epidemiology week two i guess

Random error

unpredictable error that occurs from sampling and measurement errors

reliable, precise

studies with little random error are: (2)

Systemic error

AKA bias; results in mistaken estimate of an exposure’s effect on the risk of disease, stems from the METHODS of the study

Internal validity

The results are true for the population of study subjects

External validity

AKA generalizability - findings from the study can be applied to a broader population than the one used for the study

Information bias, confounding variables, selection bias

Main types of systemic error (3):

Information bias

Arises from imperfect sensitivity or specificity of the test used to measure the exposure or outcome

Sensitivity

Ability of the test to correctly identify people with disease

Specificity

Ability of a test to correctly identify people without the disease

Imperfect sensitivity

When a test incorrectly identifies someone with the disease as not having one; false negative

Imperfect specificity

When a test incorrectly identifies someone without the disease as having the disease (false positive)

Confounding variable

observed effect of the exposure is mixed or confounded with the effect of an extraneous factor that is an intermediate step between the exposure and disease nor a consequence of the disease

Adjusted rate ratio

Used to account for confounding variables by forcing the distribution of the confounding variable to be the same in the exposed and non-exposed groups

Selection bias

results from the way people are selected or select themselves for participation or retention in a study; the process of selection produced a sample that is not representative of the population

additive, multiplicative

Risk factors can interact in these ways: (2)

Additive interaction

When the combined effect of two or more risk factors on a disease is the sum of their separate effects; assessed through risk difference

Multiplicative interaction

When the combined effect of two or more risk factors on a disease is the product of their separate effects; assessed through risk ratio

Interaction

The effect of one risk factor changes depending on the value of the second risk factor (e.g. the effect of aspirin on stroke changes based on whether the patient is a male or female).

Type 1 error

a statistical error in which we conclude that the outcomes are different (reject null hypothesis) when in reality they are NOT different

alpha

probability of type 1 error

Type 2 error

statistical error in which we conclude that the outcomes are not different (accept null hypothesis) when in reality they are different

beta

probability of type 2 error

p value

calculates the probability that the null hypothesis is true

Statistical power

the probability of concluding that there is a difference when one does exist; equal to 1-beta

1-beta

Statistical power is equal to

Central limit theorem

states that when you take a sufficiently large sample size from a population, the distribution of the sample means will be normally distributed around the population mean

Magnitude of true difference, sample size, alpha value

Statistical power depends on: (3)

Validity

ACCURACY; how well did the test measure what we wanted it to?

Reliability

CONSISTENCY; are the results reproduceable if the experiment is repeated?