Statistics Flashcards

1
Q

Categorical variable

A
  • Subject is placed into categories

- Two possible outcomes (yes/no, male/female)–> dichotomous

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

Ordinal

A
  • Inherent order
  • Ranked
  • May be summarized by a median value
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Continuous

A
  • Age, weight, height, lab values

- Means and medians used to summarize

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

Unpaired

A
  • Independent means the values of one group cannot be predicted from the other
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Paired

A
  • Paired the values of one group may be predicted from the other ( patient measured before and after therapy)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

1, 2, and 3 standard deviations

A

68.3, 95.8, and 99.7

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

Standard error of the mean

A
  • Because samples drawn from an underlying population do not each produce the same mean (but tend to cluster around the same value), one much calculate the range of where the true (unknown) population mean lies
  • The greater the sample size the smaller the SEM
  • Always less than SD
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Null Hypothesis

A

There is no relationship between two phenomena

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

Chi Squared

A
  • Also called Pearsons

- Suitable for unpaired categorical date from large samples

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

T test

A
  • Used for continuous data

- Used for comparing two sample means from either independent (non paired T test) or matched (paired T test) samples

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

ANOVA

A

Used to compare the means of several groups, instead of just two
- Looks at the difference within and between groups

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

Confidence interval

A
  • A range of values that there is a specificed probability that value of a parameter lies within it
  • 95 percent CI can be estimated if the mean and the SEm are known
  • If it crosses zero–> no good
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Type I error

A
  • Failing to reject the null hypothesis
  • Saying a relationship does except when it doesn’t
  • Alpha error set at 0.05
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Type II error

A
  • Failing to reject the null
  • Saying there is no difference when there is a difference
  • Beta error (0.2)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Which error is better? Type I or Type II

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

Absolute risk

A
  • Chance of developing a disease over a time period
17
Q

Absolute risk reduction

A
  • The absolute risks of different exposures can be compared to each other
  • Risk of drug A is 15%; risk of drug B is 5%–> 10% risk reduction
18
Q

NNT

A

The inverse of the absolute risk reduction 1/0.10 = 10

19
Q

Relative risk (risk ratio)

A
  • used to compare risk in two different groups with different risk factors exposures
  • Calculated by dividing the absolute risk in the group exposed to the risk factor by the absolute risk in the unexposed group
  • RR>1 exposure associated with increased risk of disease
  • RR=0 no association
  • RR <1 exposure is protective
20
Q

Relative risk reduction

A

1-RR

- Tells you by how much the exposure reduced or increased the risk

21
Q

Odds ratio

A
  • Used in case control ( no incidence rates).

- Calculated by dividing the odds of exposure among cases by odds of exposure among controls

22
Q

Hazard ratio

A
  • Risk ratio gives you the cumulative risk of a time span, while HR gives you the instantaneous risk at a particular point in time
23
Q

Regression analyses

A
  • One way model in which predictor independent (x) variables are thought to affect the dependent outcome (y) variable but not vice versus
  • Linear regression (continuous)
  • Logistic regression (categorical)
24
Q

Sensitivity

A
  • Percentage of people with disease who test positive
  • TP/TP+FN
  • Negative test rules out disease
  • SNOUT
25
Q

Specificiity

A
  • Percentage of people with a negative test who don’t have disease
  • TN/TN+FP
  • Positive test rules in disease
  • SPIN
26
Q

Positive predictive value

A
  • Percentage of people who test positive who do have the disease
  • TP\TP+FP
  • depends on prevalence
27
Q

Negative Predictive value

A
  • Percentage of people who test negative who do not have the disease
  • TN\TN+FN
28
Q

LIkelihood ratio

A
  • Evaluates the utility of diagnostic tests
  • LR + = sensitivty/1- spec
    = LR- = 1- sensiviity /spec
29
Q

ROC

A
  • Allows visual comparison of the performance of a set of different criteria for a diagnostic or screening test
  • The nearer the curve to the left upper corner, the better the test is
  • AUC , high number the better
30
Q

Systemic review

A
  • Use explicit methods to perform a thorough lit search and critical appraisal of individual studies
  • Prereq to meta analysis
31
Q

Meta analysis

A
  • Stat procedure for synthesizing quantitative results from different studies