# PLS-SEM Analyses

Please run the SEM model described in the figure below using the enclosed dataset (npcdata.sav). This time, model it and run it in PLS. To do so, the first step will be to convert the SAV file (SPSS) into a text file. If the version of SmartPLS you have supports up to 100 cases, then pick the first 100 cases in the file.

Please note that in PLS, as opposed to AMOS-SEM you do not need to explicitly model the errors/residuals. This is done in the background.

In 1999, NPC Research Inc. conducted a school survey of over 2,000 students in Oregon. The survey included questions about drug and alcohol use, and also questions about the student's perceptions of his or her neighborhood. Responses to the following items were on a 4-point scale (0=NO! 1=no 2=yes 4=YES!).

VARIABLE ITEM

CRIME How much does "crime" and/or "drug selling" describe your neighborhood?

FIGHTS How much does "fights" describe your neighborhood?

ABANDON How much does "lots of empty or abandoned buildings" describe your neighborhood?

GRAFITTI How much does "lots of graffiti" describe your neighborhood?

GETOUT I'd like to get out of my neighborhood.

INANDOUT People move in and out of my neighborhood.

1. Construct and estimate the model in SmartPLS.

2. Run bootstrapping to extract t-values and estimate significance.

3. What are the path coefficients between latent variables? Are they significant?

4. What is the explained variance in the endogenous constructs? (those constructs which have at least one predictor)?

5. Draw a diagram of the model with the path coefficients and explained variances included.

6. Discuss the advantages of PLS vs. AMOS-SEM for this model (if any) and its disadvantages (if any). Which approach will you use for estimating this model?