Gamla tentor del 2 Flashcards
1
Q
What is a Type I error?
A
Rejecting a true null hypothesis.
2
Q
What is a Type II error?
A
Failing to reject a false null hypothesis.
3
Q
How are Type I and Type II errors related?
A
Decreasing the risk of a Type I error increases the risk of a Type II error.
4
Q
What does a boxplot show?
A
The distribution of values of an interval-scale variable.
5
Q
When do you use Pearson’s correlation coefficient r?
A
When both variables are measured on an interval scale.
6
Q
What risk increases when performing many pairwise comparisons of means?
A
Type I error.
7
Q
What does p represent in NHST (Null Hypothesis Significance Testing)?
A
The probability of observing the data assuming the null hypothesis is true.
8
Q
Why is the null hypothesis called “null”?
A
It represents the assumption of no difference or no effect.
9
Q
What is the main factor researchers use to control statistical power?
A
Sample size.
10
Q
What does the significance level (alpha) represent in psychological research?
A
The probability of committing a Type I error.
11
Q
What is a confidence interval centered around?
A
The sample mean.
12
Q
What happens to the mean if one value in a distribution changes?
A
The mean also changes.
13
Q
What does it suggest when two or more confidence intervals do not overlap?
A
The population means are likely different.
14
Q
What does it mean if p = .067 in a hypothesis test with alpha set at .05?
A
The test may contain a Type II error (false negative).
15
Q
What kind of variables does a Chi-square test of independence involve?
A
Two nominal variables.
16
Q
What does the Chi-square test determine?
A
Whether two categorical variables are associated.
17
Q
How many levels are there in a factor if its degrees of freedom (df) is 3?
A
4 levels (df + 1).
18
Q
If the error df is 28 and there are 4 groups, what is the total number of participants?
A
32 participants.
19
Q
What is the group size if there are 4 groups and 32 participants?
A
8 per group.
20
Q
What does it mean if F(3,28) = 1.68 and p = .495?
A
The result is not statistically significant.
21
Q
What does a p-value of .0004 for an interaction effect indicate?
A
The interaction effect is statistically significant.
22
Q
What is the degrees of freedom for an interaction between two factors with 2 levels each?
A
1 (1 × 1 = 1).
23
Q
If there are 20 participants and 4 groups, how many participants per group?
A
5 per group.
24
Q
What is the F-ratio if MS = 400 and MS_error = 17.46?
A
F = 22.91.
25
What is the goal of data analysis?
To determine whether the data supports a claim about cognition or behavior.
26
What is required for effective data analysis?
A high-quality study with good sensitivity.
27
What is important when using computer-based analysis?
Understanding what and how to analyze, and being able to interpret the output.
28
What is the most well-known statistical software?
SPSS (preferably version 28 or later); JASP is also popular and free.
29
What variable types are relevant in SPSS?
* Nominal: Categorical (e.g., gender, group)
* Ordinal: Ordered categories (e.g., low, medium, high)
* Interval: Continuous numbers (e.g., temperature)
30
What are the two main types of analysis?
Experimental analysis and survey (correlational) analysis.
31
What are the steps in experimental analysis?
* Get to know the data
* Summarize the data
* Confirm what the data reveals
* Confidence intervals
* Null hypothesis testing
32
What is involved in 'getting to know the data'?
* Clean data
* Handle missing/impossible values
* Check for meaningfulness
* Transform data (e.g., units or percentages)
33
How do you summarize data?
* Central tendency (mean, median, mode)
* Spread (range, standard deviation, standard error)
* Visualize (graphs)
* Effect size (Cohen’s d: 0.2 = small, 0.8 = large)
34
What is a stem-and-leaf plot?
A way to visualize the shape and distribution of data more intuitively.
35
What is a bimodal distribution?
A distribution with two peaks, suggesting two hidden groups; problematic for standard tests.
36
What are outliers and how should they be handled?
* Extreme values increasing variability and reducing statistical power.
* Causes: measurement/input error, distractions, misunderstood instructions.
* Decide before the study which data to include.
37
What is a boxplot?
A graph showing data distribution in four quartiles.
IQR = Q3 − Q1
Outliers = values beyond Q1 − 1.5×IQR or Q3 + 1.5×IQR
38
What is standard deviation?
Indicates the spread of data around the mean.
39
What is the standard error of the mean (SE)?
Reflects how accurately the sample mean estimates the population mean.
40
What is a confidence interval (CI)?
A range where the true value likely falls.
41
How do you analyze survey (correlational) data?
1 Summarize the data
2 Create a scatterplot
3 Look for linear trends
4 Calculate the correlation coefficient (r):
* 0 = no correlation
* 1 = perfect correlation
* 0.25–0.75 = moderate correlation
5 Test significance using p-value (SPSS: “Sig.”)
42
What are the three steps of null hypothesis testing?
* Formulate the null hypothesis (no effect).
* Calculate the probability of the observed data assuming the null is true.
* Reject or retain the null based on the significance threshold (typically 0.05).
43
Why is null hypothesis testing necessary?
Because sample data is subject to variability.
44
What does it mean that hypothesis testing is inductive and indirect?
* Inductive: Generalizes from limited data.
* Indirect: Assumes no effect initially and seeks evidence to reject that assumption.
45
What happens in Step 1 of null hypothesis testing?
* Define H₀: Mean A = Mean B
* Define H₁: Mean A ≠ Mean B (alternative hypothesis)
46
What happens in Step 2 of null hypothesis testing?
* Calculate a p-value (0 to 1).
Using a t-test for two groups
Using a ANOVA for more than two groups
47
What happens in Step 3 of null hypothesis testing?
* Set a significance level (alpha, ɑ)
* Compare the p-value to ɑ.
48
What does a statistically significant result mean?
That observed differences are unlikely to occur by random chance alone.
49
What does statistical significance not tell us?
* Effect size
* Practical or scientific importance
* Degree of significance
50
What are common causes of a non-significant result?
* Small sample size
* High variability
* Unclear instructions
* Measurement issues
51
What are Type I and Type II errors?
* Type I error (α): Incorrectly rejecting a true null hypothesis.
* Type II error (β): Incorrectly retaining a false null hypothesis.
52
How should you phrase your conclusions in scientific writing?
Do not say 'prove' — instead, say: 'The results support/do not support…'
53
What can statistical analysis not replace?
Replication — findings need to be confirmed by repeated studies.
54
What are common misunderstandings about p-values?
❌ A low p-value = large effect or good method
✔️ It only means the data is unlikely under H₀.
❌ p-value = probability that H₀ is false
✔️ It's the probability of the data, assuming H₀ is true.
❌ Non-significance = H₀ is true
✔️ We lack evidence to reject H₀ — not proof of its truth.
❌ Retaining H₀ = no group differences
✔️ We just don’t have enough support to claim there is a difference.
55
What is statistical power?
The probability that a test will detect a true effect. Recommended minimum: 0.8 (80%)
56
What affects statistical power?
* Alpha level
* Effect size
* Sample size
57
What are the risks of low power?
* Increases the chance of non-significant results
* Inflates observed effect sizes
* Raises risk of Type I errors at the publication level
58
What can power analysis estimate?
The required sample size to detect an expected effect.
59
What is sensitivity in an experiment?
The likelihood that a study will detect an effect if it truly exists.
60
Should you use a one-tailed or two-tailed test?
One-tailed: Cannot detect unexpected effects; Two-tailed: Detects effects in both directions — recommended.
61
When is a t-test used?
To examine whether an independent variable with two levels has an effect on a dependent variable.
62
What are the two types of t-tests?
* Independent t-test
* Paired samples t-test (also called dependent t-test)
63
What are the assumptions of the independent t-test?
* IV consists of two independent groups
* DV is continuous
* Observations are independent
* No significant outliers
* DV is approximately normally distributed in both groups
* Equal variances in both groups (homogeneity of variance)
64
How is the t-value reported?
t(df) = t-statistic
df (degrees of freedom) = n₁ + n₂ – 2
65
What test do you use when comparing more than two means?
ANOVA (Analysis of Variance)
66
What is ANOVA, how is it written, and what effect sizes are used?
A statistical test to determine whether an independent variable has a significant effect on a dependent variable, reported as an F-test.
Effect size format: Eta squared (η²), Cohen’s f.
67
When do you use a one-way vs two-way ANOVA?
* One-way ANOVA: One independent variable (between-group design)
* Two-way ANOVA: Exactly two independent variables
68
What does a high F-value in a one-way ANOVA indicate?
A higher F-value suggests a greater effect.
F = 1 indicates no effect. F > 1 = effekt
69
What can we conclude from the overall ANOVA F-test?
That there is more variability among group means than would be expected by chance.
70
How is Eta squared interpreted?
* η² = 0.02 → small
* η² = 0.13 → medium
* η² = 0.26 → large
71
How can we determine where the effect comes from in ANOVA?
* Compare confidence intervals
* Perform pairwise comparisons of means
72
What are the assumptions of one-way ANOVA?
* DV is continuous
* IV is categorical with 2+ independent groups
* Observations are independent
* No significant outliers
* DV is approximately normal in each group
* Homogeneity of variance
73
What is the problem with many pairwise comparisons?
Increases the risk of Type I error (false positives).
74
When is a t-test vs. ANOVA used in within-subjects design?
* t-test: when comparing two conditions
* ANOVA: when comparing more than two conditions
75
What are alternative names for the paired t-test?
* Paired t-test
* Dependent t-test
* Repeated measures t-test
* Within-subjects t-test
76
What is a difference score?
The mean of one condition minus the mean of another condition.
77
What are the assumptions of a paired t-test?
* IV is categorical with two related groups (e.g., pre-post)
* DV is continuous
* No significant outliers in difference scores
* Difference scores are approximately normally distributed
78
How is Cohen’s d calculated in a paired t-test?
Cohen’s d = M_diff / SD_diff
(In SPSS, listed as “PointEstimate” in the output tables)
79
How is repeated measures ANOVA conducted?
Perform an overall F-test.
80
What does SS stand for in ANOVA?
SS = Sum of Squares.
81
What are the assumptions of repeated measures ANOVA?
* IV consists of two or more related measurements
* DV is continuous
* No significant outliers in difference scores
* Difference scores are approximately normal
* Sphericity must hold.
82
What is sphericity and how is it tested?
Sphericity means that the variances of all pairwise differences between conditions are roughly equal.
83
What is the difference between a complete and incomplete design in within-subjects ANOVA?
* Complete: Data is summarized before ANOVA
* Incomplete: ANOVA is performed directly on raw scores.
84
What is the overall F-test?
A statistical test that provides a p-value indicating whether group means differ more than expected by chance.
85
What is the name of the effect size commonly reported in repeated measures ANOVA?
Partial eta squared (η²_partial).
86
What is a two-way ANOVA?
An ANOVA used in a 2x2 factorial design, testing the effects of two independent variables.
87
What are the assumptions of a two-way ANOVA?
* IVs must be categorical
* DV must be continuous
* No significant outliers in group differences
* The DV should be approximately normally distributed
* Homogeneity of variances across all combined levels.
88
What are the main steps in a two-way ANOVA analysis?
* Perform an overall F-test
* Determine if there is a statistically significant interaction
* If yes: analyze simple main effects and pairwise comparisons.
89
What are simple main effects?
The effect of one independent variable at a specific level of another independent variable.
90
What is a Chi-square test (χ²)?
Tests whether two categorical variables are statistically associated (dependent).
91
What kind of data is used in a Chi-square test?
Categorical variables (not ordinal or continuous).
92
What are the assumptions of the Chi-square test?
* Two categorical variables
* Independent observations
* Expected cell values must be 5 or more.
93
What should you use if expected cell values are less than 5?
Fisher’s Exact Test.
94
What does a Chi-square p-value indicate?
The probability of observing the data if the variables are independent.
95
How do you interpret the strength of association in a Chi-square test?
Use Phi or Cramér’s V.
96
When do you use Phi and when Cramér’s V?
* Use Phi for 2x2 tables; use Cramér’s V otherwise.
97
How is the strength of association classified with Phi/Cramér’s V?
* 0.1 = weak
* 0.3 = moderate
* 0.5 = strong
98
When should non-parametric tests be used?
When assumptions of parametric tests (e.g., normality) are not met.
99
What do parametric tests analyze?
Means.
100
What do non-parametric tests analyze?
Medians.
101
What kind of dependent variable is needed for parametric tests?
Continuous.
102
What kind of dependent variable is needed for non-parametric tests?
Continuous or ordinal.
103
What is an example of Chi-square use in observations?
"Took the elevator (yes/no)" vs. "Used door button (yes/no)"
104
What distribution do parametric tests assume?
Normal distribution.
105
Do non-parametric tests assume any specific distribution?
No, they are distribution-free.
106
How do outliers affect parametric vs. non-parametric tests?
Parametric: sensitive; Non-parametric: robust.
107
Which type of test generally has higher statistical power?
Parametric tests.
108
What non-parametric test corresponds to the Independent t-test?
Mann-Whitney U test.
109
What non-parametric test corresponds to the Paired t-test?
Wilcoxon Signed-Rank test.
110
What non-parametric test corresponds to Between-subjects ANOVA?
Kruskal-Wallis test.
111
What non-parametric test corresponds to Repeated Measures ANOVA?
Friedman test.
112
What is an example of Chi-square use in experiments?
Two groups choosing option A or B.
113
What is an example of Chi-square use in surveys?
"Owns a pet (yes/no)" vs. "Believes animals have consciousness (yes/no)".
114
How should results from a complex design be reported?
Provide a description of variables and definitions of levels
Include descriptive statistics for each cell (means, CIs)
Report F-tests with p-values for main and interaction effects
Include effect sizes for each effect
Conduct power analysis for non-significant results
If interaction is found, analyze simple main effects
Provide a verbal explanation of interaction and main effects
Report pairwise comparisons if needed
End with a clear conclusion for the reader
115
How do you calculate effect size in a two-way ANOVA?
Use partial eta squared (η²_partial)
116
How does the calculation of F look in a separate between-subjects ANOVA?
The calculation of F (MS/MS error) uses a MS error based on one level.
117
What is the minimum number of levels required for pairwise comparisons to be relevant?
Three levels.
