Final Exam Review

P-values

100

In social science, what is the conventional rejection threshold of the null hypothesis. 

.05

200

What is the definition of a p-value?

The probability of observing a test statistic equal or more extreme than the observed test statistic assuming that the null hypothesis is true. 

300

TRUE or FALSE: 

In a two samples t-test, if a p-value is greater than .05 we can conclude the means between the two groups are equal. 

FALSE

400

p-values are the probability the result is due to random chance. 

FALSE

500

Study 1 has a p-value of .04 and study 2 has a p-value of .001. Can we conclude study 2 had a larger effect size? 

No!! p-values do not measure effect size. 

Alphabet Soup

100

How is the t-statistic computed?

The coefficient divided by the standard error

200

For variables X1 and X2, r = .4

For variables X1 and X3, r = .74. 

What can I conclude?

X1 and X3 are more correlated than X1 and X2.

300

Suppose I have a standardized a variable representing math scores. What is the mean and standard deviation of the variable? 

The mean is 0 with a standard deviation of 1. 

400

A given value has a z score of 2. Is this close too or far away from the mean?

Very far away! Only 2.5% of values are higher than 2. 

500

The distribution of the test statistic for a one way ANOVA under the null hypothesis. 

The F distribution. 

Assumptions

100

Assumptions of a two samples t-test.

1. Normality of parent population. 

2. Independence

3. Equality of Variance

200

The p-value to Levene's test is .02, groups have unequal size, what can we conclude?

The equality of variance assumption is violated.

300

QQ Plots are used to ...

Look at the normality assumption.

400

What are your options if the equality of variance assumption is violated? 

Welch test (t.test or ANOVA version)

500

When are we robust to assumptions

We are robust to normality when the sample size is large (~ 30 per cell). When group size is equal we are robust to equality of variance. 

R

100

oneway.test(lalonde$re78 ~ lalonde$educ)

This anova assumes equal variance. 

200

t.test(lalonde$re78, lalonde$u75)

This does not assume equal variance

300

Code to run a paired t-test. 

t.test(lalonde$re78, lalonde$re74, paired = T)

400

summary(lm(re78 ~ treat))

This will run a two sample t-test for treat

500

If a 95% confidence interval is between -2.58 and -.23, then we are 95% confidence the true parameter is between  -2.58 and -.23

This is false!