Inferential Statistics Flashcards

1
Q

What form can inferential statistics take?

A

Estimation
Hypothesis Testing

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

Types for estimation

A

Point estimation
Interval estimation

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

What is estimation?

A

Using sample data we estimate the distribution of a parameter in the population from which the sample was drawn

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

What is point estimation?

A

Estimate a singe value for a parameter that will be close t true value of the parameter - effect size

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

What is interval estimation

A

Find an interval that has a given probability of including the true value of the parameter within its specified range

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

What is the interval in interval estimation?

A

Confidence interval

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

What is the probability in interval estimation?

A

Confidence coefficient

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

What is hypothesis testing?

A

We test the null hypothesis that a specified parameter of the population has a specified value by looking at the samples value

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

What are hypotheses?

A

Conjectural statements that provisionally link two variables

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

What are theories?

A

Sets of definite propositions or facts that are more or less verified already

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

How does one examine the relationship between two variables?

A

Probability theory

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

What is Poppers logic re hypothesis testing?

A

To prove something is very difficult.
To disprove something is relatively easy.
Hence science does not use the method of verification but methods of falsifiability.

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

What is the null hypothesis also known as?

A

H0

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

What do statistical methods try to do with respect to H0?

A

Try to refute this statement using statistical inference

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

What is another name for the alternate hypothesis?

A

H1

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

How can one state a hypothesis?

A

One-tailed
Two-tailed

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

What is a one tailed hypothesis?

A

Refers to the statement that differences between groups occurs in one direction only e.g. A->B

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

What would the alternative hypothesis be in a one-tailed hypothesis?

A

A is not -> B

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

What is a two-tailed hypothesis?

A

Refers to the statement that differences exist between two groups but the direction of the difference is not specified i.e. may be A->B or B->A

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

What would alternative hypothesis be in a two-tailed hypothesis?

A

A=B

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

What happens to significance levels in a two tailed hypothesis?

A

They are halved

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

Which type of hypothesis needs a larger difference to reject the null hypothesis?

A

Two tailed

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

Why do two tailed hypothesis need a larger difference to reject the null hypothesis?

A

Significance levels are halved

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

Which type of hypothesis are considered more rigorous?

A

Two-tailed

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

Why are two-tailed hypotheses considered more rigorous?

A

Significance level is halved so larger differences are needed to reject the null hypothesis

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
26
Q

How is the null hypothesis tested?

A

By gathering data relevant to the hypothesis and determining how well it fits H0

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
27
Q

What is used when we test how our data fits with H0?

A

Significant level, p

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
28
Q

What is the significance level, p?

A

The probability of rejecting H0 when H0 is true

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
29
Q

What does a higher significance level, p mean?

A

The higher the p, the better the fit between the data and H0

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
30
Q

What does a low p value suggest?

A

Casts doubt upon the validity of H0

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
31
Q

What can we assume if the value of p is very low?

A

We can reject H0

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
32
Q

What are random errors?

A

Fluctuations in direction in measured data due to precision limitations of measurement devide

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
33
Q

What are random errors often a result of?

A

Researchers inability to take measurement in the same way to get the same result

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
34
Q

What are systematic errors?

A

Reproducible errors that are consistently in the same direction

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
35
Q

What errors can occur during hypothesis testing?

A

Type 1
Type 2

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
36
Q

What happens in Type 1 errors?

A

Incorrect rejection of the null hypothesis - false positive claim in favour of research hypothesis

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
37
Q

What is the likelihood os a Type 1 error?

A

alpha

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
38
Q

At what alpha level can we mainly avoid Type 1 errors?

A

<0.05

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
39
Q

What is another name for alpha?

A

Level of statistical significance i.e. p

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
40
Q

What can lead to Type 1 errors?

A

Repeated testing of hypothesis using same data
Multiple subset analysis
Secondary analysis

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
41
Q

Why does multiple testing of same data lead to type 1 error?

A

At least one test will be positive in 20 if p is set at 0.05

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
42
Q

What is a Type 2 error?

A

Incorrect acceptance of the null hypothesis - false negative rejection of research hypothesis

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
43
Q

What is the name of the likelihood of a type 2 error?

A

Beta

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
44
Q

What can lead to Type 2 error?

A

Small sample size
Large variance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
45
Q

What refers to the power of the study?

A

1 - beta

46
Q

What is the traditional level of beta?

A

20%

47
Q

What is the traditional level of power?

A

80%

48
Q

What happens as we try to lower Type 1 error?

A

Risk of Type 2 error increases

49
Q

Define power

A

Ability of a study to detect a difference between two groups if such a difference truly exists

50
Q

What does power depend on

A

Sample size
Mean effect difference (effect size)
Variability of observations
Acceptable level of p

51
Q

What variability increases power?

A

Lower variability

52
Q

What should be run to find the variance?

A

Small pilots
Or from previously published works in similar clinical examples

53
Q

What is the formula for standardised difference?

A

Target difference in means / SD of observations

54
Q

What is standardised difference an expression of?

A

Effect size

55
Q

Who created the nomogram used to calculate sample size?

A

Altman

56
Q

What is used in a nomogram to calculate sample size?

A

Standardised difference and power values

57
Q

Which error is increased as p increased?

A

Type 1

58
Q

Methods to increase power

A

Larger p value
Larger sample size
Larger effect size
Reduce variability
One-sided test
Most powerful test that appropriate assumptions will allow

59
Q

What does it mean to use a larger effect size?

A

Consider only larger deviations from null hypothesis to be significant

60
Q

When might larger effect size not be desirable? e

A

If a small difference can have a huge clinical impact

61
Q

How can one reduce variability?

A

Making more precise measurements
Matching subjects

62
Q

What must one check before choosing to use a one sided test?

A

Check if it is possible to make strong (supported) assumptions

63
Q

Which tests are more powerful?

A

Parametric

64
Q

Which type of hypothesis is more powerful?

A

One tailed

65
Q

Purpose of CI

A

To see how close the approximation of a measure in a sample is to the population

66
Q

What does a smaller CI mean?

A

The better the representativeness of the sample to the population

67
Q

What does one need to look out for when interpreting the CI?

A

Degree of confidence
Width of the interval
Upper and lower limit
Capturing the value of no difference

68
Q

What is the common degree of confidence used?

A

95%

69
Q

How does one derive the degree of confidence?

A

From the complement of conventional p value which is 5%

70
Q

What will happen to CI if there is a higher degree of confidence?

A

Wider interval will be seen

71
Q

What does a wide interval at a fixed degree of confidence indicate?

A

That the estimate is not precise

72
Q

What does a narrow interval of CI suggest?

A

Very precise estimate

73
Q

What does width of the CI depend on?

A

Size of the standard error i.e. variability, which will depend on sample size

74
Q

Which type of studies give wide CI?

A

Small studies

75
Q

What does capturing the value of no difference suggest?

A

If the 95% CI crosses the 0 point for the difference between means then the result is not statistically significant.
Similar if it crosses 1 for ratio measures or infinity for inverse ratios (NNT)

76
Q

What is the value of no difference referring to?

A

The value at which the results are not statistically significant

77
Q

Value of no difference for means?

A

0

78
Q

Value of no difference for ratios?

A

1

79
Q

Value of no difference for NNTs?

A

Infinity

80
Q

How can one reduce the width of the CI?

A

Smaller degree of confidence level e.g. 90% instead of 95%
Reduce standard deviation
Take larger sample sizes

81
Q

Value or no difference for absolute risk reduction

A

0

82
Q

Value of no difference for relative risk reduction

A

0

83
Q

Value of no difference for relative risk

A

1

84
Q

What do CI inform us about?

A

Degree of confidence in the sample
Precision of a result
Clinical significance
Statistical significance

85
Q

Formula of effect size

A

Difference in outcomes between intervention and controls divided by SD

86
Q

What is effect size a measure of?

A

Difference in point estimates

87
Q

What does effect size refer to?

A

Group of indices (independent of sample size) differing in the mode of measurement of magnitude of treatment effect

88
Q

Importance of ES in meta-analyses

A

ES measures are the common currency of meta-analyses that summarise the findings from a specific area of research

89
Q

Why are ES helpful in meta-analyses?

A

As individual studies often report outcome using different scales so using ES helps consolidate findings

90
Q

What can be used to measure ES?

A

Cohens d

91
Q

What is Cohens D?

A

Standardised difference between two means

92
Q

Calculation of Cohens d

A

Difference mean mean M1 and M2 divided by SD of either group

93
Q

Grading of ES based on Cohens d

A

0.2 = small
0.5 = medium
0.8 = large

94
Q

How can ES be interpreted?

A

assuming control and experiment group values are normally distributed with equal SDs, effect size can be interpretted just like Z scores of standard normal distribution

95
Q

What does ES of 1 mean?

A

That the score of the average person in the experimental group is 1 standard deviation above average person in control

96
Q

What does ES 0 mean

A

50% of controls would be below average person in experimental group

97
Q

What does ES 0.1 mean

A

54% of controls would be below average person in experimental group

98
Q

What does ES 0.5 mean?

A

69% of controls would be below average person in experimental group

99
Q

What does ES 1 mean?

A

85% of controls would be below average person in experimental group

100
Q

What does ES 2 mean?

A

98% of controls would be below average person in experimental group

101
Q

What does ES of 3 mean?

A

99.9% of controls would be below average person in experimental group

102
Q

Who suggested the common language effect size (CLES)

A

McGraw and Wong (1992)

103
Q

What is CLES?

A

Probability that a score sampled at random from experimental group will be greater than score sampled from controls

104
Q

If p value is 0.05, how many times does one need to calculate data to get a positive result by chance

A

20

105
Q

What is Bonferroni correction?

A

To correct for multiple testing leading to false positive

106
Q

Disadvantage of Bonferroni correction?

A

Can lead to false negatives

107
Q

Formula for Bonferroni correction?

A

Significance level for multiple tested data is altered as (normal significance level / number of statistical analyses carried out)

108
Q

What does Bonferroni correction do to the outcome?

A

Treats each outcome as an individual event

109
Q

What is a family wise error?

A

Probability that any one of a set of comparisons or significance tests is a Type 1 error

110
Q

What is a false discovery rate?

A

Instead of controlling chance of any false positives (like Bonferroni), this controls expected proportion of false positives

111
Q

What tests can be used to avoid false positives when using multiple tests?

A

Bonferronis correction
False discovery rate
Scheffe test
Tukeys honestly significant difference test
Dunnet test