Assessment and Analysis
Purpose before method
Assessment exists to understand a situation before acting on it. Questions in this domain reward candidates who select a data-collection method by matching it to the question being asked, not by picking the cheapest or most familiar tool.
Concepts you must hold
- Data types and sources: primary versus secondary; quantitative (how many, how much) versus qualitative (why, how); existing records versus newly collected data.
- Instrument quality: validity (measures what it claims) and reliability (produces consistent results). A reliable instrument can still be invalid; an invalid one cannot rescue a conclusion.
- Sampling: representativeness, sample size, self-selection and convenience-sample bias, response bias.
- Triangulation: confirming a finding across independent sources or methods.
- Analysis tools: gap analysis (current versus desired state), root-cause tools such as the five-whys and cause-and-effect diagrams, and prioritisation matrices that weigh impact against effort.
- Documentation: state assumptions, limitations, and confidence explicitly.
Traps
Confusing assessment (pre-action, understanding) with evaluation (post-action, judging). Treating satisfaction data as outcome data. Generalising from a small or self-selected group. Asserting causation from a correlation. Choosing a method because it is easy rather than because it answers the question.
Rule of thumb
Match method to question: "how many / how much" points to quantitative methods; "why / how" points to qualitative methods; "what is really causing this" means push past the presenting symptom to a root cause before recommending action.
Worked example
A satisfaction score falls in the same month a new tool is introduced, and a stakeholder concludes the tool caused the drop. The correct analysis is that timing alone establishes correlation, not causation; the analyst should triangulate with usage data, compare against a period unaffected by the change, and check for competing explanations before concluding.