Exam 2 PS 306, Fall 20021. One problem that you might encounter in a study that could lead to a confound is mortality or attrition. Carefullydescribe a hypothetical experimental scenario in which such attrition would be a confound. Then describe a scenarioin which your study might encounter attrition, but it would not lead to a confound. Clear logic and sufficient detailin your examples are both important. [10 pts]Generally speaking, participant mortality or attrition is only a problem if it is extensive(proportionally lots of data lost) and falls unevenly on your conditions. For example,suppose that you were doing a study to look at the effectiveness of two weight-lossprograms (A & B). You might gain access to a list of people who are interested in losingweight through some program and randomly assign them to one of three conditions(placebo control, A, & B). Let's assume that the "programs" would continue for a year,with weight assessments (the DV) taken at the end of the year. (Of course, you'd probablywant to take weight measurements more often.) Suppose that you start out with 50 peoplein each of the three conditions, but at the end of the year when you went to take the weightmeasurement, only 30 people were left in the placebo condition, 25 people in the ProgramA condition, and 45 people in the Program B condition. The disproportionate number ofpeople missing from the Program A condition should concern you. You might imagine thatpeople in that group became discouraged, so they dropped out of the program. Or theymight have been so successful that they discontinued the program. In either case, the 25people remaining would not be an accurate representation of the effectiveness of ProgramA.Suppose, however, that in the study you had less attrition or more similar attrition.Suppose that at the end of one year, 35 people were left in the placebo condition, 37 peoplewere left in the Program A condition, and 36 people were left in the Program B condition.Unless you could think of some reason why the similar levels of attrition are due todifferent causes (leave because of success in Program A, but leave because of failure inProgram B), I would think that the attrition is less of a problem in this scenario.2a. Suppose that you are interested in determining the extent to which either or both of two drugs (Drug A and DrugB) might lead to improved performance on problem solving tasks for mentally retarded children. To gain a sense ofthe nature of the impact of the treatments, you decide to test each child on four different types of problems chosen tobe appropriate for the mental abilities of such mentally retarded children (spatial problems, math problems, logicproblems, and manual dexterity problems). Because you are concerned about the problems of carry-over effects, youdecide to use a mixed design with the Drug factor between groups (independent groups) and the Problem factorwithin groups (repeated measures). Carefully describe the design of your study, the procedure you would use inconducting your study, the number of participants you would use in your study, and any relevant details that willmake clear that you’ve thought carefully about the design of this study. [20 pts]It is crucial that you provide sufficient detail in your depiction of the study you want to do.In this case, it is vital that you have at least three levels of your independent groups factor(Drug): Placebo, Drug A, and Drug B. Then, it is crucial that you use completecounterbalancing of the repeated measures factor (Problem). Complete counterbalancingwould involve 24 different orders: Sp, Ma, Log, Dex; Sp, Ma, Dex, Log; Sp, Log, Ma, Dex;Sp, Log, Dex, Ma; Sp, Dex, Ma, Log; Sp, Dex, Log, Ma; etc. Thus, you'd need a minimumof 72 participants (24 x 3).What might you use as the DV? Time to solve X problems? Number of problems correctlysolved in X minutes? You'd need to be sensitive to the abilities of your participants, so thatyou didn't exceed their attention span, etc. So, you'd want to try to keep your total timeunder a certain number of minutes. And I think that you'd really want to keep the studyconstrained to one testing time rather than multiple days.So, let's presume that we would have pre-tested 10 problems of each type (to ensure thatthey are appropriate) and we'll use number of problems solved in 5 minutes as our DV. Wemight also build in a brief rest/play period between problems. Thus, the whole procedureshould take around a half hour.The procedure might be something like: participant is given a pill to swallow (with theexperimenter blind to the treatment), then play for a half hour to allow the drug to takeeffect, then Math problems (5 mins) , then rest (5 mins), then Logic problems, then rest,then Dexterity problems, then rest, then Spatial problems (as one possible order).2b. In the “graph” below, indicate the means that might result from your study. Then tell me what your means wouldlead you to predict about the results of an ANOVA on the data. [10 pts]1.522.533.544.555.5Spatial Math Logic DexterityHypothetical DataPlaceboDrug ADrug BNumber SolvedProblemThese data certainly depict an interaction between the two factors. It appears thatDexterity problems aren't affected by the drugs, but the drugs both appear to help onSpatial problems, and (to a lesser extent) on Logic problems. Drug B seems to help on Mathproblems, but not Drug A. There also appears to be a main effect for Drug (Drug B > DrugA > Placebo) and a main effect for Problems (Dexterity > Spatial > Math > Logic). Ofcourse, you would focus your attention on the interaction.3. An experiment is done to test the effect of age and dress of a speaker on attitude of a college audience. Groups ofcollege students listen to a speaker talk on why Slabotnik should be president. The speaker is either neatly dressed(tie and jacket) or grubbily dressed (ripped blue jeans and torn t-shirt). Additionally, the speaker is either in his 20's,30's, or 40's. After the talk, the subjects rate their attitude about whether Slabotnik should be elected, on a scaleranging from 0 (should not be elected) to 7 (should be elected). Complete the source table below and interpret thestudy as completely as you can. [20 pts]What are your three H0’s?Age H0: m20 = m30 = m40Dress H0: mneat = mgrubbyInteraction H0: No interactionWhat value represents your best estimate of s2? MSError = .6282 2.056 1.028 1.637 .2114 3.274 .3071 14.694 14.694 23.407 <.0001 23.407 .9992