Using the SmartPLS Software Joe F. Hair, Jr. Founder & Senior Scholar Double click anywhere on this tab to close or bring back the SmartPLS navigation windows. More SmartPLS Options . . . To close the navigation windows on the left side of the screen (for “Projects,” “Outline,” and “Indicators”), double click on the tab (shown above) that has the name of your structural model. Note, do NOT click on the “X” because your structural model will close instead. By closing the navigation windows, you will have a larger area on the screen to display your model. The extra space on the screen is particularly helpful when you have a complex model, when you are trying to get your model to display appropriately to save or copy the image, or when you are assessing the model visually or viewing the SmartPLS reports. To bring the navigation windows back, simply double click on the tab again, and the windows will return. More SmartPLS Options . . . To save the SmartPLS model as a “bitmap” (.bmp) image file, select “File” “Export to Image” from the menu bar (see next slide). There are several choices to obtain an image file for your SmartPLS structural model. The first option is to select “File” “Export to Image” from the menu bar, as shown above. This allows you to export a “bitmap” image file (.bmp) of your model exactly as it is shown on the monitor. For example, if you have calculated any of the SmartPLS tasks (i.e., PLS algorithm, FIMIX-PLS, Bootstrapping, or Blindfolding) and the model is labeled with the output of the algorithm, that information will show on the .bmp file. But if your model is not visually appealing on the screen, it will not be visually appealing in the .bmp file. A nice feature of the .bmp option is that your SmartPLS model will appear on a white background, which is generally a better option for printing the image, including it in a PowerPoint presentation, or using the image as part of a journal submission. Other options are dependent on your computer’s software or operating system. For example, with Microsoft Windows you can execute the “Print Screen” command by pressing [Ctrl] + [Print Screen] on the keyboard. The keyboard combination automatically copies the information displayed on your monitor to your clipboard, where it can then be pasted into Word, Excel, or PowerPoint. A second choice is Microsoft’s “Snipping Tool” program that enables you to select the desired area: “Free-form,” “Rectangular,” “Window” or “Full-Screen.” After your desired output area is selected, the image is automatically copied to your clipboard (the same as with the “Print Screen” command). Additionally, you have the ability to save the image file (as .png, .gif, .jpg, or .html) so that it may be used later. Finally, you can paste your image in the Paint option and select the part you want. This is an example of what you get when you use the SmartPLS Export to Image option. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 4 Click here, select PLS Algorithm to calculate model results. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 5 This dialog box appears when you select the PLS Algorithm. Missing values have not been configured so you will have to Cancel this option and go to the datafile to set up. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 6 Default Settings to run PLS Algorithm – Click Finish to run Trade-off in missing value treatment: Case wise replacement can greatly reduce the number of cases but sample mean imputation reduces variance in your data. Preferred approach to deal with missing data is combination of sub-group mean and nearest neighbor, or use EM imputation using SPSS. Always use path weighting scheme Double click on the datafile to get this screen. After double clicking on the datafile you get this screen. Check the box on the left to indicate missing data in your datafile. Then change the Missing Value in the window to -99, as shown below. Finally, check the X beside the Full Data tab at top. That will close and save your changes After missing values have been All rights reserved ©. Cannot be reproduced or distributed without express written permission from to -99.0 Prentice-Hall, McGraw-Hill, SmartPLS, and session reconfigured presenters. 8 When you select the PLS Algorithm option this revised dialog box appears. It now shows the newly configured missing value option of -99.0. All other options are correct so check the Finish tab to run the model. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 9 PLS Results for Simple Example Outer loadings, path coefficients, and R2 shown on model All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 10 PLS Results for Simple Example The structural model results enable us to determine, for example, that CUSA has the strongest effect on CUSL (0.504), followed by LIKE (0.342). COMP (0.009) has little effect on the dependent variable CUSL. The three exogenous constructs together explain 56.2% of the variance of the endogenous construct CUSL (R² = 0.562), as indicated by the value in the construct circle. COMP and LIKE also jointly explain 29.5% of the variance of CUSA. Click here to obtain reports that summarize model results. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Sage, Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 12 Checking the algorithm stop criterion This is an example of the reports that are available from SmartPLS. The type of information provided is shown in the menu on the left. For example, the Stop Criterion Changes is highlighted. Above the report shows the software took 4 iterations to obtain a solution. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 13 The above table of values is the default report for the Path Coefficients. To make this table easier to understand left click on the Toggle Zero Values button at the top left. The results are shown below. To determine the statistical significance of the path coefficients, we must run the bootstrapping option. To do so click on the .splsm file tab to return to the SEM model with the tool bar. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 14 Summary of PLS-SEM Findings 1.The direct path from COMP to CUSA is 0.162 and the direct path from COMP to CUSL is 0.009. 2.The direct path from LIKE to CUSA is 0.424 and the direct path from LIKE to CUSL is 0.342. 3.The direct path from CUSA to CUSL is 0.504. 4.Overall, the model predicts 29.5% of the variance in CUSA, and 56.2% of the variance in CUSL. To determine significance levels, you must run Bootstrapping option. Look for under the calculate option. Click here, select Bootstrapping option. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. Significance of PLS-SEM Parameters = Bootstrapping PLS-SEM does not assume the data is normally distributed, which implies that parametric significance tests used in regression analyses cannot be applied to test whether coefficients such as outer weights and loadings are significant. Instead, PLS-SEM relies on a nonparametric bootstrap procedure to test coefficients for their significance. In bootstrapping, a large number of subsamples (i.e., bootstrap samples) is drawn from the original sample – with replacement. Replacement means that each time an observation is drawn at random from the sampling population, it is returned to the sampling population before the next observation is drawn (i.e., the population from which the observations are drawn always contains all the same elements). Therefore, an observation for a certain subsample can be selected more than once, or may not be selected at all for another subsample. The number of bootstrap samples should be high but must be at least equal to the number of valid observations in the dataset. The recommended number of bootstrap samples is 5,000. When you get this dialog box make sure you have chosen No Sign Changes, the number of cases is your sample size (344), and the number of samples is 5,000. Then click the Finish tab to obtain the results. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 18 SmartPLS Bootstrapping If you have missing data do not use mean replacement because bootstrapping draws samples with replacement. Use Casewise Replacement. Use individual (sign) changes option • Make sure the number of cases are equal to the number of valid observations in your dataset. • Set cases = samples size (or higher) Caution!!! It is a common mistake to set samples equal to the overall number of observations. The t values can be compared with the critical values from the standard normal distribution to decide whether the coefficients are significantly different from zero. For example, the critical values for significance levels of 1% (a = 0.01) and 5% (a = 0.05) probability of error are 2.57 and 1.96, respectively All rights reserved ©. Cannot be reproduced or distributed without express written permission from(two-tailed test) . 20 One-tailed test for 5% (a = 0.05) level is .98. Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. By clicking on Report – Default Report you get a detailed overview of the bootstrapping results. The original estimate of the outer weights is shown in the second column = Original Sample (0). If this number is divided by the Standard Deviation (STDEV) you get the t value. For example, divide 0.5361 (0) by 0.0445 (STDEV) and you get 12.047 = the t statistic – shown below. The t statistics in the table on the right indicate that all measurement model loadings are statistically significant (> 0.05). All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 21 The t statistics in the table below indicate that four of the five structural path coefficients are statistically significant (> 0.05). The only non-significant path is COMP – CUSL (t value = 0.1705). All rights reserved ©. Cannot be reproduced or distributed without express written permission from Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 22 Brief Instructions: Using SmartPLS 1. Load SmartPLS software – click on 2. Create your new project – assign project name and data. 3. Double-click to get Menu Bar. 4. Draw model – see options below: • Insertion mode = • Selection mode = • Connection mode = 5. Save model. 6. Click on calculate icon and select PLS algorithm on the Pull-Down menu. Now accept the default options (or insert your own) by clicking Finish. All rights reserved ©. Cannot be reproduced or distributed without express written permission from Sage, Prentice-Hall, McGraw-Hill, SmartPLS, and session presenters. 24