Confidence Interval EXAM 3

Question 1

What is the purpose of the Confidence Interval?

Answer 1

Help to define the variability of a normally distributed dataset

(SD also does that)

Question 2

What is a parameter used to define datasets that are skewed (not normally distributed)?

Answer 2

-IQR and Median

Question 3

Estimation process

Answer 3

1. we have the target population (f.e. diabetics)
2. take a sample out of that population (we can’t test all diabetics)
3. determine the mean of that sample

->problem: that one sample is just a point estimate and we don’t know if it represents the whole population -> If we repeat the study the value of the sample could be somewhat different

Question 4

How does CI help with the problem that comes with single point estimates (means of one sample)

Answer 4

it is the interval that would contain a sample that is able to represent the means of the whole population with a given certainty
(f.e. CI =95% -> 95% certain that the data representing the whole population is in the given interval)

Question 5

What is the Distribution of Mean?

Answer 5

It is the normally shaped curve created from generating sample means (point estimates)

Question 6

What is the SEM?

Answer 6

Standard error (SD) of the mean (multiple means)

(multiple measurement values result in one mean -> and the SEM represents the SD of multiple means)

Question 7

What is the Mean of the means?

Answer 7

Multiple means build a normal shaped curve
in the middle of that curve, there is the mean of those means

Question 8

How does SEM define the Confidence Interval?

Answer 8

if we would repeat the experiment, the REAL population mean would fall into the range containing 95% of the data

approx. 2 SEMs (1.96) of the distribution of the means = 95% of the data points

Question 9

How is the SEM (SE) determined?

Answer 9

We would have to generate sample means over and over again -> we don’t do that in reality

-> We estimate the SE by using the formula:
SE = SD (SD of the samples, not the means) / square(n)
n= number of pt in the sample
(theoretical distribution)

Question 10

What happens to the SEM if the patient’s number goes up?

Answer 10

SEM goes down

Question 11

What is a Z-score?
Review

Answer 11

The number of SD’s to the left or right

Question 12

Definition of Confidence Interval

Answer 12

-The SEM is just like a z-score

95% CI = 1.96 z-score (or approx 2) x SEM
all values within the 95% interval are reasonable estimates of the population mean

Question 13

When are Confidence Intervals used?

Answer 13

It describes if the observed data (drug) is different enough from the placebo:
-Means
-Proportions
-HR, RR, OR

Question 14

What does a Forest Plot often describe in studies?

Answer 14

Ratios

-HR, OR, RR

Question 15

What is the 0 Hypothesis in a ratio?

Answer 15

Ratios like HR, OR, RR

The H0 would be 1 -> no difference between the two groups

Question 16

When are data points considered not statistically significant?

Answer 16

When they include the H0
for HR: they include 1 (no difference)
for the difference of mean: include 0 (no difference)

Question 17

The placebo group had a decrease in BP by 1;
CI 95% (-1 - 2)
Interpret that statement

Answer 17

the true data point falls with a 95% certainty within -1 and +2 -> meaning there was also an increase in BP

also includes 0 -> not significant?

Question 18

Can cohort studies show causality between exposure and disease?

Answer 18

No, cohort studies show an association between those two

RCT can show causality

Question 19

Which of the following results are statistically siginficant?