Full-Stack Development Research & Evidence-Based Practice 5 — Questions and Answers
Question 1: A senior developer argues against writing tests because 'we've always shipped without them.' This reasoning is an example of:
- Appeal to authority
- Appeal to tradition (argumentum ad antiquitatem) (Correct answer)
- Confirmation bias
- Anchoring bias
Correct answer: Appeal to tradition (argumentum ad antiquitatem)
Appeal to tradition mistakes historical practice for evidence of correctness, ignoring whether the outcome would have been better with the alternative.
Question 2: Which method is most appropriate for discovering unknown usability problems in a new web application before its first major release?
- Reviewing server-side error logs only
- Moderated usability testing with representative users performing realistic tasks (Correct answer)
- Asking the development team to click through the app
- Reading competitor product reviews
Correct answer: Moderated usability testing with representative users performing realistic tasks
Moderated usability testing with actual representative users surfaces problems that internal testers—familiar with the system—systematically miss.
Question 3: When adopting a new security practice based on an industry report, what should a developer verify about the report's evidence?
- Whether the report has an attractive visual design
- The sample size, industries studied, methodology, and whether it was independently audited or sponsored by a vendor (Correct answer)
- Whether the report author has a large social media following
- That the report is fewer than 12 months old regardless of methodology
Correct answer: The sample size, industries studied, methodology, and whether it was independently audited or sponsored by a vendor
Vendor-sponsored reports and those with undisclosed methodologies or small samples can introduce significant bias into security recommendations.
Question 4: A developer wants to measure the business impact of reducing page load time. What metric most directly connects performance to business outcomes?
- Lighthouse performance score
- Bounce rate and conversion rate correlated with load time segments (Correct answer)
- Number of JavaScript errors per session
- Server CPU utilization
Correct answer: Bounce rate and conversion rate correlated with load time segments
Correlating user behavior metrics (bounce, conversion) with load time segments directly links performance improvements to revenue-relevant outcomes.
Question 5: A post-mortem analysis after a production outage should prioritize which approach?
- Identifying and punishing the engineer who made the change
- Blameless root cause analysis focused on systemic and process failures (Correct answer)
- Documenting a timeline and then closing the ticket quickly
- Reverting all recent changes without analysis
Correct answer: Blameless root cause analysis focused on systemic and process failures
Blameless post-mortems focus on process and system failures rather than individual blame, producing actionable improvements and psychological safety.
Question 6: A team is deciding between two state management libraries and finds a three-year-old Stack Overflow answer recommending Library A. What is the most appropriate use of this evidence?
- Adopt Library A immediately because the community endorsed it
- Treat it as a historical data point and verify whether the recommendation still holds given current ecosystem maturity and alternatives (Correct answer)
- Discard it entirely because it is more than one year old
- Post a new question asking for a current recommendation without checking recent activity
Correct answer: Treat it as a historical data point and verify whether the recommendation still holds given current ecosystem maturity and alternatives
The JavaScript ecosystem evolves rapidly; older community consensus should be re-evaluated against current library health, bundle size, and ecosystem support.
Question 7: Which of the following best describes 'construct validity' when measuring developer productivity?
- Whether the measurement tool is calibrated correctly
- Whether the chosen metrics (e.g., lines of code) actually capture the underlying concept of productivity (Correct answer)
- Whether developers can reproduce the same measurement twice
- Whether productivity scores follow a normal distribution
Correct answer: Whether the chosen metrics (e.g., lines of code) actually capture the underlying concept of productivity
Construct validity asks whether the operationalized measure (e.g., LOC, PRs merged) truly reflects the abstract concept you intend to measure.
Full-Stack Development Research & Evidence-Based Practice 5
A senior developer argues against writing tests because 'we've always shipped without them.' This reasoning is an example of: