Which test to use?
Statistical Significance
Here's a scenario, name the test.
Assumption Checks

Create this when you are trying to use what you know about one variable to predict another

What is the regression equation?


It is the statistic you are looking for to indicate if you have statistical significance.

What is p-value?


It is the type of effect named here:

Humans tend to have a higher body temperature than turtles when it is hot out, but not when it is cold out.

What is an interaction (between species and outdoor temperature)?


Each time you report a mean, you should also report this statistic that indicates the average distance of all scores from the group mean.

What is standard deviation (SD)?


In addition to statistical significance, it is always important to report this--a measure of the real world 'wow' factor, beyond your specific sample.

What is effect size?


Use this test to answer the question: Does Taco Bell provide an equal distribution of sauces?

What is a chi-square (goodness of fit)?


It is the graph that is more likely to be associated with a statistically significant finding:



What is B, with very little (or no) overlap between the means and error bars?


100 psychology students are given study materials for the GRE. Half receive two books with practice problems and exams for both the verbal and quantitative sections of the GRE. The other half receive electronic versions of the same material. The company the develops these materials predict that the electronic version is more effective. The students who had the paper materials have a mean exam score of 1280, and those with the electronic materials have a mean score of 1310.

What is an independent samples t-test?


It is the shape that we assume a distribution of means will take, with most scores falling near the mean.

What is the normal distribution?


These are two different measures of effect size (also there are many more you could also name).

What is: cohen's d, eta-squared, R2.... ?


Use this when there are more than 2 independent group means being compared.

What is a between-subjects ANOVA (test)?


In classical statistics, it is the hypothesis that we are actually testing -- we're seeking enough statistical support to reject it.

What is the null hypothesis?


Anna Fisher and colleagues (Fisher et al., 2014) predicted that kindergarten students should learn better in undecorated classrooms compared to decorated classrooms, because students should be less distracted in sparse classrooms without distractions such as posters, maps, children’s artwork, etc. They had a group of 12 kindergarten age children participate in a science lesson in a classroom without decorations and classroom with decorations. The students took a test on the material immediately after each lesson (scores range from 0 to 20). Do kindergarteners learn better in an undecorated compared to a decorated environment?

What is a dependent (paired) samples t test?

If you find that variance is not equal between your within-subjects groups, you should apply this correction (in most cases).

What is the Greenhouse-Geisser correction?


These are three reasons for why studying statistics is beneficial (although there are certainly many more you could name).

Its in government, business, politics, lotteries, research, required for psych major, understanding what is and is not meant when encountered, performing your own studies…

Developing critical thinking skills, marketing skills,....


Use this test if you want to see how one group of participants scores change over time (comparing two group means from the same individuals).

What is a paired/dependent/related samples t-test?


It is the conclusion that can be drawn about the research hypothesis when you find that < .05.

What is support for the research hypothesis (never proof).


Students at 6 different universities are asked about their attitudes towards vaccinations at the beginning and the end of the school year. Researchers want to see whether there are any differences in attitudes between universities and over time.

What is a Mixed ANOVA?


It is the sampling method consideration in the following, that makes it challenging to generalize the findings:

The manager of a movie theatre asked 100 of his customers what they like to do on a Saturday night. 

What is sampling bias? 


It is the type of variable in this example: (nominal, ordinal interval, or ratio) and (discrete or continuous)

What is ordinal and discrete?


Use this test when you have both a between-subject and a within-subject IV manipulation, and a DV that is quantitative.

What is a Mixed ANOVA?

You make this mistake when your statistical analysis suggests incorrectly that you should reject the null hypothesis; we set this probability to alpha (a)

What is a Type I error?


They are the three null hypotheses that can be tested in the following scenario: I want to know if married and unmarried individuals have similar attitudes towards different types of drinks. All participants will rate (-100 = dislike, +100 = like) their preference for beer, wine, and water.

Main Effect of marital status: Married and unmarried people show similar attitudes of drinks.

Main Effect of drink type: There is not a difference in drink preference between beer, wine, and water.

Interaction between marital status and drink type: The type of drink a person prefers does not depend on their marital status.


This is the next step you should do, if find that your Levene's Homogeneity of Variance Test reports p > .05.

What is "do nothing -- you are clear to interpret your between-subjects analysis without making a correction"?


A researcher conducts a one-way ANOVA and reports the following: F(4, 95) = 10.76, p = .03. assuming each of her groups had an equal number of participants in them, how many participants were in each group?

What is 20?

n = 20 per group

(5 groups; N = 100)