1 What is error variance and why is it a problem Key Points for Exam 3 a If a causal relationship exists the dependent variable should vary based on the independent variable b Things other than the independent variable what is manipulated can influence the dependent variable what is measured c Error Variance variability in scores caused by extraneous variable or participant variability i People are just different from each other ii Total variance observed group differences systematic variance effect of independent variable error variance effect of all other things d What are the four ways to deal with it control it randomize it across conditions increase effectiveness of your IV and stats i Reduce Error Variance 1 Treat all participants the same 2 Sample from a more homogenous group 3 Match participants on variable that you know have an effect on DV ii Increase Effectiveness of Independent Variable 1 Use levels of IV that are very different 2 Example Effect of Alcohol on Memory a Sober vs 0 05 BAC b Sober vs 0 10 BAC iii Randomize Error Variance Across Groups 1 Participants have an equal chance of being in any group a Random assignment iv Statistical Analysis 1 We ve run an experiment and the average scores are different across treatments a Participants treated for depression have a lower score on a depression index compared to untreated participants b How do we know that this difference wasn t there to begin with due to error variance c We may not be able to know whether error variance caused the effect we observed but we can estimate the probability that it did using statistics i What is the probability that the size of the effect we observed happened just by chance as a results of error variance 1 P value 2 We take a 5 chance that we are wrong Type I Error 2 What are the characteristics of advantages and disadvantages of Randomized 2 Group designs Randomized Multi Group designs Matched Pair designs and Matched Multi Group designs i Randomly assign participants to 2 different groups expose them to 2 different a Randomized 2 Group Design levels of IV b Randomized Multi Group Design c Matched Pair Design i Permits comparing 2 or more treatments to one or more control groups ii Sometimes needed to rule out alternative explanations 1 Compare an intervention group treatment to a No intervention group control b A different intervention group comparison 2 Advantages a Simple 3 Disadvantages i Relatively few participants needed ii Data and stats are easy iii No pre testing required to ensure equality of groups a Doesn t yield a large amount of info b May be insensitive to effects when participants differ greatly in performance when there is a lot of variability i Help control for error variance ii Controls participant related variability by matching groups on characteristics that influence the DV match groups on age 1 Example Age has a large effect on cognitive ability we may want to iii By accounting for this participant related variability we may be more likely to see the effect of the independent variable iv Strategy Measure Match Randomize v By matching participants on characteristics that influence the DV these other characteristics are less likely to explain why groups might differ vi Systematic variance easier to observe 1 Total variance systematic variance error variance vii Matched Pair Design measure your sample find pairs of people who match on characteristics that might influence the DV for each pair randomly assign which person goes to which group d Matched multi group design i Find similar participants then randomly assign these participants to groups ii Similar to randomized multigroup design 3 or more groups 1 Advantages 2 Disadvantages a Decreases error variance a More difficult to implement i Need to measure all participants before study ii If you have many groups may be difficult to find a lot of people who match b Requires the use of slightly less powerful statistics i So make sure the thing you are matching on really DOES influence the DV 3 What are the different types of analyses you can do with data from an experiment with one IV T test ANOVA a Between Participants Experiments i Independent variable is manipulated BETWEEN people 1 You either get one level of the IV or another ii Random assignment to conditions iii There can be 2 or more conditions iv Because we manipulate the independent variable we can say the CAUSE v If the independent variable does something we would see differences in the dependent variable between the groups 1 Example or not a Students are randomly assigned to either participate in tutoring b Independent Samples t test i Two separate groups of people c ANOVA i Analysis of variance ii More than 2 groups iii One way ANOVA only 1 independent variable d Why do you need to do posthoc tests when you have more than 2 groups i Posthoc test finds out which groups differ from one another ii Deal with Type I error inflation 4 What are the null and alternative hypotheses for t tests and ANOVAs a Null Hypothesis b Alternative Hypothesis i Independent variable will not affect the dependent variable There is no difference between groups in the population i The independent variable WILL affect the dependent variable Difference between groups in the population ii Have to say it like this 1 5 Know how to interpret results of each reject or fail to reject null statistical significance how you describe results in regular words a P value 0 05 reject null P is low null must go b Results of an independent samples t test show that the students in the tutoring group had significantly higher GPAs M 3 75 SD 1 04 than did the students in the control group M 2 13 SD 1 25 t 16 2 84 p 0 013 c We conducted a one way analysis of variance ANOVA to compare the GPA for students in the in person tutoring group the online tutoring group and the control group i The overall omnibus test was significant F 2 43 8 27 p 001 ii Post hoc comparisons using Tukey s HSD indicated that students in the control group M 2 29 SD 1 07 had significantly ps 05 lower GPAs than both students in the online tutoring group M 3 43 SD 0 94 and students in the in person tutoring group M 3 89 SD 1 28 iii Students in the in person tutoring group and the online tutoring group did not differ p 49 6 What are t and F ratios of between variance within variance How do mean differences and variability impact t and F and therefore whether you reject the null hypothesis a T ratio i ii Larger T ratio the more
View Full Document
Unlocking...