Lecture 3 Flashcards
what does confidence interval indicate
PLAUSIBLE VALUES of sample data based on the value of single sample statistic (estimated mean, size of sample) and estimated standard error
if you have greater sample size, will your confidence interval have narrower or greater range?
narrower and it would be closer to the mean
characteristic of confidence interval from a small sample size
- wide range of plausible values
- means inferences is imprecise
how to read contingency table?? (…., ….)
row first then column
row, column
what is a joint cell in contingency table?
it is a cell that contains frequency count of all students who belong to one of the categories of variable 1 and one of the categories in variable 2
what is a marginal cell
it is something like joint cell but contains individual variable on their own
what would a mosaic plot look like when there is no association between 2 variables
height is the same in each column, if there is difference in height association might be present
what does Cramer’s V measure? why do we use it instead of chi-squared null hypothesis test?
Cramer’s V helps us to directly measure the strength of association in contingency table (when at least 1 variable has more than 2 categories).
eg: maths experience variable contains VCE, uni, and none
The Chi sq test doesn’t directly tell us that.
cramer’s V characteristics
- value is from 0-1
- 0 means no association in data
- stronger data would have cramers v vaue closer to 1
- when calculated on sample data, is a sample statistic.
- when calculated on population data, it is a population parameter
what does it indicate if your confidence interval includes the value 0?
means that our data is compatible with the possibility of no association at population level.
what does confidence interval = [.00, .45] mean?
the upper bound is 0.45, indicating moderate association might be plausible. the lower bound is 0 indicated no association is also plausible .
why use confidence interval rather than p value?
confidence interval contains not only all information contained in p value (in a single null hypothesis test) but also the likely values of population parameter.
p value just tells u whether or not we an reject some null hypothesis value.
in cramer’s v function in R, there is a component called ‘method’, each method is called estimator. what is an estimator?
it is a mathematical function applied to sample scores to obtain an estimated value for an unknown population parameter.
2 kinds of estimator
- point estimator: estimator that calculates a sample statistic value
- interval estimator: estimator that calculates a confidence interval
properties of interval estimators
- unbiased (biased/unbiased): independent to the sample size.
- consistent (consistent/not consistent): depends on sample size
- efficient (continuum)
what does it mean then interval estimator is unbiased?
- not depending on sample size
- estimator is unbiased if the value is the same as population
what does it mean then interval estimator is consistent?
- depends on sample size.
- as sample size get larger, coverage rate of interval estimator would get closer to the designated % of CI
you can be consistent and biased
what does it mean then interval estimator is efficient?
- produce narrow confidence interval
what is a broader term for correlationa and cramer’s v?
effect size
what is effect size?
strength of a relationship between construct measures
what is an odd
odds is the probability of an event occurring relative to the probability of it not occurring
what does it mean when you have an odds 4:1 when you’re talking about horse?
probability of having a horse winning as supposed to it not winning is 4 to 1
can an odds ratio be a negative number?
no. only can be a positive number. range is 0 to infinity.
odds nearer to 0 or nearer to infinity indicates stronger association.
what does it mean when odds ratio is 1
means no association. because this is saying that there is NO DIFFERENCE in both cases.
