Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

Psych MCQs > Critical Analysis > Flashcards

Critical Analysis Flashcards

You may prefer our related Brainscape-certified flashcards:
Psychology 101 flashcards GRE® Psychology flashcards
1
Q

Type I error

A

False positive, ie, falsely rejecting N0.
It is linked to the p-value, which arises from α, which is usually set at 0.05 and is the probability of committing a type I error.

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

Type II error

A

False negative, ie, falsely accepting N0.
It is linked to β, which is the probability of committing a type II error, and is usually equal to 20%.

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

Case Reports/Case Series - advantages & disadvantages

A

Case report = a report on a single patient with an outcome of interest.
Case series = a collection of reports on the treatment of individual patients.
Advantages: inexpensive, quick, generates hypotheses, good for rare diseases as longitudinal studies difficult to perform.
Disadvantages: cannot establish causality as no control group, little statistical validity.

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

Cross Sectional Study - advantages & disadvantages

A

= study of a sample of the population at a single instance in time.
Advantages: cheap, easy, quick, can demonstrate an association between two variables, can establish prevalence of disease in population being studied.
Disadvantages: cannot establish causality, subject to incidence-prevalence bias (where risk factor appears to cause the disease, when in reality, it actually affects the duration or prognosis of disease).

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

What is a Case Control Study?

A

= Involves identifying patients with the outcome of interest (cases) and then selecting controls (patients without the same outcome), and then looking back to see if they have the exposure of interest. Thus sample selection is always defined by the disease or the outcome being studied. The question being asked is if the cases have a greater exposure to the risk factor in question than the controls.

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

Case Control Study - advantages & disadvantages

A

Advantages: inexpensive, good for rare outcomes, not time consuming.
Disadvantages: not useful for rare exposures, recall bias, temporality cannot be established, good controls can be difficult to identify.

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

What is a Cohort Study?

A

A cohort study always begins from exposure, ie, it is defined by the exposure (in contrast to a case control study). A longitudinal study, which begins with identification of a cohort.
Can be divided into prospective and retrospective cohort studies depending on the type of follow-up.

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

Prospective Cohort Studies - advantages & disadvantages

A

Prospective = the direction of study is into the future.
Advantages: good for rare exposures, temporality can be established, multiple outcomes can be studied, control selection not a major issue, no recall bias.
Disadvantages: time consuming, cannot be used for rare outcomes, dropout rates can be an issue due to length of study (attrition bias).

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

Retrospective Cohort Studies - advantages & disadvantages

A

Retrospective = the follow-up period occurred prior to the study being started, and the cohort is assembled from historical records.
Advantages: similar as for prospective cohort studies (good for rare exposures, multiple outcomes can be studied, control selection not a major issue, no recall bias), although in some cases, it may be difficult to establish temporality.
Disadvantages: recall bias (of confounders).

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

What is an RCT?

A

A prospective interventional cohort study, with randomisation.
Randomisation is used to ensure equal distribution of factors that may affect the outcome in each group (ie, confounders). This creates groups that only differ in terms of the exposure/intervention.

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

N-of-1 RCT

A

Randomised double blind multiple crossover in same patient involving active patient and placebo.

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

Crossover Design RCT

A

Administration of 2 or more experimental therapies one after the other, to the same group of patients. Can be affected by order of treatments. There is also a carryover effect, which may be mitigated by a washout period between treatments.

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

Factorial Design RCT

A

Multiple treatments are compared separately or combined in a single trial.

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

Pragmatic RCT

A

Broad inclusion criteria to best represent real world practice. This increases external validity, but at the expensive of internal validity.

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

Systematic Review & Meta-analysis

A

Systematic review = a review of RCTs based on strict quality control.
Meta-analysis = mathematical analysis and visual interpretation of the systematic review.

Advantages: increases power of study to find a true effect, allows for more objective appraisal of evidence, heterogeneity can be investigated.
Disadvantages: only as good as the studies that are included.

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

Network Meta-analysis

A

Allows indirect comparison of two treatments that have not been directly compared in studies. This requires as assumption of homogeneity, similarity, consistency.

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

Types of Qualitative Studies

A
  • Ethnographic - immerses subjects in an unfamiliar culture
  • Case control (differs from quantitative case control studies)
  • Phenomenological - description of how participant(s) experience as certain event
  • Grounded theory - to explain why a course of action occurred as it did, eg, patient satisfaction study
  • Historical - describes past events to better understand present and future
  • Narrative model - over extended period of time
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

Hierarchy of Evidence

A

From bottom to top:
Expert opinion, editorials
Case series, case reports
Case control studies
Cohort studies
RCTs
Systematic reviews, meta-analysis

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

MOOSE

A

A reporting guideline from Meta-Analysis of Observational Studies in Epidemiology, which improves reporting.

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

START

A

Short term assessment of risk and treatability, a risk assessment tool.

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

QUOROM

A

Quality of reporting of meta-analysis.
A statement developed to help improve the quality of reporting of systematic reviews.

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

CONSORT

A

Consolidated standards of reporting trials - intended to improve the reporting of RCTs.

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

PRISMA

A

Preferred reporting items for systematic reviews and meta-analysis.
An evidence-based minimum set of items for reporting in systematic reviews and meta-analyses.

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

Interval Data

A

Type of ordinal categorical data in which the interval between each number is also a meaningful real number, but zero point is arbitrary.
For example, patient satisfaction on a 1-10 scale.

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

Dichotomous Data

A

Type of categorical data.
Variable that has only two possible outcomes.
For example, alive vs dead, smoker or non-smoker.

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

Ratio Data

A

Type of interval categorical data in which the zero-value is of meaning. For example, age.

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

Nominal Data

A

Type of categorical data.
Nominal categories.
For example, blood group, marital status.

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

Ordinal Data

A

Type of nominal categorical data data for which the order of the variables has meaning, but there is no mathematical relationship between data points.
For example; grading of tumours, ranking of depression as mild/moderate/severe.

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

Types of Function Data

A
  1. Dependent = is the result of the action of the independent variable. For example, the outcome of interest.
  2. Independent = under control of the investigator, such as the drug or treatment. For example, the exposure variable.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
30
Q

Types of Numerical Data

A
  1. Continuous = data that may take any value within a defined range. For example, height, BMI. Note this can be converted to categorical data by separating the data into groups, such as short, medium, tall.
  2. Discrete = value can only be whole numbers. For example, number of people.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
31
Q

Mean

A

Average value.
Used on normally distributed data. If the data is skewed, the mean will not be an accurate representation of the average.

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

Median

A

Represents the average when the data is skewed (not normally distributed). It is the middle value of the data points when they are listed in ascending order. If there are an even number of data points, then average the two midpoints.

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

Mode

A

The most frequently occurring observation.

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

Standard Deviation

A

A measure of dispersion of data, which indicates the variation around the mean.
Used for normally distributed data.
1 standard deviation - 68.2% of sample
2 standard deviations - 95.4% of sample
3 standard deviations - 99.7% of sample
Standard deviation is used to calculate standard error.

35
Q

Skewness

A

Data that is not normally distributed.
If data is normally distributed, then mean = median = mode.
If data is positively skewed, then mean > median > mode.
If data is negatively skewed, then mean < median < mode.

36
Q

Kurtosis

A

Relates to the peak of the distribution curve.
Mesokurtic - normally distributed.
Platykurtic - flat peak.
Leptokurtic - tall peak.

37
Q

Normally Distributed Data

A

One way to check if data is normally distributed is to calculate = mean - (std dev x 2).
If this is negative, it may indicate the data is not normally distributed.
Log transformation can transform data that is not normally distributed (non-parametric) to normally distributed data.

38
Q

P-value

A

A measure of precision, used to convey statistical significance. It gives the likelihood that a result could have resulted by chance.
However note that statistical significance is not equivalent to clinical significance (ie, the strength of the association between the variables).
The value is usually set at <0.05, based on α.

39
Q

Bonferroni Correction

A

The more times a test is run, the more likely a given result will be achieved by chance. If there are multiple variables being tested, then a Bonferroni correction should be used to counteract the multiple comparisons problem, by making the p-value more stringent.

40
Q

Standard Error

A

Indicates how close the sample mean is to the population mean, and is used to calculate confidence intervals.
Standard error = standard deviation / √N (where N is sample size).

41
Q

Variance

A

Measure of the spread of scores away from the mean. It is the standard deviation squared.

42
Q

Confidence Intervals

A

A measure of precision, which provide a range within which the observed effect would lie in the real world population.
Usually these are 95% confidence intervals, ie, 2 standard deviations each side of the calculated value.
These link back to standard deviation, from which you can calculate the standard error of the mean.
95% CI = mean +/- (1.96 x standard error)
Larger studies tend to have small CIs as the results will be more precise.

43
Q

Value of no effect

A

If the result is a ratio (ie, odds ratio), then the ‘value of no effect’ is 1, so if the CIs include 1, the result is not statistically significant.
If the result is a difference, then the ‘value of no effect’ is 0.

44
Q

Relative Risk

A

A measure of effect.
The ratio of risk in the exposed or treated group, divided by the risk in the control group.
Usually used in cohort studies, and some RCTs.
It can only be used with dichotomous data.
RR = EER/CER = (a/a+b)/(c/c+d)

45
Q

Hazard Ratio

A

Used in studies investigating harm/survival.
Differs from RR in that it represents instantaneous risk over the study time, not cumulative risk over entire study.
HR is the relative risk of an event happening at time t.
Hazard rate ratio = treatment HR / placebo HR.

46
Q

Odds Ratio

A

Ratio of events to non-events. It is similar to RR, but is often used in case control studies, where the outcomes are rare, so the ‘a’ and the ‘c’ in the denominator of EER and CER respectively can both be equated to 0.

OR = (a x d) / (b x c)

The interpretation of an OR differs from RR, because exposure doesn’t precede outcome.

47
Q

Risk Reduction

A

Risk reduction = 1 - OR
This can be used to calculate the risk reduction when an OR is less than 0, ie, a intervention is protective. For example, an OR of 0.6 is equivalent to a 40% reduction in risk.

48
Q

Standardised Mortality Ratio

A

Ratio of observed deaths to expected deaths, adjusted for age and sex.
If <1, this indicates decreased occurrence of event.
If >1, this indicates increased occurrence.

49
Q

ABI and ARI

A

ABI: Absolute benefit increase
= EER - CER = (a/a+b) - (c/c+d)

ARI: Absolute risk increase
= EER - CER = (a/a+b) - (c/c+d)

50
Q

NNT

A

Number needed to treat = 100 / ABI
This is the number of people needed to be given the intervention for 1 additional person to get the benefit.

In a population with a different prevalence, the NNT is different. It is calculated by NNT / PEER, where PEER is patient expected event rate.

51
Q

NNH

A

Number needed to harm = 100 / ARI
This is the number of people needed to be given the intervention for 1 person to be harmed.

52
Q

Relative Risk Reduction

A

The proportional change between the two groups = (EER - CER) / CER = ARI / CER

53
Q

Correlation

A

The strength of the relationship between two variables is measured by r, the correlation coefficient.
r<0.4 = low correlation.
r>0.6 = high correlation.
From r^2, you can find r, which gives the percent of variance explained by the variable in question.
Correlation coefficients can be seen on a scatter plot, where the regression effect will slope up if positive correlated, and down if negatively correlated.

54
Q

Regression

A

Helps to quantify the association between variables, ie, the degree of correlation.
The regression equation is: y = a + bx
where y is a value on the vertical axis, a is a constant, b is the regression coefficient, and x is the value on the horizontal axis.

55
Q

Sensitivity

A

How good the test is at correctly picking up the condition, ie, the true positive rate.
= true positives / (true positives + false negatives)
= A / (A + C)
Does not change with disease prevalence.

56
Q

Specificity

A

How good the test is at correctly excluding those without the condition, ie, the true negative test.
= true negatives / (true negatives + false positives) = D / (B + D).
Does not change with disease prevalence.

57
Q

PPV and NPV

A

Positive predictive value: if a person tests positive on the test, this is the probability that they have the disease.
PPV = A / (A + B)
Sometimes also called post-test probability.

Negative predictive value: if a person tests negative, this is the probability that they don’t have the disease.
NPV = D / (C + D)

Positive and negative predictive value change with prevalence of disease.

58
Q

LR positive & negative

A

Likelihood ratio +ve: the likelihood that a positive test comes from someone with the disease compared to someone without.
LR positive = sensitivity / (1 - specificity)

Likelihood ratio -ve: the likelihood that a negative test comes from someone with the disease compared to someone without.
LR negative = (1 - sensitivity) / specificity

Likelihood ratios are a function of the test and don’t change with disease prevalence.

59
Q

RUC

A

Receiver Operating Curve (RUC) = a graph which is used to establish which test is the better choice.
y-axis = sensitivity x-axis = 1- specificity
The perfect test has specificity and sensitivity of 1, but this is impossible.
The higher the area under the curve (AUC), the better the test.
It can also be used to calculate the optimal cut off point for a test.

60
Q

Kappa

A

Helps quantify the agreement between different observers, ie, the inter-rater reliability.
= (difference between observed and expected agreement) / (1 - expected agreement).

If the result is low, there is low agreement.
If the result is high, >0.8, there is almost perfect agreement.

61
Q

Statistical Test - parametric continuous data, two groups, unpaired

A

T-test (also called student T-test)

A one-sample T-test compares the mean of a single sample to a predetermined value.
An independent samples t-test compares the mean of one distinct group to the mean of another group.

The higher the value of t, the greater the statistical significance of the result.

62
Q

Statistical test - parametric continuous data, two paired groups

A

Paired T-test
Used when two samples are taken on same patient

63
Q

Statistical test - continuous parametric data, with two or more groups

A

ANOVA
Involves calculating the F statistic

64
Q

Statistical test - parametric continuous data, test of correlation

A

Pearson’s
Denoted by r

65
Q

Statistical test - parametric regression test

A

Least squares method

66
Q

Statistical test - dichotomous data

A

Chi squared test
Distribution free, so can also be used for parametric and non-parametric data

67
Q

Tukey Test

A

Post hoc test (done following ANOVA), in that the comparisons between variables are made after that data has already been collected.
An ANOVA can tell you if your results are significant overall, but it won’t tell you where those differences lie - but a Tukey test can.

68
Q

Statistical test - non-parametric continuous data, two groups, unpaired

A

Mann-Whitney U test

69
Q

Statistical test - non-parametric continuous data, two groups, paired

A

Wilcoxin matched pairs

70
Q

Statistical test - non-parametric continuous data, two or more groups

A

Kruskai-Wallis

71
Q

Statistical test - non-parametric continuous data, correlation

A

Spearman’s
Denoted by rs

72
Q

Statistical test - non-parametric data, regression

A

Non-parametric regression

73
Q

Fisher’s exact test

A

Used with dichotomous data, if the value of one of the data cells is <5 (instead of Chi-squared test)

74
Q

Null Hypothesis

A

Forms the basis of the research question. The aim is to reject the null hypothesis.
N0 = there is no difference between the two groups, no treatment effect.
NA (alternate hypothesis) = there is a difference between the two groups, a treatment effect is present.

75
Q

A priori hypothesis

A

An a priori hypothesis is generated in advance of data analysis.

76
Q

Post hoc analysis

A

A post hoc analysis is the hypothesis generated after the study.

77
Q

Power

A

Power is equal to (1 - β). Thus power is the probability of correctly rejecting N0.
Power is linked to sample size, and a larger sample size increases power.
Power is also related to reliability of measures, variance in sample and effect size.

78
Q

Bradford Hill Criteria

A

Criteria for causation:
- Strength of association - how large is the effect?
- Biological gradient - is there a dose-response relationship?
- Experimental evidence
- Temporality - does the exposure precede the effect?
- Biological plausibility
- Specificity - does altering only the cause alter the effect?
- Consistency - has the same association been observed by others, in different populations, using a different method?
- Analogy - is the association supported by similar associations?
- Coherence - does the evidence fit with what is known?

79
Q

Explanations for association

A

True causality, reverse causality, confounders, bias, chance.

80
Q

Phase I trials

A

Researchers test a new drug or treatment in a small group of people for the first time to evaluate safety, determine a safe dose range and identify side effects.

81
Q

Phase II trials

A

The drug is given to a larger group of people to see if it is effective and to further evaluate safety.

82
Q

Phase III trials

A

The drug or treatment is given to large groups of people to confirm its effectiveness, monitor side effects, compare to commonly used treatments, and collect information that will allow the drug or treatment to be used safely.

83
Q

Phase IV trials

A

Studies are done after the drug or treatment has been marketed to gather information on the drug’s effect in various populations and any side effects associated with long-term use.

Psych MCQs (10 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions