INTR STATISTCL REASONING
INTR STATISTCL REASONING STAT 110
Popular in Course
Popular in Statistics
verified elite notetaker
This 10 page Class Notes was uploaded by Shane Marks on Monday October 26, 2015. The Class Notes belongs to STAT 110 at University of South Carolina - Columbia taught by G. Baker in Fall. Since its upload, it has received 9 views. For similar materials see /class/229647/stat-110-university-of-south-carolina-columbia in Statistics at University of South Carolina - Columbia.
Reviews for INTR STATISTCL REASONING
Report this Material
What is Karma?
Karma is the currency of StudySoup.
Date Created: 10/26/15
Experiments Good and Bad Chapter 5 How do these two studies differ I In Aiken SC a store manager wants to know which brand of coffee is most liked among the 1824 yearold population I I I Physicians on a Drug an a ty Monimring Board want to determine which f rugs is more effective in trmting HIV in the early Stags of the dise ase lg Observational or Experimental I An observational study obsenes individuals and measures variables of interst but oe s not attempt tn in uence the responss S I An experiment deliberately impose some tr 39 eatment on individuals in order tn affect the responses More Examples of Experiments I Does aspirin reduce the rate of heart attacks I Are beetles attracted to bright colors I Can we make a stronger paper bag by increasing the percentage of hardwood content I Does class size affect student performance in STAT 110 The Vocabulary of Experiments I Subject individual studied in an experiment I Treatment experimental condition applied to subjects I Response Variable variable that measures an outcome or result of a study Dependent Variable I Explanatory Variables variable that we think explains or causes changes in the response variable Independent Va riables a mu Depzvlmerl ufslznxnz u nnnn my Smlh mum Slide 5 Experiment Example Consider the Lowly Grocery Bag IWe want a strong bag to hold heavy groceries IWe think that the percentage of hardwood that goes into the pulp from which the bags are made affects the strength IWe make multiple batches of bags from pulp with 5 10 15 and 20 hardwood and measure the strength a mu Depzvlmerl ufslznxnz u nnnn my Smlh mum Slide 5 Experiment Example Consider Lowly Gro I Identify the response variable I Identify the explanatory variable I Identify the treatmenls I Identify the subjecls Experiment The Effect of Class Size on Performance in STAT 110 IThere are STAT 110 classes with 50 and 200 studenls I Will class size affect student rforrnanoe I How will we measure student performance Beware of Confounding Variables confounding Variable lurking riable is a variable mamas an te e Two variables are confounded when eir effecE on e reaaonse variable cannot be distinguished from one another When we let a student sign up for the STAT 110 class of hisher choice we increase the risk of confounding variables I Why lWhat might some confounding variables be a my Depzvlmerl ufslznxnz Univuxilvofsouh Ovulim Slide in Are there possible confounding variables in the paper bag experiment lWhat might some be I How do we decrease the likelihood of confounding variables affecting our experimental results a my Depzvlmerl ufslznxnz Univuxilvofsouh Ovulim Slide n Moral IWe reduce the likelihood of confounding variables affecting experimental results by Randomly placing subjects into groups and Randomly assigning treatments to groups a 5mm Univuxilv of 5m Slide 2 Clinical Trials I Clinical Trials are experiments that study the effect of medical treatments on actual patients I Placebo is a dummy treatment with no active ingredients I Placebo effect is a patient s favorable response to a dummy treatment a mu Depzvlmerl ofSlzliXliz Univemlyofsouh Ovolin25lide13 Clinical Trials Randomized Comparative Experiments I Use Simple Random Samples to form two or more groups I Randomly assign treatments to groups I Compare two or more groups I One group is the control group It receives a placebo or already proven treatment a mu Depzvlmerl ofSlzliXliz Univemlyofsouh Ovolim Slide u Randomized Comparative Experiment Pain Reliever A Group Compare Pain RelieverB Group a pain Why Randomized Comparative Experiments Work I Randomization produces groups of subjects that should be similar in all respects before treatment is applied We say we randomize out the effect of confounding lurking variables I Confounding variables may be both known and unknown h F Univemlyofsouh swim 5in 15 Why Randomized Comparative Experiments Work I Comparative design compares treatments to each other and to a control group I If the groups are equal to each other in all respects except for the treatments then the treatments must be responsible for any differences between the groups a mu Devzvlmerl ofSlzliXliz nivuxilyofsouh swim 5in n Clinical Trial Phases I PhaseI Laboratory testing has been completed Small number of subjects Evaluate safety and dosage I Phase II Larger number of subjects Evaluate effectiveness and safety I Phase III Large groups of subjects Con rm effectiveness and identify side effects I Phase IV Studies conducted after FDA approval treatment is on the market a as ViDevzvlmerI ofSlzliXliz Univemlyofsouh swim 5in 8 Number of Subjects in a Group I Use same number of subjects in each group I Use enough subjects so that randomization eliminates any systematic differences between groups a my Depzvlmerl ofsunxnz Univemlyofsouh Ovolim Slide is So what if we nd a difference between a treatment group and the control group I Difference in what Group average or proportion I Remember groups are only samples of a larger population so Group averages and proportions are statistics and they will change if you change the members of the group Worm Ovolim Slide zn The Real Question is Are the response differences between groups large enough to indicate something more than the expected sample to sample variability Statistical Significance lAn observed difference between group averages or proportions so large that it would rarely occur by chance because of variability between repeated samples is said to be statistically significant lThe difference should also be of practical interest 6 mm of 5m Ovolim Slide 22 Suppose n 100 9MofE1W10 5n pmm Placebo Med Trt Results are statistically significagt Depzvlmerl arsmm 5m Ovolim Slide 23 Suppose n 100 9 Ma aMam Placebo Med Trt Results are not statistically sail Suppose n 1000 9MofE1W3 Placebo Med Trt Results are statically significanratty Observational Studies I We can not always randomly assign subjects to groups and groups to treatments I In some cases we must use observational studies where subjects are already in groups I What is the major weakness of an observational study as opposed to an experimental study a mu Depzvlmerl ufsiznxnz Univemlyofsouh Ovolim Slide 25 Good Observational Studies I Subjects are not randomly assigned to groups but we can randomly choose subjects from the groups they are already in including a control group ITry to match the groups where at all possible This eliminates known confounding variables I Finally we compare the groups adjusting for known confounding variables WaitBarizsglxiaai Observational Studies I Repeated many times under different circumstances in different locations at different times lLOOk for a pattern in the results