1 User manual WEB GESCA . Version Heungsun Hwang1, Kwanghee Jung2, and Seungman Kim2. 1 McGill University 2 Texas Tech University Please cite this software as follows: Hwang, H., Jung, K., & Kim, S. (2019). WEB GESCA (Version ) [Software]. Available from 1|Page 1. How to use WEB GESCA . a. Sample data and model Bergami and Bagozzi's (2000) organizational identification data are used for illustrative purposes. The number of cases is equal to 305. The model specified for the data is displayed in Figure 1 (No residual terms are displayed in the figure). As shown in the figure, this model consists of four latent variables and 21 reflective indicators. Specifically, Organizational Prestige (Org_Pres) is measured by eight indicators (cei1. cei8), Organizational Identification (Org_Iden) by six indicators (ma1 ma6), Affective Commitment-Joy (AC_Joy) by four indicators (orgcmt1, 2, 3 and 7), and Affective Commitment Love (AC_Love) by three indicators (orgcmt5, 6, and 8).
2 Figure 1. The specified structural equation model for the example data b. Prepare a raw data file WEB GESCA is run on individual-level raw data. The raw data file is to be prepared as comma separated values file format (.csv), or ASCII file format (.txt or .dat). You can use comma, tab, semicolon, or pipe as separator in any of these cases. It is recommended that the first row contains the names of indicators. If your data file doesn't contain the names of indicators, you must indicate it (See ). The following figures show the dataset opened in Excel and Notepad for the present example: 2|Page Figure 2. A comma separated values file format opened in Excel Figure 3. A tab separated txt file opened in Notepad 3|Page c. Connect to WEB GESCA . i. Open your web browser, , Google Chrome, Microsoft Edge, Firefox, etc.
3 Figure 4. Web browser ii. Enter ' or ' in the address area. Figure 5. Address for WEB GESCA . 4|Page iii. Now you can see the front page of WEB GESCA . Figure 6. Front page of WEB GESCA . d. Data The page Data' is the first page when you visit WEB GESCA . The first step for running WEB GESCA is upload your data, which are already prepared as described in Section You can also download sample data by clicking the button. i. Upload data file - Click the Browse' button and select a data file of your choosing. Figure 7. Upload data file (step 1). - Select your data file and click the open button in the window. If the file is successfully uploaded, you will see the first six cases of the data. The Model menu will also become available in the left-hand sidebar. 5|Page Figure 8.
4 Upload data file (step 2). ii. Input specifications Once the data is uploaded, you can specify your input data on the right-hand side of the Data page. The default set up is data has header' and it is separated by comma.' If your data is not formatted as required, the data will not appear appropriately in the First 6 cases of the data' area. Note that if you uncheck the checkbox for header because your data doesn't contain the names of indicators, the program will assign arbitrary names to them. Figure 9. Input specification. 6|Page iii. Missing Data WEB GESCA currently provides users with three options for dealing with missing observations: (1) listwise deletion, (2) mean substitution, and (3) least-squares imputation (refer to Hwang & Takane, 2014, Chapter 3). Users can select one of the options in the box of Missing Data on the right side of the page.
5 Once users select an option for missing data, another box Missing value identifier will appear, where users input a numeric value that indicates missing observations in their data. The default value indicating missing observations is -9999. However, users can put any value of their choosing. Figure 10. Handling missing data iv. Analysis Type If you want to conduct a multi-group analysis, choose Multi-group analysis'. Once you check the check box, all variables' names of your data will appear. Please select a variable that you want to use as a grouping variable from those listed. Figure 11. Options for multi-group analysis 7|Page e. Model After uploading and specifying data, you need to specify your model. Click Model in the left main menu. There are two ways to build your model: using GESCA R syntax or Table.
6 If you are familiar with GESCA R syntax and want to use the syntax, select the tab Build a model by syntax' on the top. Otherwise, select the tab Build a model by table.'. Figure 12. Two methods for modeling i. Build a model by syntax This is the simplest way to run WEB GESCA . Simply enter your measurement model and structural model into the textbox of Input your model.' After you input the syntax, click the button. Then, WEB GESCA will run to analyze your model. Please refer to GESCA R package manual for the use of GESCA R syntax ( ). Figure 13. Build a model by syntax 8|Page ii. Build a model by table 1) Choose the number of latent variables Drag the slider to choose the number of latent variables in your model. If your model has second-order latent variables, choose the number of them as well.
7 After choosing the number of latent variables, click Confirm the number of latent variables'. Figure 14. Select number of latent variables 2) Input names of latent variables If you confirmed the number of latent variables, the box for inputting the name of each latent variable will appear. Input the names of your latent variables. If you don't label your latent variables, they will be named like LV_1, LV_2, . LV_n by default. Figure 15. Naming for latent variables If all the names of latent variables are entered correctly, then please click confirm the name(s)'. 9|Page 3) Specifying structural model Once you confirmed latent variables' names, a table for specifying your structural model will appear in the middle of the main contents area. You can specify a structural model by checking checkboxes.
8 The figure below shows that LV_1 affects LV_2. Figure 7. Specifying structural model If you want to fix a path coefficient to a constant, enter the constant value ranging from to to the blank textbox. Figure 8. Fixing path coefficient If you selected multi-group analysis' ( ), you can constrain all path coefficients to be equal across groups simultaneously by checking Constraining path coefficients across groups.' This checkbox only appears when you selected multi-group analysis. Figure 98. Constraining all path coefficients across groups 10 | P a g e After you completed structural model specification, click confirm the model.'. 4) Assigning indicators You can assign indicators to latent variables by using the checkboxes. Figure 10. Assigning indicators Figure 19 shows that the latent variable LV_1 is assigned the indicators cei1, cei2, and cei3, LV_2 is assigned cei4, cei5, and cei6, and LV_3 is assigned cei7, cei8 and ma1.
9 If you want to fix a loading to a constant in advance, put the constant value ranging from to in the blank textbox. Figure 20. Fixing value of loading for an indicator 11 | P a g e If you selected multi-group analysis ( ), you have two more options for constrained analyses. First, you can constrain the loadings of selected indicators to be identical by inserting a label ( , an alphabet letter or number) across groups. In the below example, three indicators were chosen, and labeled by different alphabet letters ( a , b , and c ). This indicates that each loading of the three indicators is to be held equal across groups. Note that any loadings with the same label will be constrained to be equal to each other. Figure 21. Constraining specific loadings across group Also, if you choose Constraining (fix) loadings across groups,' all loadings will be constrained to be equal across groups.
10 Figure 22. Constraining all loadings across group 12 | P a g e If formative indicators are assumed for a latent variable, you check the Formative relationship' box per latent variable located at the top of the table. If they are reflective indicators, just leave the checkboxes unchecked. Figure 23. Specifying formative relationship between the indicators and the latent variable The above figure shows that cei1, cei2, and cei3 are formative indicators for LV_1, whereas the remaining indicators for LV_2 and LV_3 are reflective. 5) Second-order latent variable modeling If you have a second-order latent variable in your model ( ), you need to specify the relationship between second-order and first-order latent variables. Another table will appear below the table for assigning indicators if your model has a second-order latent variable.