MKTG Analytics Exam 2

What is the VERY first thing you should always examine in any statistical test?

The p-value

Why do we use samples instead of the entire population?

Because studying the full population is expensive and unrealistic/unreachable

What is the "magic" sample size mentioned?

Around 400 people → ~5% error

Why doesn't increasing sample size always improve results?

After a certain point, results don't change significantly

What is a Z-test for proportions used for?

To test whether two proportions are significantly different

What information do you need for a Z-test?

Sample sizes (n) and Percentages (p)

What is a one-tailed test?

A test where the hypothesis has a direction (e.g., A > B)

What is a two-tailed test?

A test with no direction (A ≠ B)

What do p-values of 0.245 and 0.432 indicate?

Not significant for both one- and two-tailed tests

What does a T-test compare?

Means and distributions between groups

What does too much overlap between distributions mean?

No significant difference

What does little overlap mean?

Significant difference

What is the formula conceptually for a T-test?

Difference between means ÷ pooled standard error

What does X̄ (x-bar) mean?

Sample mean

What does "s" represent?

Sample standard deviation

What does a larger t-score indicate?

A more significant difference

How do you report a T-test?

t = ?, df = ?, p < ?

What is a one-sample T-test used for?

Comparing a sample mean to a known population mean

What additional data is needed to determine significance?

Sample size and standard deviation

How does variance affect significance?

Lower variance → more likely significant; Higher variance → less likely significant

What does an independent samples T-test compare?

Means between two separate groups

What is the purpose of ANOVA?

To test whether any group differs among multiple groups

What is the key difference between T-test and ANOVA?

T-test → 2 groups; ANOVA → 3+ groups

What must you do if ANOVA is significant?

Run a post hoc test

What is the purpose of a post hoc test?

To determine which specific groups differ

What post hoc test is commonly used here?

Duncan's test

What are the two steps of ANOVA?

Check if significant; If yes → find where differences are

What is a main effect?

The effect of one independent variable on the dependent variable

What is an interaction effect?

When one IV's effect depends on another IV

Why is post hoc not needed for some main effects?

Because there are only two levels

What are contrasts used for?

To analyze specific differences in interactions

What does p < .05 indicate?

Significant effect

What does a high p-value (e.g., .76) indicate?

Not significant

Do p-values prove anything?

No — they only provide evidence

What is a Type I error?

False positive

What is a Type II error?

False negative

What is regression analysis?

A method that estimates a trendline and models relationships

What does regression measure?

How well data fits a trendline

What variables are used in regression?

IVs (predictors) and DV (outcome)

How many DVs can regression handle?

Only one at a time

What is simple linear regression?

One independent variable

What is multiple regression?

Two or more independent variables

What is logistic regression used for?

Categorical outcomes

What is a residual?

Actual - predicted value

What does OLS stand for?

Ordinary Least Squares

What must the DV be for linear regression?

Numeric

What is backward elimination?

Start with all variables → remove weakest

What is forward selection?

Start with best predictor → add more

What is stepwise selection?

Adds/removes variables step-by-step

What is dummy coding?

Converting categorical variables into 0s and 1s

How many dummy variables are created?

k - 1

What is multicollinearity?

Predictors are highly correlated with each other

What is omitted variable bias?

Leaving out an important variable

What is the FIRST thing to check in model evaluation?

p-value

When is a p-value significant?

< 0.05

What should you do with non-significant predictors?

Remove them

What does R² measure?

Variance explained in DV

What is the range of R²?

0 to 1

Why use adjusted R²?

It penalizes for adding unnecessary variables

What does ANOVA test in regression?

Whether the model explains any variance

What does a coefficient (B) represent?

Change in DV for a 1-unit increase in IV

What does the sign of B indicate?

Direction of relationship

What is B₀?

Intercept

When should you interpret a coefficient?

Only if it is statistically significant

What if a coefficient is not significant?

It is not meaningful

What is standard error (SE)?

Estimate of prediction accuracy

What does a smaller SE mean?

Better predictions

If income coefficient = 10.51, what does that mean?

For every $1000 increase in income, debt increases by $10.51

What must you include in reporting results?

DV, IVs, Significant predictors, Coefficients, Adjusted R²

How should you interpret results?

Clearly explain direction and magnitude

What is linearity?

Relationship is linear

What is normality?

Errors are normally distributed

What is homoscedasticity?

Equal variance of errors

What is independence?

Observations are not related

What is associative analysis?

Examining relationships between variables

What is a monotonic relationship?

Consistent direction (increase or decrease)

What is a nonmonotonic relationship?

No clear direction

What does correlation measure?

Strength and direction of relationship

When do we interpret correlation?

Only if statistically significant

What should you report for correlation?

Only significant relationships

What is chi-square used for?

Testing relationships between categorical variables

What is the null hypothesis in chi-square?

Variables are not related

What are observed frequencies?

Actual data counts

What are expected frequencies?

Counts if no relationship exists

What does a significant p-value in chi-square mean?

Variables are related

What must be included in reporting chi-square?

Counts, χ² value, p-value

What type of data is used for correlation?

Continuous

What type of data is used for chi-square?

Categorical

What should analysis lead to?

A management recommendation

How should results be explained?

Simply (like explaining to a beginner)