Advanced statistics study guide for exam # 2
Advanced statistics study guide for exam # 2 PSYC 4317
Popular in Advanced Statistics
Popular in Psychlogy
This 4 page Study Guide was uploaded by Alejandra D. Rocha on Sunday October 25, 2015. The Study Guide belongs to PSYC 4317 at University of Texas at El Paso taught by Anthony Blum in Summer 2015. Since its upload, it has received 142 views. For similar materials see Advanced Statistics in Psychlogy at University of Texas at El Paso.
Reviews for Advanced statistics study guide for exam # 2
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 10/25/15
PSYC 4317 Spring 2014 Study Guide 2 Analysis of Covariance 1 What is a covariate The covariate is a continuous variable that is a quantity not a categorical factor 2 In just looking at the design how does a design in which you would perform an analysis of covariance differ from a design in which you would perform an analysis of variance n ANCOVA for each subject you have group membership dependent variable score and a covariate score quantity 3 What type of relationship do we want between our covariate and our dependent variable Why The covariate should be correlated with the dependent variable because this is an important fact in increasing the power of your test 4 What would happen to the denominator of the F in such a circumstance Why The denominator chance variability would be a small number because ANCOVA computes variability by subtracting each Y score from the Y score predicted from the covariatedependent variable Y regression line As long as the correlation between the covariate and the dependent variable is greater than zero this should lead to a smaller measure of variability 5 What type of relationship do we want between our covariate and our independent variable Why The covariate should not be correlated with your Independent variablewhich might make your test less powerful than the corresponding ANOVA This is why it is recommended that you measure your covariate before the experiment pretest 6 What would happen to the numerator of the F ratio if the covariate and the independent variable were related The denominator would be a bigger measure of variability leading to a smaller F 7 What are two ways that ANCOVA could lead to a loss of power Working against the trend to increasing power are situations where you have a covariate associated with the Independent Variable Also you lose a degree of freedom with ANCOVA as opposed to ANOVA which will also reduce the power of your test 8 How does ANOVA compute variance when compared to ANCOVA ANOVA computes the variance by subtracting every Y dependent variable score from the mean and ANCOVA computes variability by subtracting each Y score from the Y score predicted from the covariate dependent variable Y regression line 9 Describe the adjustment of group means estimated group means procedure that Analysis of Covariance performs ANCOVA will adjust the means on the dependent variable of the different experimental groups The adjustment gives us the group means that would have occurred if all subjects received the same score on the covariate 10 What key assumption has to hold for this adjustment procedure to be valid Describe this assumption It is a form of statistically equalizing our experimental groups the assumption of homogeneity of regression slopes must hold compute a separate covariate dependent variable regression line for each group the assumption holds if the regression lines have the same slope parallel 11 What is random assignment random assignment is a form of experimental equalizing our different groups 12 When is the key assumption mentioned in 10 likely to hold When it is not likely to hold The problem is that the situation in which researchers wish to use the adjustment procedure namely when comparing preexisting groups quasiIndependent variable is very likely to be a situation where the Homogeneity of Regression Slopes Assumption is violated e The situation where the Homogeneity of Regression Slopes Assumption is likely to hold is when we have formed our different experimental groups via random assignment a True Independent Variable Of course in this case we have likely already equalized our groups experimentally Repeated Measures ANOVA 1 What is the key difference between repeatedmeasures withinsubjects designs and between subjects designs Repeated measures design uses the same subjects for all the levels of independent variable 2 Give me an example of an experiment with a onefactor repeatedmeasures design be clear about what subjects are in what conditions Subjects rats Dependent variable running time Independent variable maze color Levels white red blue 12 subjects go under the three levels of the independent variables and their running times are measured under every condition 3 What are two advantages of a repeated measures design vs corresponding between subjects design a use less subjects b more powerful statistical test 4 What type of problem arises with a repeated measures design that did not arise with a between subjects design Subjects in study a long time Carry over effects previous treatment is affecting performance on a subsequent treatment confound variable 5 Give me examples of different kinds of carryover effects describe them Learning if a subject learns how to perform a task in the first treatment performance it is likely to be better if the same or similar task are used in subsequent treatments Fatigue if performance in earlier treatments leads to fatigue then performance in later treatments may be deteriorated regardless of any effect of the independent variable Habituation under some conditions repeated exposure to a stimulus leads to reduces responsiveness to that stimulus Sensitization sometimes exposure to one stimulus can cause subjects to respond more strongly to another stimulus In a phenomenon called potentiated startle for example a rat will show an exaggerated startle response to a sudden noise if the rat has recently received a brief foot shock in the same situation Constrast because of contrast exposure to one condition may alter the responses of subjects in other conditions If you oay your participants a relatively large amount of success performance on one task and the pay them less or make them work harder for the same amount in a subsequent task they may feel underpaid Adaptation if subjects go through a period of adaptation becoming adjusted to the dark for example then earlier results may differ from later results because of adaptive changes 6 What is a counterbalancing In counterbalancing you assing the various treatments of the experiment in a different order for different subjects Each treatments level occurs equally often in each position and each precedes and follows every other treatment on equal number of times 7 Give me an example of a design with counterbalancing Table EH MCnunterbalanced Single I Factor Design With Three Treatments l TREATMENTS SUB ECTS T T2 T1 gaminlmlgpm l L 8 What are the two aspects of counterbalancing that allow us to cancel out carryover effects Present all possible orderings of the treatment every treatments follows every other treatment equally often across subjects and every treatment appears equally often in each position 9 Give me an example of a two factor repeated measures design which subjects go where Subjects 12 Ducks Dpt vbl Swimming speed Ind pt vbls Factor 1 Liquid type Levels of factor 1 water olive oil duck sauce Factor 2 Light source Levels of factor 2 strobe sunlight and infrared All the subjects 12 ducks go under all the levels of the factor 1 liquid type and the levels of the factor 2 light source Hypothesis Ho 1 strobe u sunlight u infrared Ho 1 water 1 oil 1 ds Ho there is no light type X liquid type interaction 10 In terms of the statistical computations think sumof squares how does the Repeated Measures ANOVA reduce error variance By having the same subject across the independent variables All the subject related factors such as age weight religion personality sex are identical across treatments reducing error variance Mixed Designs 1 What is the defining feature of mixed designs This type of design includes a repeated measures independent variable and betweensubjects independent variable 2 Give me an example of a mixed design be sure to show what the distribution of subjects is Subjects 15 platypus of kills dependent variable Independent variables Diet type fish carrots cheese Location of kills water land The 15 subjects platypus go under the two levels of locations of kills water and land Then three groups of 5 platypus each go under one level of the diet type the first group eats fish the second group eats carrots and the third group eats cheese The diet type is a within subjects independent variable and location of kills is a repeated measures variable 3 Be sure to know your designstatistics pairings handout that is presented on blackboard Nonparametric Tests 1 What are some standard assumptions for parametric tests 1 Random independent sampling 2 Equality of variance homogeneity assumption 3 Normality assumptions normal distribution 4 Ratio or interval scale data 2 In what cases would you use a nonparametric test If the assumptions do not hold if they are violated I would use a nonparametric test 3 With what design would you use the MannWhitney Test What parametric test would you use the MannWhitney Test in place of A betweensubjects one factor two level design independent sample ttest 4 What is the null hypothesis for the MannWhitney Test Ho the medians for the levels of the between subjects factor are the same 5 Describe the logic of the MannWhitney test in terms of ranked scores Since the null hypothesis involve the medians of each level we ran order all scores from highest to lowest If all or most of the high ranks are in one group and low ranks in the other we reject null hypothesis 6 With what design would you use the Wilcoxon Signed Ranks Test What parametric test would you use the Wilcoxon Signed Ranks Test in place of Repeated measures one factor two level design and the parametric test would be paired samples ttest 7 What is the null hypothesis of the Wilcoxon Test Ho there is no difference in the medians between before and after conditions 8 Describe the logic of the Wilcoxon Test in terms of ranked differences Compute subtraction of before scores minus after scores for each subjects 9 With what design would you use the KruskalWallis Test What parametric test would you use the KruskalWallis Test in place of Betweensubjects one factor multiple level design One way ANOVA univariate 10 What is the null hypothesis for the KruskalWallis Test Ho mean ranks of the groups are the same 11 Describe the logic of the KruskalWallis test in terms of ranked scores You convert the measurement observations to their ranks in the overall data set the smallest value gets a rank of 1 the next smallest gets a rank of 2 and so on 12 With what design would you use the Friedman Test What parametric test would you use the Friedman Test in place of Repeated measures one factor multiple levels design on one way repeated measures ANOVA 13 What is the null hypothesis for the Friedman Test Ho there are no differences between the conditions 14 Describe the logic of the Friedman test in terms of ranked scores Rankordering the measures for each subject For the present example we will assign the rank of quot3quot to the largest of a subject39s three measures quot2quot to the intermediate of the three and quot1quot to the smallest
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'