Lecture 4: Statistical Significance

Null Hypothesis

Assertion that there is NO relationship between the independent and dependent variables 

disease in exposed = disease in unexposed

RR = 1.0

Alternative Hypothesis

Assertion that there is some relationship between the independent and dependent variables

Two-sided hypothesis

The occurrence of disease is not the same in the exposed and unexposed groups

RR not equal to 1.0

One-sided hypothesis

The occurrence of disease in the exposed group is greater or less than the occurrence of disease in the unexposed group

RR > 1.0 or RR < 1.0

Measure of Association

measures that assess the strength of the statistical relationship between a given study factor and disease, disability, or death

Relative risk

Compares risk of disease in exposed group relative to the risk of disease in the unexposed group

“The risk of disease among the exposed is X times higher/lower than the risk of disease among unexposed” 

Risk in exposed / risk in unexposed

Relative Risk Equation

Risk in the exposed / risk in the unexposed

Risk Ratio Equation

Risk Difference

The amount of risk of disease among the exposed that can be attributed to the exposure

The number of cases of disease among the exposed that would not occur if the exposure was eliminated

Risk in the exposed - risk in the unexposed

P-Value

calculated from statistical models that are thought to describe the pattern of observations when chance alone is the sole reason for their variability

the probability of obtaining a result at least as extreme as that observed in the study by chance alone

*Assuming no association between the exposure and outcome (null is correct)

0-1

influenced by both the magnitude/strength of the association and the sample size

small

P-value reflects both the magnitude of difference between the groups and the sample size:

in a large sample size, a ___ difference may be statistically significant

large

P-value reflects both the magnitude of difference between the groups and the sample size:

in a small sample size, a ___ difference may not achieve statistical significance

statistically significant

if the p-value is below the cutoff, the result is said to be ___ ___ (you should have less faith that your null hypothesis/assumptions are correct)

leaning towards concept that there is a relationship vs there is NOT

95% confidence interval

a level of certainty about our estimate RR

if we repeatedly make new estimates using exactly the same procedure, the confidence intervals would contain the true value 95% of the time

A single estimate that, if repeated indefinitely, would result in 95% of the confidence intervals formed containing the true value

how “good” an estimate is

NOT statistically significant

if the interval contains the null value, the result it…

statistically significant

if the interval does not contain the null value, the result is…