Unit 3: Analytical and Critical Thinking

I. Orientation

Analytical and critical thinking provide the reasoning foundation for design thinking and complex problem solving. Analytical thinking breaks a complex situation into manageable elements, while critical thinking evaluates the quality of evidence, assumptions, interpretations, and conclusions. Together, they support decisions that are transparent, defensible, and responsive to uncertainty.

• Governing principle: A good decision connects a clearly defined problem to relevant evidence, explicit criteria, logical analysis, and proportionate action.
• Analytical thinking: Separates a system into components, identifies relationships, detects patterns, and compares alternatives.
• Critical thinking: Questions assumptions, tests claims, recognizes limitations, and distinguishes evidence from opinion.
• Complexity assumption: Problems may have multiple causes, stakeholders, feedback loops, uncertain data, and consequences that emerge over time.
• Decision convention: Conclusions should be conditional on stated assumptions; a change in evidence or priorities may justify a change in decision.
• Evidence hierarchy: Direct, relevant, reliable, and independently corroborated evidence is generally stronger than anecdote, intuition, or unsupported assertion.

II. Analytical Reasoning

A. Application of analytical reasoning and critical thinking for evaluating alternatives

Application of analytical reasoning and critical thinking for evaluating alternatives means systematically comparing possible actions against explicit objectives, constraints, evidence, and consequences.

• Define the decision: State what must be chosen, by whom, and by when; “select a transport option for a 12-month pilot” is more analyzable than “improve transport.”
• Separate facts from assumptions: A fact may be “the budget is $50,000,” while an assumption is “users will adopt the service if it is free.”
• Identify criteria: Common criteria include cost, effectiveness, equity, feasibility, risk, sustainability, and user acceptance.
• Compare consequences: Examine both first-order effects, such as lower cost, and second-order effects, such as reduced quality or increased maintenance.
• Use proportional reasoning: A high-cost, irreversible decision requires stronger evidence than a low-cost, reversible experiment.
• Make reasoning visible: Record the evidence, weighting, assumptions, uncertainty, and rationale rather than presenting only the final choice.
• Evaluate alternatives jointly: A solution that performs well on cost but poorly on safety should be compared with balanced alternatives, not judged by cost alone.

B. Applications and limitations

Analytical reasoning is most useful when alternatives and evaluation criteria can be made explicit, but it cannot eliminate uncertainty or value conflicts.

• Application—weighted comparison: If cost has weight 0.4, impact 0.4, and feasibility 0.2, an option scoring 8, 7, and 6 respectively receives:
  
  TEXT

    Overall score = (0.4 × 8) + (0.4 × 7) + (0.2 × 6) = 7.2

  
  Weights represent decision priorities; scores should use a defined scale, such as 1–10.
• Limitation—false precision: A score of 7.2 does not prove that an option is exactly 20% better than one scoring 6.0.
• Limitation—unmeasured values: Cultural appropriateness, dignity, or trust may be difficult to quantify but still ethically important.
• Critical safeguard: Conduct sensitivity analysis by changing uncertain assumptions and checking whether the preferred alternative changes.

III. Cognitive Biases

A. Understanding cognitive biases

Understanding cognitive biases involves recognizing systematic tendencies that distort perception, judgment, and choice, especially when people face uncertainty, time pressure, or emotionally significant outcomes.

• Confirmation bias: People favor information supporting an existing belief. A project team may seek only positive user comments after deciding that a prototype is successful.
• Anchoring bias: An initial number influences later judgments. A quoted price of $10,000 can make $8,000 seem cheap even when the market value is $5,000.
• Availability bias: Easily recalled events seem more probable. A recent cyberattack may cause a team to overestimate that risk relative to less visible operational failures.
• Overconfidence bias: Decision-makers overestimate their knowledge or control, producing narrow forecasts and insufficient contingency planning.
• Sunk-cost effect: Past investment improperly influences future action. Continuing a failing project because $200,000 has already been spent ignores whether future benefits justify future costs.
• Framing effect: Equivalent information produces different choices depending on presentation; “90% survival” may appear more attractive than “10% mortality.”
• Groupthink: Desire for agreement suppresses dissent, reducing the examination of alternatives and warning signs.
• Hindsight bias: After an event, its outcome appears predictable, making learning inaccurate unless decisions are assessed using information available at the time.

B. Applications and limitations

Bias awareness improves judgment when it is converted into deliberate procedures rather than treated as a claim that other people are biased.

• Pre-mortem analysis: Assume the chosen plan failed and list plausible causes; this counters overconfidence and encourages attention to weak signals.
• Devil’s advocate: Assign a person to construct the strongest case against the favored option, reducing confirmation bias and groupthink.
• Independent estimates: Obtain forecasts before group discussion to reduce anchoring and conformity.
• Reference classes: Compare a proposed project with outcomes from similar completed projects instead of relying only on an optimistic internal forecast.
• Blind evaluation: Remove irrelevant identity information when assessing applications or proposals, reducing status and similarity biases.
• Limitation—bias cannot be removed completely: Procedures reduce predictable errors but cannot guarantee objectivity; values, incomplete data, and genuine ambiguity remain.

IV. Decision-Making Approaches

A. Decision-making approaches

Decision-making approaches are structured ways of selecting action, ranging from fully analytical models to approaches designed for uncertainty, urgency, or competing stakeholder values.

• Rational or analytical approach: Define the problem, generate alternatives, establish criteria, assess evidence, select, implement, and review. It works best when objectives and data are reasonably clear.
• Bounded rationality: People choose a satisfactory option because time, information, attention, and computational ability are limited. “Good enough” may be more realistic than optimization.
• Incremental decision-making: Make small, reversible changes and learn from feedback; this is appropriate when causes are uncertain and large commitments are risky.
• Intuitive or expert judgment: Experienced practitioners rapidly recognize patterns, but intuition is more dependable in stable environments with regular feedback than in novel situations.
• Participatory decision-making: Involves affected stakeholders to improve legitimacy, local knowledge, implementation, and acceptance.
• Multi-criteria decision-making: Combines different dimensions, such as cost, safety, environmental impact, and equity, using a transparent scoring or ranking process.
• Expected-value reasoning: When probabilities and outcomes can be estimated, compare alternatives using:
  
  TEXT

    Expected value = Σ (probability of outcome × value of outcome)

  
  The symbol Σ means “sum across all possible outcomes.”
• Ethical decision-making: Tests options against duties, consequences, fairness, rights, and the interests of vulnerable groups rather than relying only on efficiency.

B. Applications and limitations

Selecting an approach should depend on the decision environment, not on loyalty to a single model.

• Stable and measurable setting: Use analytical comparison when reliable performance data and clear objectives exist, such as selecting among suppliers.
• Uncertain and changing setting: Use pilots, staged investment, and feedback loops when future conditions cannot be predicted confidently.
• High-stakes setting: Combine quantitative analysis with ethical review, stakeholder consultation, and independent challenge.
• Emergency setting: Use predefined thresholds and delegated authority because delay may create greater harm than imperfect information.
• Limitation—trade-offs are unavoidable: A decision may improve efficiency while reducing equity; the chosen balance must be justified rather than hidden inside a score.

V. Structured Analysis Techniques

A. Structured analysis techniques

Structured analysis techniques organize complex information into visible components, relationships, causes, priorities, and decision paths.

• Issue tree: Break a broad question into mutually distinct branches. “Why are deliveries late?” may divide into supplier delay, processing delay, transport delay, and demand fluctuation.
• MECE principle: Branches should be mutually exclusive and collectively exhaustive as far as practical; overlapping categories can double-count causes.
• Five Whys: Repeatedly ask why to move from symptom to underlying cause. If “users abandon registration” leads to “the form is too long,” investigation should continue to the design or policy cause.
• Fishbone diagram: Organizes possible causes under categories such as people, process, technology, materials, measurement, and environment.
• SWOT analysis: Records strengths and weaknesses inside an organization and opportunities and threats in its environment; it is exploratory, not proof of causation.
• Pareto analysis: Ranks causes or categories by contribution. A bar chart may show that 70% of complaints arise from two of eight service failures.
• Decision matrix: Lists alternatives by criteria, assigns scores and weights, and calculates comparative totals; it makes priorities inspectable.
• Systems mapping: Shows actors, resources, delays, feedback loops, and unintended consequences, helping analysts avoid treating an interconnected problem as linear.

B. Applications and limitations

Structured tools are useful for organizing thought and discussion, but their outputs depend on the quality of categories, data, and assumptions entered into them.

• Technique selection: Use an issue tree for scope, Five Whys for causal exploration, a matrix for alternatives, and a systems map for interdependence.
• Triangulation: Apply more than one technique; a fishbone diagram can generate causes, while process data tests whether those causes actually occur.
• Avoid premature closure: Treat early diagrams as hypotheses and revise them when interviews, observations, or measurements contradict them.
• Worked example—decision matrix: For three options, calculate each weighted criterion:
  
  TEXT

    Weighted score = Σ (criterion weight × option score)

  
  A weight of 0.5 for effectiveness and 0.5 for cost means the organization values both equally; changing the weights tests how priorities affect the result.
• Limitation—visual order can mislead: A neat causal diagram may suggest certainty where relationships are only suspected.

VI. Evidence-Based Evaluation Methods

A. Evidence-based evaluation methods

Evidence-based evaluation methods determine whether a claim, intervention, or alternative is effective by specifying indicators, collecting appropriate data, and comparing observed results with a credible baseline or counterfactual.

• Operationalize outcomes: Convert broad aims into measurable indicators; “better access” could become average waiting time, completion rate, or proportion of underserved users served.
• Baseline measurement: Record conditions before intervention. If average waiting time is 40 minutes before implementation, later values can be compared with that starting point.
• Process evaluation: Examines whether an intervention was implemented as intended, including reach, participation, fidelity, and resource use.
• Outcome evaluation: Measures short- or medium-term changes attributable to the intervention, such as reduced errors after a redesigned form.
• Impact evaluation: Investigates longer-term effects and attempts to separate intervention effects from external influences.
• Randomized comparison: Random assignment to intervention and control groups can reduce selection bias when ethically and practically feasible.
• Quasi-experimental comparison: Difference-in-differences compares changes over time between an intervention group and a comparable non-intervention group:
  
  TEXT

    Effect estimate = (After − Before)intervention − (After − Before)comparison

  
  “After” and “Before” are outcome measurements at two time points.
• Qualitative evidence: Interviews, observations, and open-ended responses explain how and why outcomes occurred, complementing numerical indicators.
• Data quality checks: Assess validity, reliability, completeness, timeliness, sampling, and potential measurement bias before interpreting results.

B. Applications and limitations

Evidence-based evaluation supports learning and accountability when findings are interpreted in context and linked to decisions.

• Use mixed methods: Combine a numerical change in waiting time with interviews explaining whether the change resulted from staffing, seasonality, or user behavior.
• Report uncertainty: Use ranges, confidence intervals, or cautious language when samples are small or measurements are variable; a single average can conceal unequal effects.
• Check equity: Disaggregate results by relevant groups, such as age, income, disability, or location, because an overall improvement may coexist with harm to a subgroup.
• Distinguish correlation from causation: Two variables changing together does not establish that one caused the other; consider alternative explanations and comparison evidence.
• Evaluate unintended consequences: A policy that reduces processing time but increases exclusion or error may not represent genuine improvement.
• Limitation—generalizability: Results from one location, population, or period may not transfer automatically to another; assess contextual similarity before scaling.
• Decision rule: Continue, adapt, pause, or stop an intervention according to predefined indicators, thresholds, stakeholder impacts, and the strength of the evidence.