Exam 3 Flashcards Preview

PSY 3392 > Exam 3 > Flashcards

Flashcards in Exam 3 Deck (49)
Loading flashcards...

used to confirm whether IV has produced an effect in an experiment; used because of nature of control provided through random assignment

inferential statistics


an outcome is one that has only a small likelihood of occurring if the NH were true; the difference obtained in experiment is larger than would be expected if error variation (chance) were responsible for outcome

statistical significant


assumes that IV has no effect on DV / there's no difference between groups / there's no association between variables

null hypothesis


assumes IV does have effect on DV / there is difference between variables / there is association between variables

alternative hypothesis


NHST is a probability statistic, so we can't __ a hypothesis, only __ it

prove vs support


probability of obtaining the observed effect if NH were true; finding a difference between groups if there's no real difference; probability of obtaining observed effect by chance



p-value threshold that needs to be crossed in order to reach statistical significance (.05 in psychology); chosen before we begin study to avoid experimenter bias; based on statistical probabilities

alpha level


NH is actually false and NH is rejected

NH is actually true and NH is rejected

NH is actually false and fail to reject NH

NH is actually trie and fail to reject NH


false positive

false negative



false positive; reject NH but NH is really true; saying there's an effect when there really isn't; most common cause of this is conducting too many statistical analyses; can reduce it with a stricter p-value; most common problem in psychology today

Type I Error


false negative; failing to reject NH when it's actually false; usually caused by a small sample size; something is there and we don't see it

Type II Error


an index of strength of relationship between variables / differences between groups; is mostly independent of sample size; many types for different statistical analyses; most common are Cohen's d and Pearson's r

effect size


strength of association between two variables; sign gives us direction ; closer to |1| os effect; large -> r > .37, med -> r > .24, small r > .10, none -> 0-.1

Pearson's r


ES of different groups; depends on size of difference between groups and amount of variability within groups; large -> d > .8, med -> d > .30, small d > .20

Cohen's d


probability that NH will be correctly rejected when it is false (a hit); the ability to detect statistically significant effects; 1 - TIIE; want it to be .80 greater; determined by significance level, effect size, and sample size; used to determine sample size



the way we determine sample size; tells you appropriate sample size you'll need for effect size to reach that level of statistical significance; based on 1) type of analysis to be conducted 2) estimated ES and 3) p-value / alpha level

power analysis


refers to research design; ability to detect effect of IV even if effect is small; likelihood that it will detect effect if IV does not have effect based on specific design



we are publishing results that have false hits (TIEs)

1) incentive structure
2) publication bias
3) confirmation bias

crisis in psychology and possible causes


involves manipulating variables and the researcher(s) having control; the goal is to establish causal relationships; research study that allows us to infer causality through manipulation and control

experimental method


the variable that gets manipulated

independent variable


the variable that is measured

dependent variable


a single group is test before and after some treatment; there is manipulation but no control (not an experiment; no comparison / control group, and many potential confounds; lacks a basis for comparisons and does not offer control

pretest-posttest designs


degree to which diffs in DV can be attributed to IV vs another variable; extent to which you can make causal inference based on experiment; usually comes at cost to external validity; threatened by confounds, intact groups, attrition, participant reactivity, experimenter effects, and lack of appropriate controls

internal validity


DV value differs at different levels of IV; experimental and control score differently



temporal precedence; IV before DV

time-order relationship


confounds; any variable other than IV that could be affecting changes to DV

confounding variables / extraneous variables


used to give control to an experiment; established by random assignment because it averages individual differences across conditions; enables us to rule out alternative explanations due to any differences among participants

balanced groups / samples


groups are formed so they are similar on all important characteristics at the start of the experiment; used to account for individual differences among participants; used in a random groups design; most effective way to design an independent groups design

random assignment


is a way to deal w/ confounds; each block includes a random order of conditions; there are as many blocks as there are participants in each condition of experiment; should be determined in advance; used because it 1) gives us groups of equal size in each condition 2) controls for time related variables in studies that take a year or two 3) can help balance unknown differences between experimenters

block randomization


a threat to internal validity; pre-existing groups (could be differences in groups that already exist)

intact groups


a threat to internal validity; a loss of participants as study continues; sometimes okay
(often okay; ex: if computer crashed
not okay; ex: if one condition is impacted more than another)

attrition (mechanical loss / selective subject loss)