EBM Quiz #2

sample

a subset of the population, selected so as to be representative of the larger population (smaller that represents the population (infer it will happen to the whole population))- took 1000 kids, if kids exposed to second hand smoke they get asthma. so now we can infer the same will happen to all kids in the country (pick kids that are just like the rest of the kids in the country)

1. it's faster and easier

2. it's less expensive

3. study of the whole population is impossible in most situations

4. sample results are more accurate than results-based observations of a population

5. can estimate the error in resulting statistics for samples

6. samples can reduce heterogeneity

What are some reasons why we would use a sample rather than a whole population for a study?

simple random sample

probability method of sampling:

- every subject has an equal probability of being selected for the study (everyone has the same probability of being selected (putting names in a hat))

systematic random sample

probability method of sampling:

- every Kth item is selected

- K is determined by dividing the # of items in the sampling frame by the desired sample size (count us up (every 4th person might get picked))

stratified random sample

probability method of sampling:

- the population is 1st divided into relevant subgroups, and a random sample is then selected from each subgroup (break people into subgroups and pick people within the subgroup, in medicine it will be to break them up into age group and break it down based on each age)

cluster random sample

probability method of sampling:

- 2-stage process in which the population is divided into clusters and a subset of the clusters is randomly selected (divide population into clusters and pick some subsets of the clusters - row 1,2,3,4,5 (row 5 wins))

nonprobability samples

those in which the probability that a subject is selected is unknown and may reflect selection biases of the patient doing the study

probability is unknown

random assignment

the assignment of subjects to treatments is done by using random methods

- helps to ensure that the groups receiving the different treatment modalities are as similar as possible

computer programs to randomize patients

double-blind randomization

someone other than the investigator must keep the list of random assignments, so the investigator and subject both don't know which participants are in which groups

someone external party does the randomizing (BEST ONE for most probability)

H0 Null Hypothesis

a statement claiming that there is no difference between the assumed or hypothesized value and the population mean

- means "no difference"

example of H0 null hypothesis

What type of hypothesis is this an example of?

Vaccine experiment: The infection rates for the control and treatment groups in a vaccine experiment are equal.

H1 alternative hypothesis

a statement that disagrees with the null hypothesis

- your real, true hypothesis

Type I Error

an error in which the null hypothesis is rejected, when it's actually really true

- "say nothing happened when something happened"

Type II Error

an error in which they fail to reject the null hypothesis, when it is actually false

- "say something happened when nothing happened"

power

the probability of rejecting the null hypothesis when it is indeed false

- the ability of a study to detect a true difference

- want this to be high

P-value

the probability of obtaining a result as extreme as (or more extreme than) the one observed, if the null hypothesis is true (p<0.05 or 0.01)

- the probability that the observed value is due to chance alone

<0.05 or <0.01

- what we found is truly due to another variable and is not just by chance

What P-value are you aiming for with your results? What does this mean?

dependent variable

the outcome or response; "depends" on other variables

independent variable

the variable that occurs first; data that is input into a study

1. difference between the observed mean and the norm

2. amount of variability among subjects

3. number of subjects in the study

What are the 3 factors that play a role in deciding whether an observed mean differs from the norm?

T-test

- a test used a great deal in all areas of science

- to use this, observations should be normally interested

one-tailed (directional) t-test

when you have an expectation of what you want

- can be used when investigators have an expectation about the sample value, and they want to test only whether it is larger or smaller than the mean in population

two-tailed (nondirectional) test

can be used when investigators do not have a prior expectation for the value in the sample; they want to know if the sample mean differs from the population mean in either direction

- i.e. yes it does differ from the expected mean OR no it doesn't differ from the expected mean

Chi-square test

the test is used for 2 separate groups

chi-square test

used with counts or frequencies when 2 groups are being analyzed

- uses observed frequencies and compares them to frequencies that would be expected if no differences existed in proportioins

analysis of variance (ANOVA)

test for 3 or more groups

- asks if any differences exist at all among the means of the groups

- if yes, the investigator then makes comparisons among pairs of combinations of groups

linear regression

involves determining an equation for predicting the value of the outcome from values of the explanatory variable

regression will help you predict values; correlation is a simple yes/no, these two variables are/are not related (correlation describes the relationship)

What is the difference between regression and correlation?

multiple regression

the extension of simple (linear) regression to 2 or more independent variables