Download Statistical Tests and Analysis: A Comprehensive Guide with Exercises and Solutions and more Exams Advanced Education in PDF only on Docsity!
CMN3 102 Lab Exam With
Complete Solution
What does a chi-square test? - ANSWER Relationship between two categorical variables When do you accept the null hypothesis for a chi square? - ANSWER If observed matches expected, accept null hypothesis If observed does not match expected, reject null hypothesis
Assumptions for a chi square (4) - ANSWER - 2 categorical variables
- Each participant can only contribute to one cell in the contingency table
- The expected frequencies for any given cell cannot be below 5
- Not using dependent/independent variables - only looking for a relationship
How to measure effect size for chi square + levels - ANSWER Cramer's V Falls between 0 and 1
.5 = large association .3 to .5 = moderate association .1 to .3 = small association
0 to .1 = no association
What statistics do you report for a chi square? - ANSWER X2 (df, N = XX) = XX.XX, p = .XXX Cramer's V: .XXX
What does a t-test measure? - ANSWER Test if a categorical variable predicts a continuous outcome
How many categories can the categorical variable have in a t-test? - ANSWER Must be dichotomous - 2 categories
What is a paired samples t-test? - ANSWER Comparing the same group of people at two points in time
What is an independent t-test? - ANSWER Comparing two different groups of people on the same outcome
Assumptions for paired samples t-tests - ANSWER - Outcome is measured at the continuous level
How do you report statistics for the Mann Whitney U test - ANSWER Statistic: U = X.XX, p = .XXX Group information: median value for each group (Mdn = XX.XX) Effect size: r (instead of Cohen's d), interpreted same way as Cramer's V
What test do you run if you violate the assumption of homogeneity of variances for an independent t-test? - ANSWER Welch's T-Test
How do you report statistics for Welch's T-Test? - ANSWER Statistic: t(df) = XX.XX, p = .XXX Group information: mean + SD for each group (M = XX.XX, SD = XX.XX) Effect size: Cohen's d
What test do you run to see if your analysis has a normal distribution? How do you interpret it? - ANSWER Shapiro-Wilk test p < .05 = non-normal
How do you determine effect size for a t-test? What are the levels? - ANSWER Cohen's d - between 0 and infinity
.8 - large difference
.5 to.8 - moderate difference .2 to .5 - small difference 0 to .2 - no difference
What statistics do you report for a t-test? - ANSWER t(df) = XX.XX, p = .XXX Cohen's d = X.XX M = X.XX, SD = X.XX
How do you calculate df for each type of t-test? - ANSWER Paired samples: df = n - 1 Independent samples: df = n - 2
What does a correlation analysis test? - ANSWER Relationship between two continuous variables
What does a positive relationship indicate? - ANSWER As one variable increases, so does the other
What does a negative relationship indicate? - ANSWER As one variable increases the other decreases
How do you report statistics for a correlation? - ANSWER df = n - 2 r (df) = .XX, p = .XXX r2 = .XX
What does a regression analysis test? - ANSWER Test if one continuous variable predicts another continuous variable
What does the F-ratio represent? - ANSWER How well the line of best fit explains the data
What are the assumptions for a regression analysis? - ANSWER - Both variables measured at the continuous level
- No outliers (extreme scores) in the data (this would ruin the best fit line)
- Linearity: The association should be linear
Is normality required for a regression analysis? - ANSWER No
How do you determine effect size for a regression? What are the levels? - ANSWER Coefficient of determination (r2)
.26 = large association
.13 to .25 = moderate association .03 to .12 = small association 0 to .02 = no association
How do you calculate the df for a regression? - ANSWER dfM will be 1 for a simple regression dfE will be n - 2 dfT (total) = dfM + dfE
How do you report the statistics for a regression analysis? - ANSWER β = .XX, t = X.XX, p = .XXX F(dfM, dfE) = XX.XX, p = .XXX; r2 = .XX)
What is the required level of power for any analysis? - ANSWER .8 or higher
