Entering Data with Imported Files
There are four methods for importing data into ConstructMap.
- Use the Import Wizard with an Excel file.
- Import a single tab-delimited file.
- Import a single Conquest-formatted file.
- Use the standard method with two tab-delimited files.
The first three methods assume a unidimensional model with one instrument. However, once imported with any of these methods, you can modify your model in Construct Map. For example, to facilitate data entry, you may place all of your data in a single Excel spreadsheet for importing. After the data are imported, you can divide the items into different instruments or assign items to multiple constructs using the Constuct Map features.
Import Wizard
The Import Wizard imports one Excel-formatted file that contains response data.
Requirements for acceptable Excel formats
- ConstructMap only supports Microsoft Office Excel; it does not support iWork Office software.
- For Windows: xls version 97-2003
- For Macintosh: Up to 2001 and version 2004 is experimental to be used at your own risk
- For both: make sure that the excel file is not password protected
Each row contains data for one case; columns contain a respondent ID, item data, or a case-level variable (i.e., to be used for grouping data). The columns can be in any order. ConstructMap assumes the items belong to a single instrument and measure a single construct. You can edit the model in ConstructMap after the data have been imported.
Example of Excel data with heading information in Row 1:
In the example above, column A contains a respondent ID, column B contains a case-level variable, and columns C through N contain item data.
Example of Excel data without a heading row:
In the example above, column A contains a respondent ID, columns B through I contain item data and column J contains a case-level variable.
Converting raw response codes to IRT categories
When the item data are imported, ConstructMap attempts to map reponse values to IRT categories for you. Within an item, it will map consecutive integral values to integers 0, 1, 2, ... Non-consecutive values will be mapped as missing values. ConstructMap reads all of the data for each item and then interprets the range of valid score values for that item. Each item is treated independently of other items. ConstructMap only maps contiguous responses to IRT categories. It maps the lowest-value contiguous response to 0 and then assigns subsequent integral values to the remaining responses, in numerical order.
ConstructMap does not interpret '.' as a missing value indicator. Instead, leave cells blank to indicate missing values.
In the first example above, there are no numeric values in the item columns, so ConstructMap will not attempt to map the response codes to IRT categories. After importing, the user should modify the Answer Key to indicate the correct response for each item.
In the second example, if we assume the data we see are the only response values present in the data, ConstructMap will translate the response codes 1, 2 and 3 for the first item to IRT categories 0, 1, 2. For the second item, the response codes 2, 3 and 4 will be translated to 0, 1 and 2.
Note that ConstructMap retains the raw responses that were originally imported. As a user-defined option, these values can be displayed on most of the item reports such as WrightMaps.
How to use this feature
To import data:
- Select File - Import Wizard.
- Use the Browse button to select the Excel file to import, then click on the Open button.
- If your spreadsheet includes column headings in Row 1, check the checkbox at the bottom of the import window.
- Click the Next> button to continue.
- Select appropriate radio buttons (Respondent ID, Case-level variable, Item) to indicate the contents of each column.
- After defining the last column, the Next> button will be activated. Click it to continue with the import.
- Read the messages on the Wizard Finished window, then press the Finished button.
- If it is not already selected, select the Answer Key tab at the bottom of the data table window.
- Review the IRT categories the wizard automatically selected for your items and make any necessary changes.
- Select the Students tab at the bottom of the data table window to see your imported data. Note that you will see the original raw codes that were imported. To see the IRT categories, select View - Show Recoded Values from the menu at the top of the window.
- Select File - Save to save your data.
That completes the import of your Excel-formatted data.
Tab Delimited Method
The Tab Delimited Method imports a single file that is formatted as tab delimited. The first column must contain unique respondent IDs. All the other columns must contain item responses. Missing values should be represented as blank cells; the import feature does not understand '.' as a missing data indicator.
There are no header rows in the tab delimited file for item names etc. Instead the item names are automatically assigned as Item 1, Item 2, ... etc.
Tip: Create this file in Excel then save it as tab-delimted format.
An example of a tab-delimited file created in Excel:
How to use this feature
- Select File – Open - Tab Delimited. Answer Yes when prompted about deleting the current project.
- Browse to the tab delimited file you created and click Open.
- When the importing process finishes, your data will be displayed in the ConstructMap response data window.
- Select File – Save to save your data.
Converting Raw Response Codes to IRT Categories
ConstructMap imports all of the data for each item and then interprets the range of valid score values. ConstructMap only recognizes contiguous integral responses for each item. It then maps the responses to integral scores starting at 0.
In the example abvoe, if we assume the data we see are the only response values present in the data, ConstructMap will translate the response codes 1, 2, 3 and 4 for the first item to IRT categories 0, 1, 2 and 3. For the second item, the response codes 2, 3, 4 and 5 will be translated to 0, 1, 2 and 3.
If an item has two sets of continuous data, for example 1 through 3 and 7 through 9, only the first set are considered score values while the second set are considered missing data.
Note that ConstructMap retains the raw responses that were originally imported. As a user-defined option, these values can be displayed on most of the item reports such as WrightMaps.
ASCII (Conquest) Method
This method uses only one file which is formatted very similar to a Conquest data file.
Response data for this import method must be consistent with the following assumptions that are applied by ConstructMap:
- Items are from a single instrument;
- The items are indicators of a single latent trait;
- The responses are integers from 0 through 9; and
- All valid responses are represented in the data (i.e., no categories are omitted).
- Create the response data in a text format (not tab-delimited). We recommend that you use a text editor, such as Notepad or SimpleText, to create the file (see
C:\Program Files\ConstructMap43\projects\simple_import\simple_import.txtfor an example). - If another project is active, export the data and then select File – New Project. Answer Yes when prompted about deleting the current project.
- Select File – Simple Import.
- Browse to the file you created and click Open.
- When the importing process finishes, your data will be displayed in the ConstructMap response data window. Select File – Save.
The first row in the file defines the format of the data by specifying the column in which response data begins and the number of items. The response data is then entered as single-digit numerical values, with the data for each case entered on a single row. In the example shown below, the data located in columns 1-4 is ignored and the responses are imported from columns 5-23.
ConstructMap imports all of the data for each item and then interprets the range of valid score values. ConstructMap only recognizes contiguous responses for each item. It then maps the lowest-numbered contiguous responses to scores starting at 0. In Figure 83 the first item shows responses of 7, 2, and 1. ConstructMap translates the 1 to a 0, the 2 to a 1, and the 7 is considered missing data because it is not contiguous with the 1 and 2. The third item shows responses of 1, 2, and 3 and ConstructMap translates these to scores of 0, 1 and 2. The fourth item shows responses of 2, 3, and 4, which are also translated to scores of 0, 1 and 2. If an item has two sets of continuous data, for example 1 through 3 and 7 through 9, only the first set are considered score values while the second set are considered missing data.
Standard Method
The Standard Method of importing involves the creation of two files, the model specification file and the response data file. The model specification file must be imported first, as it describes the variables and items. Currently, the files must be in tab-delimited format. Typically, users use a spreadsheet application like Excel to generate the files.
Creating the Model Specification File
(Variables, Items, Population data etc.)
Create the model specification file – the variable, item, and population settings – in a tab delimited format. We recommend that you use a spreadsheet such as Excel for creating the file (see C:\Program Files\ConstructMapxx\projects\Ex1_dich\items.txt for an example).
Each model specification component is entered on a single row, with each type of specification having its own format:
System/Model information (one row only, optional)
| Column A | s (single lower case character) |
| Column B | Minimum scale value to display on graphs – correlates to min. logit. |
| Column C | Maximum scale value to display on graphs – correlates to max. logit. |
| Column D | Minimum logit value for estimation computations |
| Column E | Maximum logit value for estimation computations |
| Column F | Number of quadrature points for estimation (not used for Monte Carlo integration) |
| Column G | Model type, r = rating scale, d = dichotomous, p = partial credit (use p for mixed dichotomous and partial credit items). ConstructMap cannot mix dichotomous and rating scale items. |
Variable entries (i.e., dimensions, constructs, progress variables, etc.)
(One row per variable)
| Column A | v (single lower case character) |
| Column B | A short abbreviation for the variable. |
| Column C | A full title for the variable. |
| Column D | The maximum score for the variable (e.g., 4 indicates that scores 0-4 are valid). |
| Column E | The mean of the population distribution on the variable; optional. |
| Column F | The variance of the population distribution on the variable; optional. |
| Column G | The tau parameters for the variable (in a rating scale model), separated by commas (the number of parameters must equal the maximum score); optional. NOTE: This field should be in text format. |
| Column K (NOTE columns H, I, J are left blank) | Covariances; optional. If multiple covariances are required, they should be comma-separated. This field should be in text format. |
Item Set entries (e.g., groups of items such as an activity, test, assignment, etc.)
(one row per Item Set)
| For top-tier item sets: | |
| Column A | is (lower case characters) |
| Column B | name of this Item Set |
| For lower-tier item sets: | |
| Column A | is (lower case characters) |
| Column B | this Item Set parent’s name |
| Column C | name of this Item Set |
Item entries
(one row per item)
| Column A | i (single lower case character) |
| Column B | The Item Set this item belongs to. |
| Column C | The short variable abbreviation to associate this item with a variable. |
| Column D | A short abbreviation to identify the item (this name will appear in the column heading for the Item Set in the data table; it only needs to be unique within the Item Set, however several maps and reports are easier to interpret when every item has a unique name). A length of 4-6 characters is recommended to keep the maps easy to read. |
| Column E | A full title for the item. |
| Column F | The (average) item difficulty. Optional. |
| Column G | Item comments. Optional. |
| Column H | Item administration date in mm/dd/yyyy format. A blank entry indicates that the item is inactive and it will not be used in computing estimates. This field should be in text format. |
| Column I | Tau parameters for the item, separated by commas (tau(j) values rather than delta(i,j) values). The number of steps must equal the maximum score value for the item. This field should be in text format. Optional. |
| Column J | Maximum score for the item. If this field is left empty, the maximum score for the variable associated with the item will be used. For calibration purposes, this should be the actual maximum score found in the data rather than the theoretical maximum score. |
-
Save the file in tab-delimited text format.
Template for the Model Specification File
The figure below is an example of a minimally defined model specification file. Only the variable and item definitions appear. ConstructMap will assume a partial credit model with one assessment stored in the default item set, named "base." The variable mean will be 0.0 and the variance will be 1.0. The item parameters will be set to 0.
Figure 3 shows a more fully specified model. This specification defines a unidimensional dichotomous model using quadrature integration with 10 quadrature points to produce estimates. The reports will transform -3.0 to 3.0 logits to a scale of 1000 to 2000. The variable, “abil1” has a maximum score of 1, a mean of 0 and a variance of 1.0. Two item sets are defined, the top tier named “is1” and the second tier named “act1.” Two items are defined; both are “active” and will be included in all estimations.
The items.txt files for Example 1 and Example 2 projects have simple model specifications, while the model specifications for Example 3, Example 4, and the SEPUP Demo project are more detailed. Examples 3 and 4 include a system specification row and the SEPUP Demo project includes a system specification row and Criterion Zone definitions.
Creating the Response Data File
Create the response data in a tab-delimited format. Again, we recommend that you use a spreadsheet for creating the file (see C:\Program Files\ConstructMapxx\projects\ for several examples of students.txt files).
The response data for each case is entered on a single row, with the item names used as column headings. The file should use the following format:
Required heading section (see Sample 1 below):
| Row 1 | The date the items were scored. The current date is recommended. |
| Row 2 | Titles for the Item Sets. |
| Row 3 | The first column should be entitled "Name". |
| The next set of columns should indicate any demographic field names. | |
| The final set of columns should indicate an item short name. |
Optional demographics section:
Start the section with <DEMOGRAPHICS> in the first column of a new row (including the < > delimiters). This will be the only entry on that row. Subsequent rows will contain demographic field specifications, one demographic field per row:
| Column A | Demographic name |
| Column B | Default value for this demographic |
| Column C | true for a discrete field or false for continuous |
| Column D | true if users can add new categories or false if they cannot. An example of a demographic field that might need additional categories added later is Primary Language, where English and Spanish might be defined initially, but you may want to add other languages later. |
| Columns E, F, G, etc. | Categories this demographic has. Leave blank for continuous demographic fields or those that have no predefined categories. Separate predefined categories with commas, e.g., English, Spanish (for a demographic field for Primary Language). |
End the section with </DEMOGRAPHICS> in the first column of a new row (including the < > delimiters).
Optional answer key section (see Sample 2 below)
Start the section with <ANSWERKEY> in the first column of a new row (include the < >). This will be the only entry on that row. Subsequent rows will contain the answer key specification. Create a separate row for each valid response value.
| Column A | A valid response |
| Columns B, C, D, etc. | The recoded value for the particular item. For example, an "A" in column 1 might be scored as a 0 for item 1 and as a 1 for item 2. You need a separate row for each value you want to provide an answer key for . Use . to denote a missing score or a missing recoding for a particular item. |
End the section with </ANSWERKEY> in the first column of a new row (include the < >).
- Note: When an answer key is provided, unrecognized responses in the data file will be treated as missing data.
Required response data section (one row per student):
| Column A | Case name or ID (Required. ConstructMap does not automatically assign IDs). |
| Columns B, C, D, etc. | Demographic and response data associated with the column headings. |
Save the file in tab-delimited text format.
Figure 1 is an example of a response data file that might accompany the minimal model specification file for dichotomous items. New as of version 4.3.12:
"." in an answer key represents a missing response, that is a valid input by the user. However, "N/A" in an answer key
means an invalid entry. An invalid entry means a particular value can not be put as a valid response to an item for a student.
The Item Set is “base” with item short names of “i1” and “i2.”
Figure 2 is an example that includes an answer key section. In this example, a score of A+ on item i1 will be translated to a score of 3, and a score of A will be translated to a missing score. Scores of A or OK on item i2 will be translated to a score of 2.
Figure 3 is an example that includes a demographic field section. Two demographic fields are defined: Class and Gender. In this example, Class represents the section the student is enrolled in. Gender is self-explanatory. Note on row 3 that Name does not have a default value, and it is identified as a discrete (categorical) field in which users can add new values. In row 4, Class is defined with no default value and is also identified as a discrete field in which users can add new values. Gender is defined in row 5, with a default value of “Male” and is also identified as a discrete field, but users cannot add values other than “Male” and “Female.”
The students.txt files for Example 1, Example 3 and the SEPUP Demo projects have simple response data. The response data for Example 2 includes demographic fields and the data for Example 4 includes an answer key.
Importing the Data
Once the model specification and response data files have been created as tab-delimited files, they are ready to be imported. The model specification must be imported first.
- Select File - Open - Standard (2 files) from the menu.
- Browse to the tab-delimited model specification file you created and click Open. Note: The system will report any difficulties encountered during the import process by displaying them on the screen and also writing them to a file named
filename_Errors.txtwhere 'filename' is the name of the file you tried to open. This error file will be located in the logs folder. - Browse to the tab-delimited response data file you created in the previous section and click Open. If you do not have response data available and would only like to import the model specification, click Cancel. Note: The system will report any difficulties it encountered during the import process by displaying them on the screen and also writing them to a file named
filename_Errors.txtwhere 'filename' is the name of the file you tried to open. This error file will be located in the logs folder. - When the importing process finishes, your respondent data will be displayed in the Students tab and the model specifications will be displayed in the Items tab.
- Select File – Save to save your project.
- If you did not enter item difficulties in the model specification file, they must be computed before any reports or maps can be generated.
