t-Tests

Three types of analyses in this note

Correlation

• Two variables

Linear Regression

• Two variables

Multiple Regression

• Three or more variables

Single Sample t Test

• Two variables

• One factor

• Two levels

Independent t Test

• Two variables

• One factor

• Two levels

Paired Dependent t Test

• Two variables

• One factor

• Two levels

Causality

• Something causing something

Quasi

• Kind of, but not fully

• An experiment without random assignment

• Example:

  • Comparing stress levels between students in two different classes
    (you didn’t randomly assign students → quasi)

  • Comparing people before vs. after a policy change
    (no random assignment → quasi)

Example:

Final Grade in 3830

Fall vs. Winter

What would we name the factor? (factor is another way of saying independent variable)

• Factor - Semester:

  • Level 1: Fall

  • Level 2: Winter

We’re looking to see if we can reject the null

• 0.5

One tailed

• The direction we think is going to happen (Winter is going to do better than Fall)

• Increasing your power

• Direction is predicted

Two tailed

• A two-tailed test assesses whether there is a significant difference in either direction, meaning Winter could perform significantly better or worse than Fall.

• Direction is not predicted

3rd year Stats class

Winter only

78-81 (known mean)

*What would you be a type of study you would run

Normality

• A normally distrubuted set of data (think of a bell curve (mean. median, mode is the same)

• Know what it is and how to define it

• We literally assume YES

• In theory if NO = ‘non parametric’

• You do not want an inherent difference (if it is significant_

• 👉 You do NOT want Levene’s to be significant

Lab

Factor

• Breed Type

  • Level 1: Large

  • Level 2: Small

Independent Variable

• Breed

Dependent Variable

• Yappiness

  • 1-5

Factor

• Location

  • Level 1: Home

  • Level 2: Park

Independent Variable

• Location

Dependent Variable