Lecture 15: Flashcards
If the variable is categorical, we’re likely to be interested in what more between proportions and means?
Proportions more than means
Parameter we want to estimate isn’t the population mean but rather the population _______
Proportion
We can estimate the population proportion by taking a ___ and then computing the _____ proportion
Sample
Computing a sample proportion:
Proportion = no. Of individuals with trait in the group/ total number of individuals in the group
How do you compute the odds of a group?
Odds = number of individuals with the trait in the group/number of individuals without trait in the group
How do you calculate Proportion differences in groups?
Proportion difference = Group 1 proportion - Group 2 proportion
The sample proportion can also be called:
- The probability
- The risk
We can also compute a ______ ______ for the population proportion so we know how precise our estimate is
Confidence interval
How do you compute a proportion ratio comparing 2 groups?
Group 1 proportion/Group 2 proportion
How would you compute the odds ratio between 2 groups?
Group 1 odds/Group 2 odds
Overall what are the 5 statistics we can compute from a contingency table?
- Sample proportion
- Sample odds
- Sample proportion difference
- Sample proportion ratio
- Sample odds ratio
We have to remember that our computations are only _____
Estimates
Eg. Sample risk is only an estimate of the true risk of the population
What is essential for getting decent estimates of proportions and odds?
Large sample (n in the hundreds/thousands)
The smaller the proportion difference/ proportion you’re estimating, the ______ the sample needs to be to get reliable estimates/reasonably narrow confidence intervals
Bigger
The farther from 1 the odds/odds ratio are that you’re estimating the _____ the sample needs to be
Bigger
Often the proportion ratios and odds ratios you see are “_______”
Adjusted = they adjust the ratio to remove the influence of a potentially confounding variable
