Chapter 6: Measuring Association with Continuous Outcomes

Difference of Means

Ex) comparing average birthweight between two large groups

Group A (nonsmokers) Mean birthweight: 3.4 kg

Group B (smokers): mean birthweight: 3.1 kg

compares the average outcome between two groups

Answer: 3.4 - 3.1 = -0.3.

Interpretation: Babies born to smokers weigh on average 0.3 kg less than those born to non-smokers

Useful for the average effect of an exposure

Standardized mean difference (SMD) uHHHHH

Ex)

Group A (nonsmokers) Mean birthweight: 3.4 kg

Group B (smokers): mean birthweight: 3.1 kg

expresses difference relative to variability (standard deviation) within the population

we need the pooled std: weigh each groups std by its sample size

Formula = Difference of the means / pooled std deviation

Answer: (3.4-3.1)/0.5 we obtain 0.6

Interpretation: The birthweight for smokers is 0.6 standard deviations lower than the non-smoking group. This shows a moderate association between smokers’ babies’ birth weights having reduced weight

Standardized mean differences (SMD) associations

SMD < or = 0.2 is small association

SMD > 0.2 but < 0.8 is moderate association

SMD > or = 0.8 is large association

Correlation coefficient ®

measures strength and direction of linear relationshop between two continous variables

Range -1 (perfect negative) to +1 (perfect positive)

an r close to 0 is no association between the two variables

Regression coefficient B (Beta)

Ex) Interpret B in this

Uses one variable to estimate the size of change in the other

B tells us how much the outcome changes for ach one-unit increase in the predictor

Answer: If we are prediciting blood pressure from salt intake, a B of 1.5 means that for every additional gram of salt, blood pressure goes up by 1.5 mmHg.