After learning how to clean data in JMP, start basic statistical analysis. In this article and the following articles, we will introduce the analysis process of linear regression one by one.
Before performing linear regression multivariate analysis, it is often necessary to perform univariate analysis first. However, the univariate analysis of most of the commonly used software operations is cumbersome, and it is necessary to continuously replace the independent variables to repeat the operation in order to carry out univariate analysis of multiple independent variables.
However, in JMP The software can realize the univariate analysis of all independent variables at one time. Today's article is going to share how to use JMP Complete a one-time multivariate one-way analysis.
Why do linear regression?
Multivariate analysis is more common in practice than the univariate analysis with only one group variable described earlier, because the vast majority of outcomes are affected by more than one factor. For example, comparing the efficacy of two different drugs in the treatment of diabetes, even if there is a difference in efficacy between the two groups, it may not be attributable to drug factors, because many factors such as BMI, blood pressure, diet, etc. will also affect the efficacy of drug treatment. If these Unbalanced factors between the two groups will affect the evaluation effect of the drug. Therefore, in practice, we often use multivariate analysis.
What we usually say is "adjusting" for factors such as age and gender, or that smoking is an "independent prognostic factor" for lung cancer, these refer to multivariate analysis. Linear regression is often used in multivariate analysis of quantitative outcomes, and is mainly used to describe the linear relationship between one or more independent variables and quantitative outcomes. When we want to explore risk factors for a quantitative outcome, we usually first consider linear regression, and then consider other alternatives if linear regression is not suitable.
Univariate Analysis of Linear Regression
The so-called univariate analysis, that is, every time you do a regression, put it first 1 independent variables, and observe the relationship between each independent variable and the dependent variable.
At present, many statistical softwares can only put one independent variable in for univariate analysis, and then make the result; if you want to see the effect of another independent variable, you need to go back and do it again.
Imagine if there are 20 independent variables, you need to do the same work back and forth 20 times. This process is very cumbersome and time-consuming. JMP The software provides tools for one-time implementation of univariate analysis of all independent variables .
Take the data in Figure 1 as an example to explore the possible influencing factors of physical fitness scores.
Figure 1 Example data (partial display)
In JMP, click the "Analyze" menu → "Fit Y with X" (Figure 2), and in the pop-up dialog box, import the physical health score into "Y, response", and import all independent variables into "X, factor" ( image 3).
Figure 2 Schematic diagram of the operation of "Fit Y with X" The first step
Figure 3 The second step of the operation diagram of "Fit Y with X"
The default output is shown in Figure 4 (only 3 independent variables are shown, and the rest are omitted), and only the scatter plot of each independent variable and dependent variable is displayed on the page.
Figure 4 The output of "Fit Y with X"
If you want to View the test results of the univariate analysis of each independent variable , you need to perform some subsequent operations in JMP. It is important to note that the steps are different for continuous variables and categorical variables.
continuous variable
For continuous variables, take age as an example, click the red down arrow to the left of "Binary Fitting" in the results interface → select "Fit Line" (Figure 5).
Figure 5 Schematic diagram of the operation of univariate analysis of continuous variables
The output results are shown in Figure 6. It can be seen from the figure that there is a negative linear relationship between age and physical health score. From the statistical test results, the linear relationship between age and physical health score was statistically significant at the test level of 0.05 (P=0.0275).
Figure 6 Univariate analysis results of continuous variables
Categorical variables
For categorical variables, taking drinking as an example, click the red down arrow to the left of "...one-way analysis" → select "mean/ANOVA/pooled t" (Figure 7).
Figure 7 Schematic diagram of the operation of univariate analysis of categorical variables
The output results are shown in Figure 8. The results of the t-test and the one-way ANOVA are also given in the results, and the results of the two are consistent. The results showed that the relationship between alcohol consumption and physical health score was not statistically significant (P=0.8871).
Figure 8 Univariate Results for Categorical Variables
Seeing this, you may ask: In the case of many independent variables, each independent variable has to operate "fitted line" or "mean/variance analysis/pooled t", will the process be very cumbersome?
JMP There is a convenient way to solve this problem, which can display the statistical test results of all independent variables very quickly .
- for Continuous variables, first press ctrl key , then click the red down arrow to the left of "Binary Fit" for any independent variable → select " fitted line ", which can realize the output of univariate analysis results for all continuous variables.
- for Categorical variables ,same with Press first ctrl key , then click the red down arrow to the left of any categorical variable, choose " mean / Analysis of variance / combined t " , which enables the output of univariate analysis results for all categorical variables.
The above is the univariate analysis based on JMP to realize all independent variables at one time shared today. We will then describe how to perform a multivariate analysis of linear regression in JMP. Stay tuned!
>> Learn more about JMP Multivariate Analysis
Recommended articles for statistical analysis:
Original link: https://mp.weixin.qq.com/s/uHau9nfIKPAmk9YaPoAb5Q
