Stats Test Decision Tree

Explanatory Variable is either

Categorical or Numerical

Categorical Variable

can give numbers to, but don’t always

incudes nominal and ordinal

Numerical Variable

can do math with.

includes continuous and discrete

Explanatory Variable

the variable we suspect is causing the change of the response variable

Response Variable

the variable that may change as a result of the explanatory varaible

Stat Tests associated with categorical explanatory variables

chi square

two sample t-test

anova

Stat Tests associated with explanatory numerical variables

correlation

regression

multiple regression

logistic regression

Response variables can be

categorical or numerical

If the explanatory variable is categorical, and the response variable is categorical, which stats test(s) can we run

Chi square

If the explanatory variable is categorical and the response variable is numerical, what stats test(s) can we run

T-test

Anova

If there are three or more categorical explanatory variables and the response variable is numerical, what stats test(s) can we run

anova

If there are less than two categorical explanatory variables and the response type is numerical, what stats test(s) can we run

t-test

If the explanatory variable is numerical and the response variable is categorical, what stats test(s) can we run

logistic

regression

If the explanatory variable is numerical and the response variable is numerical, what stats test(s) can we run

correlation

regression

multiple regression

If the explanatory variable is numerical and the response variable is both numerical and categorical, what stats test(s) can we run

correlation

regression

multiple regression

If there is one numerical explanatory variable and the response variable is numerical (or numerical and categorical), what stats test(s) can we run

correlation

regression

If there are two or more numerical explanatory variable and the response variable is numerical (or numerical and categorical), what stats test(s) can we run

multiple

regression

Observational studies cannot be used to

determine cause and effect

(establish an association only)

Dots Plots are useful for visualizing

one numerical variable

Scatterplots are useful for visualizing

the relationships between two numerical variables

Visualizing Numerical Variables

dot plots, scatterplots, histograms, box plots

Specific Stats for Numerical Data

variance, standard deviation, median, mean, median

Visualizing Categorical Data

Bar plots, contingency tables, mosaic plots, pie charts, side by side box plots

Types of Bar Plots

stacked, side by side, standardized stacked

Null Hypothesis

No relationship; observed difference in proportions is simply due to chance

Alternative Hypothesis

there is a relationship; observed difference proportions is not due to chance

We use a chi-square test when

dealing with counts and investigating how far the observed counts are from the expected counts, we use a chi-square statistic

Chi-Square’s parameters

degrees of freedom

Conditions of a Chi-Square Test

1. Data is categorical

2. Independence

3. Same Size

4. df >1

Chi-square tests are used with

categorical data

T-tests investigate

the difference between groups

One Sample T-test is used when the explanatory variable is ____ and the response variable is ____

explanatory: categorical

response: continuous (numerical)

Paired t-tests are used when

two groups share participants/subjects

Null Hypothesis for Paired t-test is always that

the difference is 0

A one-tailed t-test is only appropriate with

a directional hypothesis

ANOVAs are used with

categorical variables

Correlation is used to

describe the strength of the linear association between two variables

Simple linear regression is used when you have

two variables

Multiple Regression is used when you have

more than two variables

Logistic regression can be used to perform a regression when you have

a binary explanatory varaible

Descriptive statistics

mean, standard deviation, frequency, describe, pnrom