we will assess which formats are likely to yield the most appropriate insights (e.g. focus groups may be appropriate to gain insights into general information but in-depth interviews are more appropriate specialist knowledge or sensitive issues - which can range from business in confidence through to personal or cultural issues). 

For quantitative methodologies, sampling strategies take into careful consideration accessibility of the target population in order to maximise representativeness and generalisability for statistical inference. For sub-populations that are important to the research problem, web based surveys may not be appropriate if a key sub-population cannot be readily accessed.

Examples include: people with limited literacy, individuals who are highly mobile such as truck drivers, or people living in remote locations, and so on. Appropriate sampling strategies and collection methods are considered such as us quota sampling to ensure appropriate representation of small but important sub-populations and alternative or supplemental collection methods such as paper-based questionnaires, telephone or face-to-face interviewing.

Phase 3: Identifying data and data collection Methods

Data already being collected may have the potential to fully or in part address the research question. With the wide availability of data analytics solutions, many organisations already employ automated reports, dashboards and alerts in order to provide ready access to indicators of performance. 

This phase can also provide recommendations for improvements for routine data collections being undertaken. Even modest design adjustments to what data are routinely collected or how data bases are coordinated and communicated can yield substantial gains for reporting insights.  

After examining existing data collections, our consultants may gather further data with more traditional collection methods from secondary and primary sources to ensure the research problem can be appropriately addressed and that data is representative of target populations. 

Secondary sources may include publicly available data such as published statistics, socio-economic data and industry trends. 

For primary data collection, our consultants will custom design data collection instruments that operationalise (design measures for) the gaps in knowledge established in the literature review. Instrument selection and design will depend on the nature of the subject matter and the target populations being assessed. Examples of data collection instruments may include structured questionnaires undertaken as postal or web-survey, telephone surveys or face-to-face, observational techniques, focus groups, through to in-depth interviews.

Phase 4: Data coding and analysis

For qualitative data, coding and analysis involves drawing out key themes from the data which may be explicitly expressed or may be implicit, requiring considered judgment from the researcher in uncovering themes and in minimising bias in reporting.

For quantitative analysis, data are audited and transformed ready for the analysis phase. Auditing of data for potential errors, such as out of scope or unexpected values, can require considered judgement by the analyst as such values may elude to systematic or larger problems with the data set that require further investigation.

Transformation for data requires grouping of variable categories that may be sparse (have few responses), data that may not be normally distributed (e.g. skewed), or secondary variables may need to be constructed from original variables within the data set. Missing data can require investigation for potential bias (e.g. is a particular sub-group in the population not represented or not answering questions).

For complex data, entire layouts may also require transformations such as collapsing (summing or averaging over subgroups) or reshaping between wide and long formats. If more advanced analysis techniques are used, data also needs to be inspected for potential model violations such as high correlations (collinearity), outliers, and uneven distribution of observations (heteroscedasticity) that can bias results in a model.

Once the data is “clean” the data is ready for the analysis phase. Our consultant will recommend the most appropriate analysis approaches depending on the nature of the data and research problem being answered.

Phase 5: Drawing inferences and recommendations

In this phase, findings are placed in context of the literature review and the entire scoping process. This ensures that implications drawn from the research lead to astute and pragmatic recommendations.