Answer: A. Correct errors and inconsistencies. Cleaning checks data for errors, duplicates, inconsistent entries, and other quality issues.
How can researchers code open-ended answers?
A Delete every unusual response
B Replace words with names
C Count only the longest answer
D Group responses into clear themes
Answer: D. Group responses into clear themes. Coding assigns responses to categories or themes using documented rules.
What should analysts do with missing data?
A Replace blanks with guesses
B Assess patterns and handling choices
C Assume every blank is zero
D Delete all incomplete records
Answer: B. Assess patterns and handling choices. Missingness should be examined because its pattern can affect conclusions and appropriate methods.
Which measure resists extreme values better?
A Standard deviation
B Range
C Median
D Mean
Answer: C. Median. The median is less affected by extreme observations than the mean.
Which results appear in a cross-tabulation?
A A website’s loading path
B A single average only
C A focus group’s audio
D Counts across category combinations
Answer: D. Counts across category combinations. Cross-tabs show how two or more categorical variables are distributed together.
Why does correlation not establish causation?
A Variables cannot move together
B Other explanations may exist
C Experiments never use data
D Correlation is always negative
Answer: B. Other explanations may exist. A correlation can reflect confounding, reverse direction, or coincidence rather than a causal effect.
What can regression analysis estimate?
A Relationships among measured variables
B A guaranteed causal effect
C A perfect forecast
D Every customer’s motive
Answer: A. Relationships among measured variables. Regression estimates associations or predictions, with interpretation depending on assumptions and design.
What does a p-value describe?
A Probability a hypothesis is true
B Size of a marketing effect
C Chance of a customer purchase
D Compatibility with a null model
Answer: D. Compatibility with a null model. A p-value measures how unusual data would be under a specified null model and assumptions.
How does a confidence interval communicate uncertainty?
A A proof of causation
B A guaranteed population boundary
C A range of plausible estimates
D A response rate
Answer: C. A range of plausible estimates. A confidence interval communicates uncertainty around an estimate under the method’s assumptions.
Which variants can an A/B test compare?
A A product and trademark
B Two experience variants
C A forecast and budget
D Two unrelated populations
Answer: B. Two experience variants. An A/B test compares variants using a chosen outcome measure.
Why randomize participants in an experiment?
A Remove every measurement error
B Guarantee identical outcomes
C Increase response bias
D Reduce systematic group differences
Answer: D. Reduce systematic group differences. Random assignment helps balance other influences across groups on average.
Which group provides an experimental comparison condition?
A Collect only positive comments
B Set the campaign budget
C Provide a comparison condition
D Replace the treatment group
Answer: C. Provide a comparison condition. A control group gives researchers a comparison for interpreting the treatment or intervention.
Why create a holdout group?
A Estimate outcomes without exposure
B Collect customer passwords
C Measure package durability
D Increase repeated impressions
Answer: A. Estimate outcomes without exposure. A holdout group is kept out of an intervention to help estimate what might happen without it.
What does effect size communicate?
A A study’s interview count
B A question’s reading level
C A sample frame’s coverage
D Magnitude of an observed difference
Answer: D. Magnitude of an observed difference. Effect size communicates the magnitude of a difference or relationship, beyond whether it is detectable.
Which dashboard view best shows a trend?
A Values plotted over time
B A list without dates
C One unlabeled total
D A logo collage
Answer: A. Values plotted over time. A time-series view helps readers see how a measure changes across periods.
What should an executive summary highlight?
A Only the study title
B All coding notes
C Every raw response
D Key findings and implications
Answer: D. Key findings and implications. An executive summary presents the most important results, implications, and relevant limits.
How does an insight differ from a finding?
A It removes uncertainty
B It is a larger sample
C It explains meaning and implication
D It is a survey question
Answer: C. It explains meaning and implication. A finding reports evidence; an insight interprets why it matters for a decision.
What makes a recommendation actionable?
A A broad inspirational phrase
B Clear owner and next step
C A list of raw data
D An unsupported prediction
Answer: B. Clear owner and next step. Actionable recommendations specify a decision or step that a team can realistically take.
Why report sampling limitations?
A Avoid explaining methods
B Guarantee acceptance
C Flag limited generalizability
D Undermine every finding
Answer: C. Flag limited generalizability. Limitations help readers judge the scope and transferability of findings.
How should researchers protect participant privacy?
A Limit access and exposure
B Publish identifiable responses
C Share raw records widely
D Collect unrelated details
Answer: A. Limit access and exposure. Access controls and limited collection help reduce unnecessary exposure of participant information.
Why obtain informed participant consent?
A Explain participation and data use
B Replace privacy safeguards
C Guarantee favorable answers
D Remove the right to leave
Answer: A. Explain participation and data use. Consent helps participants understand the study and the way their information will be handled.
Which identifiers does de-identification seek to remove?
A All missing records
B Every response difference
C Direct personal identifiers
D All analytical value
Answer: C. Direct personal identifiers. De-identification removes or transforms identifiers that could directly link data to a person.
Which practice supports transparent analysis?
A Document decisions and methods
B Change metrics after results
C Hide excluded records
D Report only favorable segments
Answer: A. Document decisions and methods. Documented methods and decisions make analyses easier to review and reproduce.
Which practice is called p-hacking?
A Pretesting survey wording
B Searching analyses until significance appears
C Checking data accuracy
D Using a random sample
Answer: B. Searching analyses until significance appears. P-hacking selectively searches analyses or outcomes for statistical significance, inflating false-positive risk.
Why preserve a clear research audit trail?
A Replace participant consent
B Enable review of decisions
C Guarantee a significant result
D Prevent all data changes
Answer: B. Enable review of decisions. An audit trail records key decisions, versions, and methods for later review.