Guide-linked quiz · Medium Level

Probabilistic Genotyping and AI in Crime Labs Quiz

Test probabilistic genotyping: mixture interpretation, likelihood ratios, propositions, drop-in and drop-out, validation, transfer, and limits.

Question 1 of 8

A software report gives a large likelihood ratio. What does the ratio compare?

Preview every question8 questions

Answers and explanations appear as you take the quiz.

  1. A software report gives a large likelihood ratio. What does the ratio compare?
    • The probability that the suspect is guilty
    • The chance the laboratory made no error
    • The probability of the observed DNA evidence under two stated propositions
    • The probability that DNA was deposited during the crime
  2. Why is an LR not a probability of guilt?
    • It evaluates evidence under propositions and does not combine all evidence or decide guilt
    • It is calculated without DNA data
    • It always assumes the suspect is innocent
    • It measures only the analyst’s confidence
  3. What does DNA drop-out mean in mixture interpretation?
    • A DNA contributor intentionally removes a sample
    • The software deletes a person from the database
    • A peak appears when no allele is present
    • An allele that is present may fail to be detected in the measured profile
  4. Why can two laboratories obtain different mixture results?
    • DNA evidence has only one possible interpretation
    • Software, configuration, models, and interpretation assumptions can differ
    • Likelihood ratios do not use propositions
    • One laboratory must always be wrong
  5. A source-level LR supports that a person may have contributed DNA. What remains unresolved?
    • Whether DNA was detected at all
    • Which propositions were compared
    • When or how the DNA was deposited
    • Whether a profile contained multiple contributors
  6. What should a laboratory do before using a software update in casework?
    • Assume the update preserves all prior behavior
    • Use it only on the most complex sample
    • Delete earlier results to avoid comparison
    • Validate the update and document version, configuration, and performance
  7. Which validation set is most useful for a laboratory?
    • Known mixtures that represent relevant contributor counts and conditions in its casework
    • Only easy single-source samples
    • The same unknown case evidence used to tune the model
    • Examples chosen by a software vendor without local review
  8. Why must a report state the competing propositions?
    • Propositions identify the jury’s verdict
    • The LR’s meaning depends on the hypotheses being compared
    • They remove model uncertainty
    • They prove when the sample was deposited

Keep testing yourself

More quizzes picked for your level.

View all quizzes