Brainscape's Knowledge GenomeTM


Top Flashcards (Certified)
All Flashcards

fucking stats > finals > Flashcards

finals Flashcards

(89 cards)

Start Studying
1
Q

ratio and interval

A

continuous data. full range of mathematical computation, height, weight, IQ, numbers of errors on a task

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

ordinal

A

ranks or ordered categories. likert scales, coffee cup sizes, rankings.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

nominal

A

named categories. gender, major, occupation

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

what tests do you not need to report two tailed

A

chi squared and F tests

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

list the steps of the hypothesis testing procedure

A

state hypothesis (null and alternative), locate critical region by defining the alpha value or level of significance, calculate sample statistics, make a decision, report

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

why would you want to study more than 2 groups

A

allows comparison of multiple IVs, allows comparison of multiple levels of an IV, can observe curvilinear effects, prevent interpolation and extrapolation errors,

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

what is the formula for type 1 error

A

1 - (1-alpha)^c where c is the number of comparisons

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

whats the issue with bonferroni correction

A

reducing type 1 error increases the likelihood of type 2 error (decreasing power)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

what are the chance effects comprised of

A

individual differences and experimentor error

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

larger df how influences the f distribution

A

less spread out to the right, so smaller critical values and easier to be significant

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

what is sample variance

A

chance differences de to random factors such as individual differences, experimental error

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

what does tukey tell you

A

honestly significant difference is the min distance between means needed for statistical significance. basically, how big the difference between two means need to be. instead of critical value, critical mean difference.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

if you want to indicate a positive correlation you would say

A

people who score higher on x tended to also score higher on y

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

what are requirements for ANOVA (independent and repeated)

A

random sampling, independent observations (for independent only), ratio/interval, normal distribution, homo variance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

pros of repeated anova

A

participants are their own controls, need fewer participants

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

cons of repeated anova

A

order effects, practice effects, may guess hypothesis, longer studies, limits possibilities for experimental manipulations

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

how are individual differences removed from f ratio of repeated anova

A

the individual differences are mathematically removed from denominator, it is methodologically removed from numerator

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

what is SSerror also called

A

residual

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

can you have a nonmanipulated IV for factorial anova

A

yes

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

how many interactions for a 3 IV factorial anova

A

3: axb, bxc, axbxc

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

spreading

A

effect exists at one level of IV and is weaker or nonexistent on another level

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

crossover

A

no main effects because effects are opposite at different level of IV (the X situation)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

what is a three way interaction

A

there is an interaction between two IVs but this interaction only exists under a certain IV. e.g. war crimes committed is significantly heightened if a durge player romances astarion, but war crimes committed are the same between durge and nondurge for nonastarion romances. however, this effect is only present if the durge sides with gortash.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

what is the goal of an anova

A

explain total variance by seeing how much is from IV, how much is natural variability

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
main effect null hypotheses
A has no effect on outcome
26
interaction null hypotheses
there is no interaction, all mean differences between treatment conditions are due to main effects of the factors. so the mean difference of A1 and A2 at B1, B2 are the same. (must be parallel when graphed)
27
do factorial anovas tell you the effect of one variable when controlling for the other
yes
28
what is covariance
the part of variance in x that overlaps with variance in y
29
assumptions for pearson's r
quantitative data, independent observations, random sampling, linear relationship
30
convergent validity
how strongly does the measure correlate with other measures of the same construct
31
discriminant validity
how strongly does the measure correlate with measures of unrelated consturcts
32
list the things that fuck with correlation
outliers, nonlinear relationships, restriction of range, heterogenous samples, conclusion of causation
33
how do heterogenous samples fuck with correlation
you might have a curvilinear relationship and a straight line put together... but then it would be analyzed like a straight line
34
what should you do before you look for the correlation and find the significance
plot it to see the scatterplot so you're not as sketchy
35
nonlinear and nominal/ordinal data can be analyzed for correlation using
point-biserial and spearman
36
in order to logically infer that x caused y, what do you need
x must preceed y (even if shortly before), x must covary with y, there cannot be confounding factors
37
what can the line you got by doing a linear regression do
make relationship easier to see, show central tendency of relationship, predict relationship
38
can you extrapolate regression
no, unless you have other supporting evidence beyond the regression itself
39
the more covariance the less or more precise a regression line is
more precise
40
what is the standard error of estimate
a measure of standard distance between regression line and the actual data point
41
what is variance
average of the squared differences from the mean
42
what is standard deviation
square root of the variance (average squared difference from mean)
43
what is goodness of fit
the proportion of the total sum of squares that is explained by our model (r squared)
44
what is the analysis of regression trying to answer
is the amount of variance predicted by the regression equation significantly greater than what we would expect by chance?
45
what is residual variation
variance in y that is not related to changes in x
46
what is regression variation
variance in y that is related or associated with changes in x
47
what are you testing for hypothesis testing for regression
is the slope (b) signficiantly different from zero? is b significantly different from: a historical valuee, a particular measure as defined by things like IQ, another sample?
48
how does larger n influence regression standard error
it decreases it. check the formula
49
how does larger standard deviation of the product of y and x influence standard error
it increases it. larger Sy.x indicates that there is more variability in the sample and thus cannot predict the slope with less error.
50
what does larger standard deviation of x do for standard error
more stable sample variances, so smaller error
51
simultaneous multiple regression
a standard multiple regression where you put all predictor variables in at once
52
hierarchical multiple regression
where you add predictor variables in predetermined sets. like first ones you want to control for etc.
53
stepwise multiple regression
computer program adds or removes predictor variables one at a time to optimise R^2. completely data driven
54
what does unstandardized slope tell you for multiple regression
what is the relationship between y and x? t test to see if significant. it is the change in y for every one unit change in x, holding all other xs constant
55
what does standardized slope tell you in multiple regression
the relative importance of each x
56
third variable problem
relationship between two variables is actually due to a third confounding one
57
how do you know its multiple regression by looking at the report
"even when... is controlled for" "independent of..." "correcting for..." "adjusted for..." "taking into account..."
58
mediation analysis
whether a third variable explains the relationship between x and y, identifies possible causal mechanisms. show IV predicts DV, IV predicts mediator, and that both IV and mediator prdicts DV. the IV when predicting DV along with mediator, should have a lesser predictive power (intuitively).
59
moderation analysis
assesses whether a third variable changes the relationship between x and y. possible interactions. you need an interaction term in the multiple regression with its own slope.
60
requirements for parametric tests
normal distribution, homogeneity of variance, numerical score per individual
61
h0 for chi squared test of independence
the frequencies for one variable are the same across levels of the other variable
62
which fe is for chi squared independence
the fancy one with fcfr/n
63
what does chi square statistic do
whether a set of proportions differ from another
64
cohen's w
effect size for both types of chi squared tests
65
phi coefficient
proportion of variance accounted for like r^2 for 2x2 matrix
66
cramer's v
modification of phi coefficient for larger matrixes for chi squared
67
requirements for chi square
independent observations (random sampling, each observed frequency is from a different participant), cochran's rule (expected frequencies should be greater than 5); more lenient version is cannot be less than 1, less than 20% of expected cell frequencies less than 5
68
how to deal with violating chi square assumptions
increase sample size, collapse categories
69
as a nonparametric test, what does the chi square subsitute?
pearson correlation, T, and ANOVA
70
when do you use nonpara tests
when standard para test requirements not met, and data consist of nominal or ordinal
71
aside from measurement data types and violated assumptions, why else might you want to do nonparametric tests
when there is too much variance. you want to convert the noise to a rank. when there are undetermined scores due to a person having to be stopped midway, etc.
72
multilevel modelling
carry out regression separately for different environments, average regression coefficients across courses.
73
factor analysis
find factors that are correlated with each other but not with others
74
factor loading
correlation coefficient of variable with a factor. variables may have loadings on each factor but only high loadings on one typicall
75
factor analysis
looks at different how lots of different observations correlate, determines theoretical constructs
76
meta analysis
quantitative summary of scientific literature with stat analysis
77
review paper
qualitative version of meta analysis
78
questionable research practices, what is it
decisions in design, analysis, and reporting that increase the likelihood of achieving a positive result (increasing type 1 possibility). basically, things people do with good intentions but still cause misinterpretation.
79
examples of QRPs
include: harking, phacking, not correcting for multiple comparisons, not reporting all measures, rounding off p values, only including data that worked out
80
does a p value of 0.01 mean 1% chance of false alarm?
no you bufoon
81
harking
hypothesizing after results are known. basically write it as if what you found was expected all along. reverse engineering of hypothesis
82
phacking
fishing around in data for significant results. redefining variables and running unplanned analyses. this is why you need to know what analyses you wanna do before doing it.
83
how do people discourage p hacking
more journals welcome null results. allow exploratory hypotheses and analyses. p curve analysis
84
not correcting for multiple comparisons
test-wise vs experiment wise alpha issue.
85
what can increase the chance of false positives (type 1 error)
having small samples, collecting additional dependent variables, peeking at data, dropping experimental condition
86
file drawer problem
only publish statistically significant results. thus hard to know if its a ture effect or a ton of type 1 error.
87
how to mitigate replication crisis
replication studies and p-curve analyses (plot p values to look for file drawer problem). preregistration of hypotheses, data collection rules, analytic strategies. calculate power a priori, report effect size and confident intervals, consider conditional probabilities
88
good research traits
open data, open source, open access, open methodology, open peer review, open educational resources. adequately powered, strong and isolate a question of interest research methods. reproducible computationally, methodologically, results.
89
how to consume science
dont believe everything you read (if effect unbelievable, it could be), how big is the sample (effects are unreliable if sample size too low. 2000 person study more reliable than 50). has it been replicated? is data avaliable?

fucking stats (2 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2025 Bold Learning Solutions. Terms and Conditions