This homework calls on you to create a SmartPLS model predicting confidence in institutions from gender, education, and media use. The data are from the survey_sample.sav example data files provided with the installation of SPSS. Note no attempt has been made to filter or adjust data to assure compliance with assumptions of PLS - output is purely for instructional purposes, not purposes of substantive research.
The model is to look approximately like this:
Download the comma-delimited version of the data file, from survey_sample.csv, placing it in c:\temp or some other local directory on your computer.
Launch and run SmartPLS. When it comes up it will ask you for a workspace. Give C:\temp or some other directory location on your computer.
Select File, New, Create new project. Give the project a name like "ch14". Check the input indicators box.
When asked to "Please enter a path" browse to the data file you downloaded in step 1. Click Finish.
In the Projects window, click on survey_sample.csv. On the right, click on "Comma" where it asks to choose the delimiter. Click the Validate button to make sure there are no illegal cell entries such as blanks. Check that data do contain missing values and enter -99 as the missing values code.
In the Projects window, double-click on the .splsm file. The path diagramming window will replace the data window on the right.
Click on the "Switch to insertion mode" button and draw the first latent. Indicators will appear in the "Indicators" window. Right click to rename the latent as "Education".
Drag and drop the indicators educ and degree onto the Education latent.
Repeat to create the Gender latent, which has only sex as an indicator.
Repeat to create the latent Media Use, with indicators news, news2, news4, and news5 (these are news from different sources).
Repeat to create the latent Confidence, with these indicators: coneduc, confinan, conmedia, conpress, and contv (these are confidence in different types of institutions).
Click on the Connect tool. Draw the arrows connecting the latents as in the figure above.
Select Calculate from the menu, then select PLS Algorithm. Make sure the Data Metric is set to Mean 0, Var 1. Click Finish.
From the menu select Report, HTML Report (print). Warning! The full output is about 1,000 pages, mostly taken up by listings of the thousands of cases. Do not try to print directly. Instead bring up Word or your word processor, then cut and paste all output elements except the case listings (except the first several cases). Your final output in Word should be about 10 pages.
