What is the problem with running post-hoc tests?

as you compare individual groups each one has a chance to be a Type I error

What is an alpha level?

probability of making an error when the null hypothesis is true

Define a between-participants design.

participants randomly assigned to 1 & only 1 level of the IV

What does the histogram of results from 10,000 runs of the simulation represent?

distribution of results that the model predicts after 10,000 runs if [z]

Why do they matter when interpreting statistical claims?

allow us to better understand the likelihood of certain events occurring

How do you describe the shape of a histogram?

symmetric, skewed, unimodal, bimodal, multimodal

What does it mean when we reject the null?

the null hypothesis doesn't provide a compelling origin story for our data

Define a paired-participants design.

participants are carefully assigned to 1 & only 1 level of the IV

How do we model the null hypothesis for correlations?

use parametric stats to generate results & be able to interpret the results

Given a regression constant and coefficient, how can you predict the criterion variable given a predictor?

with an equation: ŷ = a + bx -> given a specific x value, we can predict ŷ

you made to the end! if no one else has told you this: I'm proud of you & everything you've done this semester <3

<3 <3 <3, now onto the final!, omgggg, this final is gonna ruin me istg

How do we estimate the population variance?

We estimate the population variance via a formula: SD2 = ∑(X-M)2/N-1

How can you construct them?

y-axis: a measurement of how often data values popped up in the data set, x-axis lists each of the possible data values

What are confidence intervals?

estimate of the range of reasonable values of the parameter being reported

What can correlational analysis tell us?

if there's a relationship between 2 (or more) variables

What does one run of a simulation represent?

the # of [x] out of [y] that got it right if the model is true

What is within-groups variance?

measures how noisy the DV measurement is in this population, How spread out are the scores within each individual group?

What do effect sizes tell us?

It tells us how far the population would be shifted

What are dependent variables?

variable expected to change as a result of IV manipulation & measured

What does a high p value tell us?

we can't rule out the null hypothesis as a reasonable explanation

variable that systemically varies along w/ the IV so can't tell IV effects

new theory/hypothesis explains the relationship in the data, researcher noticing relationship in data, creating & presenting new theory, Hypothesizing After the Results are Known

What are some of the shortcomings of null hypothesis significance testing?

small p value only tells us that results were unlikely due to the null, p values don't tell us anything about the research hypothesis, a statistically significant result isn't necessarily a big/important result, Large p values only tell us that we can't rule out the null hypothesis

What are the different ways we can classify variables?

Qualitative, Quantiative, Discrete, Continuous, Nominal, Ordinal

Define a within-participants design.

all participants experience all levels of the IV

What can we conclude about a model if the data we have is consistent with the model?

model could be an origin story for our data

What is a correlation?

measuring the relationship between 2 variables

Why does the drawer effect matter?

higher probability of Type I error, we don't know how many tests occured that weren't statistically significant

What does the standard deviation tell us about a data set?

tells you about how far you expect any one value to be from the mean

What does a low p value tell us?

we can rule out the null hypothesis as a reasonable explationaion

studbud

K300 Statistics Final Exam Review

93 termsStudBud Administration21 plays

Imported from external platform

Flashcards93 cardsFPractice quizPHost liveL

Questions 93

Answers

1
What is the problem with running post-hoc tests?
as you compare individual groups each one has a chance to be a Type I error
as you compare individual groups, they all start to look the same
as you compare individual groups, they all start making no sense
2
What is an alpha level?
probability of making an error when the null hypothesis is wrong
probability of making an error when the null hypothesis is true
3
Define a between-participants design.
all participants experience all levels of the IV
participants are carefully assigned to 1 & only 1 level of the IV
participants randomly assigned to 1 & only 1 level of the IV
4
What does the histogram of results from 10,000 runs of the simulation represent?
the # of [x] out of [y] that got it right if the model is true
distribution of results that the model predicts after 10,000 runs if [z]
stimulation of wrong answers made by IU admins
5
Why do they matter when interpreting statistical claims?
force us to fail to reject the null
allow us to better understand the likelihood of certain events occurring
allow us to reject the null
6
How do you describe the shape of a histogram?
flat
symmetric, skewed, unimodal, bimodal, multimodal
7
Z scores are useful because this standardization doesn't allow you to easily compare scores from different distributions
True
False
8
Alpha levels are important because they're used to judge if a test statistic is statisically significant
True
False
9
What does it mean when we reject the null?
we can't rule out the null as a reasonable explanation for our data
the null is completely wrong
the null hypothesis doesn't provide a compelling origin story for our data
the null is absolutely right
10
Define a paired-participants design.
all participants experience all levels of the IV
participants randomly assigned to 1 & only 1 level of the IV
participants are carefully assigned to 1 & only 1 level of the IV
11
How do we model the null hypothesis for correlations?
use the matching applet to model the distribution of results
use an ANOVA to generate results & be able to interpret the results
use parametric stats to generate results & be able to interpret the results
use the coin-flipping applet to model the distribution of results
12
The "best fitting" line can be predicted by minimizing the distance b/w y values of predicted line & data's y values
False
True
13
Rejecting the null hypothesis doesn't tell us anything about how strongly the manipulation of the IV affected the DV
True
False
14
A Type II error is when the null is correct but we concluded that we should reject the null
False
True
15
Given a regression constant and coefficient, how can you predict the criterion variable given a predictor?
with an equation: SE = σ / √ (n)
with an equation: ŷ = a + bx -> given a specific x value, we can predict ŷ
with an equation: y = mx + b
16
How do you calculate a Z score for a single value?
with a formula: z = (x-μ)/σ
with a graph
with a modal
with an ANOVA
17
Parametric statistics are different from the simulate-build-model approach because software can setup models for us
True
False
18
you made to the end! if no one else has told you this: I'm proud of you & everything you've done this semester <3
<3 <3 <3
now onto the final!
omgggg
this final is gonna ruin me istg
19
How do we estimate the population variance?
with an ANOVA
with a model
We estimate the population variance via a formula: SD2 = ∑(X-M)2/N-1
20
How can you construct them?
y-axis: a measurement of how often data values popped up in the data set
x-axis lists each of the possible data values
i have no idea man
21
The 3 causal relationships are: A causes B, B causes A, or C has some effect
True
False
22
What are confidence intervals?
estimate of the mean of reasonable values of the parameter being reported
estimate of the mode of reasonable values of the parameter being reported
estimate of the median of reasonable values of the parameter being reported
estimate of the range of reasonable values of the parameter being reported
23
What can correlational analysis tell us?
if one variable caused another to occur
if there's a relationship between 2 (or more) variables
24
What are measures of central tendency?
Mean, Median, Mode
range
variance
standard deviation
25
What does one run of a simulation represent?
the # of [x] out of [y] that got it right if the model is true
distribution of results that the model predicts after 10,000 runs if [z]
26
Pre-registration helps with better & more thoughtful study designs and reduces the temptations of p-hacking and HARKing
False
True
27
What is within-groups variance?
measures how much IV manipulation affected the DV
How spread out are the mean values of each group or condition?
measures how noisy the DV measurement is in this population
How spread out are the scores within each individual group?
28
What do effect sizes tell us?
It tells you how far the null is off
It tells you how far the mean would be shited
It tells us how far the population would be shifted
29
What are dependent variables?
variable expected to change as a result of IV manipulation & measured
the variable that the researcher manipulates and has levels attached to it
variable that systemically varies along w/ the IV so can't tell IV effects
30
you interpret and “calculate” the probability notations used in Bayesian analysis via: P(H|D) = P(H) * P(D|H)/P(D)
False
True
31
What does a high p value tell us?
we can rule out the null hypothesis
we can't rule out the null hypothesis as a reasonable explanation
32
What are confounds?
the variable that the researcher manipulates and has levels attached to it
variable expected to change as a result of IV manipulation & measured
variable that systemically varies along w/ the IV so can't tell IV effects
33
A posterior is probability of an event before new data is collected, a prior is probability of an event after new data
True
False
34
What is HARKing?
new theory/hypothesis explains the relationship in the data
researcher noticing relationship in data, creating & presenting new theory
Subsequent pairwise comparisons to see exactly where the difference lies
Hypothesizing After the Results are Known
35
The null hypothesis says that the differences in our data are there because of noise, dumb luck, random chance, etc
True
False
36
Meta-analysis- running another experiment to see if replication results are extreme enough to reject the null hypothesis
False
True
37
What are some of the shortcomings of null hypothesis significance testing?
small p value only tells us that results were unlikely due to the null
p values don't tell us anything about the research hypothesis
a statistically significant result isn't necessarily a big/important result
Large p values only tell us that we can't rule out the null hypothesis
38
What are the different ways we can classify variables?
Qualitative, Quantiative
Not helpful, stupid
Discrete, Continuous, Nominal, Ordinal
confusing
39
Define a within-participants design.
participants randomly assigned to 1 & only 1 level of the IV
all participants experience all levels of the IV
participants are carefully assigned to 1 & only 1 level of the IV
40
Bayesian analysis is different from NHST because we update our beliefs after collecting some new data is observed
False
True
41
Replication drastically increases our belief that the original results weren't due to a Type I Error
True
False
42
Z scores differ from a “raw” score because they're sometimes called "standardized scores"
False
True
43
What can we conclude about a model if the data we have is consistent with the model?
model could be an origin story for our data
model is very unlikely to be the origin story for our data
model is wrong and doesn't tell us anything about our hypothesis
44
How can you gauge your level of confidence in the prediction made using a regression line?
r2
s2
c2
t2
45
What is a correlation?
manipulating 1 variable & measuring the other
measuring the relationship between 2 variables
46
Replication is method of combining results from multiple related studies
False
True
47
A formal results statement for t tests/ANOVAs is: There was/was not a significant effect on [x & y], t/F(N-1), p = #
False
True
48
A post-hoc test is subsequent pairwise comparisons to determine exactly where the difference lies
False
True
49
Why does the drawer effect matter?
higher probability of Type I error
we don't know how many tests occured that weren't statistically significant
we know how many tests were done without being statistically significant
50
What does the standard deviation tell us about a data set?
how many SDs a certain score is above/below the distribution mean
how much error we expect when estimating population mean w/ size N mean
tells you about how far you expect any one value to be from the mean
51
What does a low p value tell us?
we can't rule out the null hypothesis as a reasonable explanation
we can rule out the null hypothesis as a reasonable explationaion
the null is right
the null is wrong and stinky
52
A con of correlational studies is that correlation doesn't imply causation
False
True
53
What is the shape of the null distribution for an ANOVA and why is it shaped that way?
Positively skewed, F-ratios can't be 0
Symmetric, F-ratios can go under & above 0
Negatively skewed, F-ratios can't be over 1
54
We move from testing 2 groups to testing more than 2 groups by using variance measures to analyze more than 2 groups
True
False
55
What is a Z score?
measure of how far, on average, each score is from the mean
how much errorwe expect when estimating pop mean w/ N size sample mean
measure of how many SDs certain score is above/below the distribution mean
56
What is the file drawer effect?
only publishing not statistically significant results
Only publishing statistically significant results
publising results that have nothing to do with your original hypothesis
57
What are the null/research hypotheses for hypothesis testing in linear regression?
research: knowing the x value helps predict the y value
knowing the y value doesn't help predict the x value
null: knowing the x value helps predict the y value
null: knowing the x value doesn't help predict the y value
58
We measure "effect size" for a correlation via the correlation constant
False
True
59
How does linear regression differ from correlation?
it doesn't matter which one we call x & y
must decide which one will be used to predict & which will be predicted
z will always play a factor in the correlation
60
The equation for population variance is different from the equation for a sample's variance because of the "N-1"
False
True
61
A p value is the probability that we'd get a result at least as extreme as our experimental result if the null were true
False
True
62
HARKing is forcing researchers to make experimental decisions BEFORE they collect their data
False
True
63
A relative statistic is when you want to make a small change look big
True
False
64
What is between-groups variance?
measures how noisy the DV measurement is in this population
How spread out are the scores within each individual group?
How spread out are the mean values of each group or condition?
measures how much IV manipulation affected the DV
65
Why are models useful?
allow us to estimate the p value
b/c they stimulate complex process & create distribution of likely outcomes
allow us to estimate the variance of a data set
66
What are “base rates”?
probability of getting a result at least as extreme as the experimental one
# of times we'd expect some outcome to occur in non-intervention population
67
alright y'all... the end of the semester is here and you made it! just a little heads up: this quiz is long. get ready
how long is this?
huh?!
yay!
omg i'm so ready to be out of here
68
What are the sources of variability in the data from an experiment?
Variability due to Experimental Manipulation
Variability due to Individual/Random Differences
Variability due to political affiliation
69
What are independent variables (IVs)?
the variable that the researcher manipulates and has levels attached to it
variable expected to change as a result of IV manipulation & measured
variable that systemically varies along w/ the IV so can't tell IV effects
70
How can we fix those problems with post-hoc tests?
selecting a medium alpha level so that overall alpha level stays the same
selecting a bigger alpha level so that overall alpha level is above 5%
selecting a smaller alpha level so that overall alpha level is below 5%
71
What does rejecting the null hypothesis for an ANOVA tell us?
the result was unlikely to have been due to the null hypothesis
the result was unlikely to have been due to experimental manipulation
the result was likely to have been due to the null hypothesis
72
The correlational coefficient tells us the direction & strength of the correlation
True
False
73
Given a data set, what’s the right measure of central tendency to use? Why?
always use the mean
always use the median
always use the mode
it depends on the shape of the distribution
74
What is an ANOVA and how does it differ from a t test?
ANalysis Of RAnge, uses S statistic
ANalysis of Mean, uses SD
ANalysis Of VAriance, uses F statistic (aka F-ratio)
75
Correlational studies differ from experimental studies because they simply measure the variables & their relationship
False
True
76
A Type I error is when the null is incorrect but we concluded that we should fail to reject the null
True
False
77
The goal of linear regression is to use a line to make an exact prediction
False
True
78
What is a model?
formal description of the process that generates data
hypothesis that tells us that our research hypothesis is right
hypothesis that says that differences in your data are there b/c of noise
79
What if the data we have is inconsistent (very unlikely) with the model?
model is very unlikely to be the origin story for our data
model is right and tells us that our hypothesis is correct
model could be an origin story for our data
80
What are histograms?
ANalysis Of VAriance
frequency distributions
when researchers make a hypothesis after data is collected
81
What advantages does Bayesian analysis offer over NHST?
estimate probabilities of different models & see which ones are most likely
Bayesian analysis gives you a lot more info than testing a null model
Bayesian analysis gives you a lot less info than testing a null model
82
Effect size is measure of how far manipulation of the IV shifted the population measured in terms of the population’s SD
False
True
83
We evaluate a model once we have data by creating a simulation of the model
False
True
84
An absolute statsitic is when you want to make a bigger value look small
False
True
85
What is a t statistic? How does it relate to statistical significance?
tweak results after collecting data to have a statically significant result
parametric test statistic used to look at the difference between two groups
86
Define pre-registration.
stating what the reasearchers will do before the experiment starts
Subsequent pairwise comparisons to see exactly where the difference lies
researchers clearly defining the different parts of their experiment
87
P-hacking is tweaking your analysis before you collect data in order to make your results stastically significant
True
False
88
In linear regression, which variable is the predictor and which is the criterion variable?
predictor is y & criterion is x
predictor is x & criterion is y
89
How do we measure variability in a data set?
Standard Deviation (SD), range, variance
z scores
90
What is the standard error?
distance between the highest & lowest points
how many SDs a certain score is above/below the distribution mean
how much error we expect when estimating pop mean w/ size N sample mean
91
Sample size matters because data from large sample is likely to deviate away from the expected result
True
False
92
Meta-analyses make it easier to estimate effect sizes & allow the results to be generalized to a wider population
True
False
93
also some questions aren't listed here but those questions are posted on the quizlet! alright let's go!
ok
let's get started
thanks for the heads up!
alr cool

K300 Statistics Final Exam Review Flashcards | StudBud