*Geek Box: Statistical Methods

*Geek Box: Linear Regression

You’ll likely come across the statistical method known as ‘linear regression’ very regularly when reading research. So what is it? Linear regression is a way to model the relationship between a dependent variable and one or more independent variables.

A dependent variable may also be known as an outcome variable or response variable: this is the factor whose variation we want to understand. An independent variable(s) may also be known as exposure variables or risk factors: these are the factors that may influence the occurrence of the outcome, or the size of the effect of the outcome.

In a simple linear regression, only one independent [exposure] variable is modelled for its association with the dependent [outcome] variable, while in a multiple [also known as multivariate] linear regression, more than one independent variable is modelled for their associations with the dependent [outcome] variable. A linear regression analysis predicts how the outcome either increases or decreases with an increase in the exposure.

For example, you could want to model how blood glucose levels are affected by increasing carbohydrate content, or how likely heart disease is to occur with increasing levels of LDL-cholesterol, i.e., you can predict the value of the outcome from the value of the exposure variable.

So lets take the present study to bring this concept to life; we know that intrahepatic triglycerides increased by 39% on the SFA diet, but we also know that participants gained ~1.5kg over the course of the intervention, when the goal was weight maintenance. So, to see whether the increase in IHTAG was more related to weight gain rather than diet, the authors conducted a linear regression to determine the relationship between IHTAG [the outcome, dependent variable], and the change in bodyweight [the exposure, independent variable]. Because the analysis predicts the value of one variable from another, it indicated that the change in bodyweight [the exposure variable] only predicted 17% of the increase in IHTAG [the outcome variable]. This indicates that it was the dietary intervention that was responsible for the majority of the increase in IHTAG.