L3 - Inferential Analysis Flashcards
What is operationalization?
What is content validity?
Is the measure a good implementation of the theoretical construct?
–> Content validity
Confounding variable
A variable that might be correlated with the IV and might actually be the influence on the DV
Moderator
Might affect the relationship between IV and DV
Mediator
Establishes the relationship between the IV and the DV
Difference between experimental and correlational research
Experimental: IV is explicitly manipulated
Correlational: IV and DV vary naturally in sample
What is within-subjects design
Same person is presented with several levels of the manipulated variable
Between-subjects design
Different people are presented the different levels of the manipulated variable
Is a universal hypothesis verifiable?
If you only have a sample from population then no.
, “All employées in tech companies are short-sighted.“
Is an existential hypothesis verifiable with a sample from the population?
Yes. But not falsifiable.
“There are employées in tech companies that are short-sighted.“)
Can proportions or trends be verified or falsified with a sample from the population?
No both not.
“55% of employées in tech companies are short-sighted.”
Can we make a decision on whether to accept or reject an hypothesis based on statistical inference?
yes
What is Fisherian approach
set up probability distribution under H0
What is the Neyman-Pearson approach?
set up H1 and H2
What is the effect size?
- Difference between H1 and H2
What is type 1 error?
If the state of the world H1 true but you accept H2
What is type 2 error
If the state of the world is H2 but you accept H1.
What is power?
1-ß
When H2 is state of the world and you accept H2
What is the p-value
probability of observed test statistic under H0
Is the p-value a measure of the evidence of the hypothesis
no
Q6: Imagine you ran a statistical inference test and obtained a p-value of p = .01. Which of the following statements correctly describes your situation (multiple responses possible)
You have found the probability that if the null hypothesis is true, you would get results as extreme (or more extreme) as what you have observed.
Is the effect size dependent on the p-value?
No
What is a small effect size?
.2
What is a large effect size?
.8
Which test to compare means of two independent groups?
Independent-samples t-test
Which test for comparing the means of two related sets of observations?
Dependent samples t-test
Example: Happiness of a group of people before and after an election
(i.e., each person measured twice)
Which test for comparing the mean of one group against a single value?
one sample t-test
Example: Protein content of sample of packages of an energy bar to see whether they contain 20g of protein (as indicated on the package)
Computing the expected frequency under H0 of two nominal variables
What is Chi-squared used for?
Analyzing the association of two nominal variables. Chi-squared is the normalized different between observed and expected frequencies
What does Cramer’s V measure
effect size between two nominal variables
With a larger sample size, the actual values in the population are better reflected; variability around the true value becomes smaller.
What is this called?
Central Limit Theorem
Reminder: What is the power?
Does the power vary for different effect sizes?
Yes. Larger effect size, larger power
Does the power depend on the sample size?
Yes, because larger samples are more peaked. The larger the sample, the higher the power.
What is the a priori power analysis about?
How large should the sample size be to have a good change to detect the hypothesized effect?
