If anyone knows about the estimated values after fitting the model and the contents of the effect tests, I would appreciate it if you could let me know.
I started using JMP to use ANVOVA to compare the quantitative values of healthy and diseased subjects for each atlas using a certain quantitative index.
At that time, the value of one atlas for one subject could not be measured and was treated as a missing value.
However, when I wrote out the estimated equation after fitting the model, the missing values were reflected as estimated values. I would like to exclude the missing values from the test when I test the effect.
Do you treat them as missing values by performing some processing? Or are you just calculating an estimate and treating them as missing values?
If anyone knows, please let me know.
Hi @pe1 ,
The model that is made in the column is always going to calculate for every row, agnostic of the presence of any other missing values in other columns. That being said, you can always hide and exclude the rows where those missing values exist. Just click the empty cell in the original column, right click and click "select matching cells", then go to Rows>Hide and Exclude.
Thanks!
Ben
I'm not familiar with medical analysis, so I don't understand your question, but is ANVOVA "ANCOVA" (analysis of covariance)? Specifically, what model are you using, and which of the target variables (response variable, Y variable, dependent variable) and explanatory variables (predictor variable, X variable, factor) have missing values? Does this mean you're performing an ANCOVA with the "parametric value" as the response variable, including "atlas," "healthy group/disease group," and other covariates?
■When fitting a linear regression model such as an analysis of covariance model (ANCOVA model) using "Fit Model," data rows with missing values in the response variable are not used in calculating parameter estimates (regression coefficients, weights, parameter estimates). However, even if the response variable has missing values, you can still calculate predicted values for the response variable.
■When fitting a linear regression model such as an analysis of covariance model (ANCOVA model) using "Fit Model," the behavior of data rows with missing values in the explanatory variables will differ depending on the "Treat missing values as categories" option.
- If the "Treat Missing Values as Categorical" option is off, any data row with even one missing value in any of the explanatory variables is excluded from the calculation. In this case, the row with the missing value is not used in computing parameter estimates, and the predicted value for the response variable is also missing.
- If the "Treat missing values as categorical" option is enabled, missing values in the explanatory variables are treated as a single category and used in the analysis. In this case, rows with missing values are used to compute parameter estimates and predictions for the response variable.
The Treat Missing Values as Categorical option can be toggled on or off using the options available by clicking the Model Specification red triangle in the upper left corner of the Fit Model launch dialog.
