This tutorial will help you set up and interpret a multiple linear regression in Excel using the XLSTAT software. Linear regression is based on Ordinary Least Squares (OLS).
Not sure this is the modeling feature you are looking for? Check out this guide.
Dataset for running a multiple linear regression
The data have been obtained in Lewis T. and Taylor L.R. (1967). Introduction to Experimental Ecology, New York: Academic Press, Inc.. They concern 237 children, described by their gender, age in months, height in inches (1 inch = 2.54 cm), and weight in pounds (1 pound = 0.45 kg).
The Linear Regression method belongs to a larger family of models called GLM (Generalized Linear Models), as do the ANOVA. This dataset is also used in the two tutorials on simple linear regression and ANCOVA.
Goal of this tutorial
Using simple linear regression, we want to find out how the weight of the children varies with their height and age, and to verify if a linear model makes sense. Here, the dependent variable is the weight, and the explanatory variables are height and age: we have two of them so we choose multiple linear regression.
Setting up a multiple linear regression
Open XLSTAT
In the ribbon, select Modeling data / Linear Regression.
Select the data on the Excel sheet. TheDependent variable(or variable to model) is here the "Weight". Thequantitative explanatory variablesare the "Height" and the "Age".
Since the column title for the variables is already selected, leave the Variable labelsoption activated.
Go to the Outputs tab and activate theType I/III SSoption in order to display the corresponding results.
Click on OK to begin computation.
How to interpret the results of a multiple linear regression in XLSTAT?
Just as a reminder, multiple linear regression enables you to predict a variable depending on several others, on the basis of a linear relationship inferred by a supervised learning algorithm. If you want to establish a linear relationship between only two variables, do not hesitate to check our tutorial on simple linear regression.
The first table displays the goodness of fit coefficients of the model. The R² (coefficient of determination) indicates the % of variability of the dependent variable which is explained by the explanatory variables. The closer to 1 the R², the better the fit.
In this particular case, 63 % of the variability of the Weight is explained by Height and Age. The remainder of the variability is due to some effects (other explanatory variables) that have not been included in this analysis. These effects could be gender, geographical region, life habits, etc.
It is important to examine the results of the analysis of variance table (see below). The results enable us to determine whether or not the explanatory variables bring significant information (null hypothesis H0) to the model. In other words, it's a way of asking yourself whether it is valid to use the mean to describe the whole population, or whether the information brought by the explanatory variables is of value or not.
The Fisher's F test is used. Given the fact that the probability corresponding to the F value is lower than 0.0001, it means that we would be taking a lower than 0.01% risk in assuming that the null hypothesis (no effect of the two explanatory variables) is wrong. Therefore, we can conclude with confidence that the two variables do bring a significant amount of information.
The next tables display the Type I and Type III SS. These results indicate whether a variable brings significant information or not, once all the other variables are already included in the model.
The following table gives details on the model. This table is helpful when predictions are needed, or when you need to compare the coefficients of the model for a given population with the ones obtained for another population (it could be used here to compare the models for girls and boys). We can see that the 95 % confidence range of the Height parameter is very narrow, while we notice that the p-value for the Age parameter is much larger than the one of the Height parameter and that the confidence interval for the Age almost includes 0. This indicates that the Age effect is weaker than the Height effect. The equation of the model is written below the table. We can see that for a given Height, age has a positive effect on the Weight: when the Age increases by 1 month, the Weight increases by 0.23 pounds.
The table and the chart below correspond to the standardized regression coefficients (sometimes referred to as beta coefficients). They allow us to directly compare the relative influence of the explanatory variables on the dependent variable, and their significance.
The next table shows the residuals. It enables us to take a closer look at each of the standardized residuals. These residuals, given the assumptions of the linear regression model, should be normally distributed, meaning that 95% of the residuals should be in the interval [-1.96, 1.96]. All values outside this interval are potential outliers, or might suggest that the normality assumption is wrong. We have used XLSTAT's DataFlagger to bring out the residuals that are not in the [-1.96, 1.96] interval.
Out of 237, we can identify that 15 residuals are out of the [-1.96, 1.96] range, which makes 6.3% instead of 5%. A more in depth analysis of the residuals has been performed in a tutorial on ANCOVA The chart below allows us to compare the predicted values to the observed values.
The histogram of the residuals enables us to quickly visualize the residuals that are out of the range [-2, 2].
Conclusion for this multiple linear regression
As conclusion, the Height and the Age allow us to explain 63 % of the variability of the Weight. A significant amount of information is not explained by the model we have used. In a tutorial on ANCOVA, the Gender is added to the model to improve the quality of the fit.
The following video explains how to run a multiple linear regression in XLSTAT.
Was this article useful?
- Yes
- No
