PSYCH306 Exam Study Guide A+ Graded
What is an example of a theoretical construct operationalised as a measurable variable?
- ANSWER Anxiety. Test rating anxiety on numerical scale
What is a sampling distribution? - ANSWER The distribution of test statistic values that
we expect under the null hypothesis
What is the critical region within a sampling distribution? - ANSWER Extreme end ie tail
of the sampling distribution where we have unlikely values ie unlikely to occur under the
null hypothesis. Accept them as evidence for rejecting the null hypothesis.
What does it mean if your test has low statistical power? - ANSWER The probability of
getting a true positive if there really is an effect in the data is low ie lower than 80% ish.
Therefore a higher chance of getting false negatives and lower chance of getting true
positives.
How can you fix low test statistical power ? - ANSWER Increase the sample size
What does the formula for variance tell us ? - ANSWER Sum of squares= a raw measure
of how much datapoints vary around mean
Variance= sum of squares divided by number of points to get average measure of how
much they vary around mean
Standard deviation = take square root of variance to get back in original scale units
What does formula for the F value tell us ? - ANSWER How large the differences between
group differences are compared to within group differences.
F value can only be positive, asymmetric
What does the confidence interval tell you ? - ANSWER
What does Bayes rule allow us to figure out in a research setting? - ANSWER It lets us
figure out the probability of various hypotheses given the data that we have observed.
Not something we can figure out form null hypothesis significance testing.
What does a Bayes factor tell us ? - ANSWER The ratio of the evidence for one
hypothesis versus the other
What do we conclude if the Bayes factor is very large? - ANSWER Lots of evidence for
the alternative hypothesis
What do we conclude if the Bayes factor is very small ? - ANSWER Lots of evidence for
our null hypothesis
, What do we conclude is the Bayes factor is exactly 1? - ANSWER No greater evidence
for either the null or the alternative hypothesis
What is an issue with P-hacking? - ANSWER Increasing your false positive rate
How is pre-registering useful? - ANSWER If you commit to what your experimental
procedure and analyses will be ahead of time, you have a lot fewer researcher degrees
of freedom. Cannot mess with the data etc
Axiology - ANSWER Driven by values
"The reason we should do research is to improve the human condition" - ANSWER
Axiology- driven by values
One variable with a categorical outcome with 3+ levels - ANSWER Chi-square goodness
of fit test
Do frequencies differ from expected (null) values? - ANSWER Chi-square goodness of fit
test
Two variables, both categorical with 2+ levels - ANSWER Chi square test of
independence
Is there an association in frequencies between variables? - ANSWER Chi square test of
independence
One variable with a continuous outcome - ANSWER T-Test (one sample)
Does mean differ from expected (null) value? - ANSWER T-test (one sample)
Two variables: a categorical predictor with 2 levels and a continuous outcome.
- There are different individuals in each predictor category - ANSWER T-test
(independent samples)
Do means differ between categories ?
- there are different individuals in each predictor category - ANSWER T-test
(independent samples)
Two variables: a categorical predictor with 2 levels and a continuous outcome
- there are the same individuals in both predictor categories - ANSWER T-test (paired
samples)
Do means differ between categories ?
-same individuals in both predictor categories - ANSWER T-test (paired samples)
Two variables: a categorical predictor with 3+ levels and a continuous outcome variable
What is an example of a theoretical construct operationalised as a measurable variable?
- ANSWER Anxiety. Test rating anxiety on numerical scale
What is a sampling distribution? - ANSWER The distribution of test statistic values that
we expect under the null hypothesis
What is the critical region within a sampling distribution? - ANSWER Extreme end ie tail
of the sampling distribution where we have unlikely values ie unlikely to occur under the
null hypothesis. Accept them as evidence for rejecting the null hypothesis.
What does it mean if your test has low statistical power? - ANSWER The probability of
getting a true positive if there really is an effect in the data is low ie lower than 80% ish.
Therefore a higher chance of getting false negatives and lower chance of getting true
positives.
How can you fix low test statistical power ? - ANSWER Increase the sample size
What does the formula for variance tell us ? - ANSWER Sum of squares= a raw measure
of how much datapoints vary around mean
Variance= sum of squares divided by number of points to get average measure of how
much they vary around mean
Standard deviation = take square root of variance to get back in original scale units
What does formula for the F value tell us ? - ANSWER How large the differences between
group differences are compared to within group differences.
F value can only be positive, asymmetric
What does the confidence interval tell you ? - ANSWER
What does Bayes rule allow us to figure out in a research setting? - ANSWER It lets us
figure out the probability of various hypotheses given the data that we have observed.
Not something we can figure out form null hypothesis significance testing.
What does a Bayes factor tell us ? - ANSWER The ratio of the evidence for one
hypothesis versus the other
What do we conclude if the Bayes factor is very large? - ANSWER Lots of evidence for
the alternative hypothesis
What do we conclude if the Bayes factor is very small ? - ANSWER Lots of evidence for
our null hypothesis
, What do we conclude is the Bayes factor is exactly 1? - ANSWER No greater evidence
for either the null or the alternative hypothesis
What is an issue with P-hacking? - ANSWER Increasing your false positive rate
How is pre-registering useful? - ANSWER If you commit to what your experimental
procedure and analyses will be ahead of time, you have a lot fewer researcher degrees
of freedom. Cannot mess with the data etc
Axiology - ANSWER Driven by values
"The reason we should do research is to improve the human condition" - ANSWER
Axiology- driven by values
One variable with a categorical outcome with 3+ levels - ANSWER Chi-square goodness
of fit test
Do frequencies differ from expected (null) values? - ANSWER Chi-square goodness of fit
test
Two variables, both categorical with 2+ levels - ANSWER Chi square test of
independence
Is there an association in frequencies between variables? - ANSWER Chi square test of
independence
One variable with a continuous outcome - ANSWER T-Test (one sample)
Does mean differ from expected (null) value? - ANSWER T-test (one sample)
Two variables: a categorical predictor with 2 levels and a continuous outcome.
- There are different individuals in each predictor category - ANSWER T-test
(independent samples)
Do means differ between categories ?
- there are different individuals in each predictor category - ANSWER T-test
(independent samples)
Two variables: a categorical predictor with 2 levels and a continuous outcome
- there are the same individuals in both predictor categories - ANSWER T-test (paired
samples)
Do means differ between categories ?
-same individuals in both predictor categories - ANSWER T-test (paired samples)
Two variables: a categorical predictor with 3+ levels and a continuous outcome variable
