CRM/LAW C162 Lecture Notes - Lecture 6: Aphorism, Randomized Controlled Trial, Experiment
6 Apr 2017
School
Department
Course
Professor
Document Summary
Makes a lot more sense to prevent then to react to crime. Negativa non sunt probandan (means finding nothing proves nothing: kansas said prevenative patrolling was ineffective but it actually proved nothing. Sherman and weisburd: difference between true experiment and randomize trial (a trial has a placebo control) these experiments really are not randomized, placebo fools subject and experimenter, police still patrol because response time is faster. The design of the kansas city experiment does not justify the strong conclusion that preventive patrol has no effect on street crime: the statistical problem: low statistical power makes it difficult to find the effect of. Statistical problem if the(cid:455) (cid:272)o(cid:374)(cid:448)i(cid:272)t a(cid:374) i(cid:374)(cid:374)o(cid:272)e(cid:374)t that"s the false positi(cid:448)e error (cid:894)we tr(cid:455) to keep it at 5%(cid:895) On other hand if jury acquits an innocent defendant 80% statistical power error: if they convict a guilty they have 95% confidence in verdict.
