Audit trail navigation
In addition to the Model Summaries worksheet, RegressIt provides several other tools for viewing and navigating the audit trail of analyses in the file. When a sequence of regression models is fitted within RegressIt by successively adding or removing variables, they can be navigated by parent-child relationships via the Relatives button on the ribbon. This allows the user to jump to the parent of the currently visible model or to any of its descendants. An instructor can use this tool to trace the paths that students followed in exercises that may have required a lot of searching in the model space. It's also good for re-navigating your own work.
In addition to the Model Summaries worksheet, RegressIt provides several other tools for viewing and navigating the audit trail of analyses in the file. When a sequence of regression models is fitted within RegressIt by successively adding or removing variables, they can be navigated by parent-child relationships via the Relatives button on the ribbon. This allows the user to jump to the parent of the currently visible model or to any of its descendants. An instructor can use this tool to trace the paths that students followed in exercises that may have required a lot of searching in the model space. It's also good for re-navigating your own work.
The History procedure provides a scrollable pop-up list
of all worksheets currently in the file, with some summary information
about what each one contains. This is not only a display tool but also
a navigation tool. The user can jump directly to any worksheet in the
file by clicking on its entry in the list and hitting Move-to-Sheet.
There are also a couple of export options included in this procedure. The Export-List tool will export a table to the clipboard that contains all the information in the history window and more, from which it can be pasted to an Excel worksheet. (File Explorer must be closed when doing this.) The two screen shots below show some (not all) of the fields. On the spreadsheet these are side by side in one large table.
The table includes complete details of every regression model, such as its key parameters and statistics, the names of the computer and file in which it was created, the run time, the name of the computer on which the table was downloaded, and the model equation (list of variables). If this worksheet is located in the original analysis file or in another file with the same defined variables, the cell with the list of variables can be used to re-launch the model: just select the cell and hit the Linear Regression button on the ribbon. This will jump to the model worksheet if it exists in the file, otherwise it will pre-select the listed variables for a new model.
The Export-Folder tool (available on PC's only) can be used to draw the same information from all files in a selected folder, such as a folder into which all submitted files for a homework assignment have been downloaded. (Teaching platforms such as Sakai and Canvas offer this download option for grading purposes.) The tool exports a single large table to the clipboard, which contains one row for every worksheet in every file, and that row contains all of the same fields as in the Export-List table. This table is extremely useful for grading the work of an entire class. When using this tool, all files in the target folder should be closed, File Explorer should be closed, and a file in a different folder with a blank worksheet for pasting the table should be open.
Here's an example of a portion of the table produced from a folder containing three user files. Each one contains one or more descriptive analyses and a sequence of regression models:
The Export-Folder tool (available on PC's only) can be used to draw the same information from all files in a selected folder, such as a folder into which all submitted files for a homework assignment have been downloaded. (Teaching platforms such as Sakai and Canvas offer this download option for grading purposes.) The tool exports a single large table to the clipboard, which contains one row for every worksheet in every file, and that row contains all of the same fields as in the Export-List table. This table is extremely useful for grading the work of an entire class. When using this tool, all files in the target folder should be closed, File Explorer should be closed, and a file in a different folder with a blank worksheet for pasting the table should be open.
Here's an example of a portion of the table produced from a folder containing three user files. Each one contains one or more descriptive analyses and a sequence of regression models:
For purposes of grading the selection of a regression model, students can be told to use a standard name such as _FINAL_MODEL for their final model, and the table can then be sorted on this field to group all of the final student models together. If an underscore is the first character in the name, it will sort first in the list, as shown here (other columns are hidden):
In addition to statistics and
variable selections for each model, the table includes the name of the computer
on which the model was run, the date and time at which it was run, and the name
of the analysis file at the time the model was run. Students should also be told to personalize
the names of their analysis files and data worksheets before beginning their
work, as in the first column in the table above. With
this amount of detail, it is virtually impossible for one student to submit
another's work, and the instructor can also get an overview of the modeling
process followed by each student, for example, the sequence of models and the
time spent on the whole analysis. If
this table is copied to a worksheet in a file that contains the data, the text
in the model equation cell can be used to launch any of the models.
With this tool it is feasible to take a helicopter tour of regression analyses submitted by very large numbers of students, and it may be interesting to sort the table on other fields (e.g., model equation, number of variables, computer name, date/time, etc.). On assignments that involve fitting multiple regression models starting from a large candidate set of variables, it can be quite interesting to see the distribution of answers.
One more feature of the Export-List and Export-Folder tools is the ability for the user to include comments in the analysis file on one or more worksheets. If a worksheet has the word "Comments" in cell A1, the program will look for text strings in cells A2, A3, A4, etc.. These text string strings will be entered in cells in columns D, E, F, etc., in the row in the table for that worksheet (i.e. a column of text strings on a comments worksheet is mapped into a row of text strings in the exported table of workbook contents). This feature could be used for annotating the models in the file or for entering answers to homework assignment questions: a comments worksheet could be included in the original data file with questions in even-numbered rows and with corresponding answers to be entered in the adjacent odd-numbered rows. While model-building is still going on, comments worksheets ought to be placed to the right of the Model Summaries worksheet in order to avoid interfering with the placement of new model worksheets in sequence.
With this tool it is feasible to take a helicopter tour of regression analyses submitted by very large numbers of students, and it may be interesting to sort the table on other fields (e.g., model equation, number of variables, computer name, date/time, etc.). On assignments that involve fitting multiple regression models starting from a large candidate set of variables, it can be quite interesting to see the distribution of answers.
One more feature of the Export-List and Export-Folder tools is the ability for the user to include comments in the analysis file on one or more worksheets. If a worksheet has the word "Comments" in cell A1, the program will look for text strings in cells A2, A3, A4, etc.. These text string strings will be entered in cells in columns D, E, F, etc., in the row in the table for that worksheet (i.e. a column of text strings on a comments worksheet is mapped into a row of text strings in the exported table of workbook contents). This feature could be used for annotating the models in the file or for entering answers to homework assignment questions: a comments worksheet could be included in the original data file with questions in even-numbered rows and with corresponding answers to be entered in the adjacent odd-numbered rows. While model-building is still going on, comments worksheets ought to be placed to the right of the Model Summaries worksheet in order to avoid interfering with the placement of new model worksheets in sequence.