Lecture 9-11 Flashcards
An adjust R-squared value is often reported because
It prevents misrepresentation of the results
Standardized coefficient determine
The relative strength of the independent variables on the dependent variable
Raising the statistical significance cut-off from 0.05 to 0.1 would
Increase the probability of making a type I error
Which independent variable in the regression below has the strongest impact on the dependent variable:
Beta coefficients:
Sex: .033
DAYS OF POOR MENTAL HEALTH PAST 30 DAYS -.132
The linear equation based on the Ordinary Least-Squares Regression output below is:
Education Years= 13.931 + (0.033Sex) + (-0.132MentalHealth)
What is mean centering?
.Creating a whole new variable of (usually) age by substracting the original value from the mean value
What is an absolute value?
the magnitude of the value as opposed to directionality or positive/ negative sign
What are we trying to achieve when “Comparing Means”
We are trying to compare different means within the sample i.e. various sub-groups of the sample: sexes, ethnicities, union groups
What do the “significance tests for comparing means” do?
To decide if differences between the means of sub-groups is generalizable to population
What are the types of “significance tests for comparing means”?
T-tests (three kinds)
ANOVA(analysis of Variance) tests
Within t-tests, Define the three types of t-tests and their concepts- including one sample, independent sample, paired samples?
one sample: Compare just one unit sample to population
Independent sample: two sub-groups in the sample are compared to each other.
Paired samples: after an intervention (i.e. control experiment), compare the before-after sample means
What do Student’s t-test allow us to do?
If an observed difference in sample means also generalizable to population
Cannot tell about strength of correlations, etc.
What is a cross-sectional data collection method? What kind of t-test is used?
Collecting data one-time( Member Service Survey open for 3 months). Uses the independent sample t-test
What is the longitudinal design? Pros and cons for cross-s and Longitudinal
Longitudinal: collecting data several times over period of time. increases reliability. But more difficult & expensive. Uses the paired samples t-test
When do we need to have a null and research hypothesis?
Anytime we do inferential stats and try to draw a conclusion about a population.
What is the signif. cut-off for t-test?
0.05
What helpful concepts are there when you try to measure or quantify difficult concepts such as happiness, depression and life satisfaction?
Helpful concepts- validity, reliability and measurement error
What is validity? An example of validity in a question?
Accuracy of measurement, and reducing measurement error
I.e. when we ask if folks are white, black, Asian, Latinx, Indigenous- are we asking citizenship, language, religion,
genetics, origin of ancestry?
How do you get reliability?
test-retest reliability.
Inter-rater reliability
What is the relationship between reliability and validity?
a measurement can be reliable but still not valid.
But if it’s not reliable, it’s definitely not valid
What is ANOVA? How is it different from independent sample T-tests? why is it more complicated?
Significance test. independent sample T-tests: compare 2 subgroups. ANOVA: compare the variation between 3 or more subgroups. i.e. rural, urban and sub-urban groups. If the signif level is met, then the variation of means can be generalized to the population level
What are the values in ANOVA that you need to notice??
F-ratio, Sigfi ratio
What does ANOVA NOT calculate?
Since its calculating 3 or more groups, Identify which groups are significantly different from one another
What are the similarities between Chi-square and Anova?
Both use a null hypothesis, and operate under the assumption of the null hypothesis
What are the significant critiques of T-tests and ANOVA?
They don’t tell the strength Nor the direction of the correlations, unlike OLS Regressions.
They only test bivariate relationships and cannot control other variables.
When is it appropriate to use a T-test?
Small sample sizes
What test can control different variables?
Multi-variate OLS regressions.
Why do we have to use qualifiers in interpretations i.e. there is likely a difference in the average income between different education levels in the population…
Cuz we need to realize that when generalizing to the population, there’s a degree of error involved