basic data analytic techniques - pt 2 · pdf file( ... open a new spss syntax window ......
TRANSCRIPT
BASIC DATA ANALYTIC TECHNIQUES
PART 2
Below are some instructions for how to:
1) Decide which data analytic technique(s) to use to answer your research questions and
test your hypotheses.
• Find the kind of research question you are asking below.
2) Conduct your analyses using SPSS, which is available to all students through the Virtual
Lab (vLab) and on computers in many St. Norbert College lab spaces; images of some
graphical user interfaces (GUIs) in SPSS are shown below.
** Note that you may not need to conduct analyses (e.g., if you are completing the General
Psychology Application Paper).
Begin by opening your data in SPSS. If there are any “filters” you need to turn on (e.g., if you
are only interested in some participants in your sample, such as females or people who have a
history of a particular experience).
• Note that significance values ≤ .05 are generally considered “statistically significant.”
How does ____________ influence the association between ___________ and ____________?
HIERARCHICAL MULTIPLE REGRESSION – INTERACTIVE EFFECTS (MODERATION)
** Moderation means that the association between your DV and an IV
is influenced by another possible IV. **
• Before completing your analysis, you need to standardize your IVs
• Click on Analyze
• Scroll to Descriptive Statistics, then choose Descriptives
• Move your IVs into the Variables box
• Check “Save standardized values as variables” and click OK
• Your standardized variables are now at the end of your data file
• Next, you need to compute the interaction:
o Click Transform
o Scroll to Compute Variable
o You will see this GUI:
• Create a name for your interaction variable in the Target Variable box
• Scroll to the bottom of your data list, and set them up to be multiplied (*) in the
Numeric Expression box
• Click OK
• Your interaction variable is now at the end of your data file
-----------------------------------------------------------------
• Now, complete the initial steps described above for Multiple Regression analyses
• This time, when you see this GUI:
• Be sure to enter the standardized versions of your IVs in the Independent box; do not
enter the interaction variable yet
• Click the Next button, then enter the interaction variable
• Remember to click the Statistics button and check R squared change to turn it on
• Click Continue, then click OK
• In the Model Summary table,
o The first R Square Change value (and its Sig F Change) tells you the magnitude of
the association between the DV and both/all of the IVs, combined
o The second R Square Change value (and its Sig F Change) tells you the
magnitude of the association between the DV and the interaction variable(s)
• In the Coefficients table,
o The first set of Standardized Coefficients (Beta) tells you the magnitude of the
associations between the DV and each of the IVs, separately
o The second set of Standardized Coefficients (Beta) tells you the magnitude of
the associations between the DV and interaction variable
• If your interaction is statistically significant, there are additional analyses you can
complete in order to interpret the interaction. These additional analyses (simple slopes
tests) are described in Part 4.
Does ____________ account for the association between ____________ and ____________?
MEDIATION
** The approach described below is based on the Preacher & Hayes (2008) method for
conducting mediational analyses. It will require you to obtain the INDIRECT macro for SPSS,
which is available on Hayes’ website:
(http://www.afhayes.com/spss-sas-and-mplus-macros-and-code.html) **
• First, open and run the INDIRECT macro for SPSS
• Next, open a new SPSS Syntax window – SPSS does not have GUIs for these analyses
• The example codes below examine the role of one Mediator (M) in the association
between an IV (X) on a DV (Y)
• The example codes below examine the same as above, but when taking into account a
relevant Covariate (C)
• Each path shown in the model below will be examined in this analysis
a b
c/c’
• The results will look different from other analyses; scroll down to the section that looks
like this:
Mediator
Indep. Variable Dep. Variable
** In this example, the association between posttraumatic stress and physically
aggressive acts is accounted for by trait shame
• First, check to see if there is an association between X and Y (c path)
• If the c path is statistically significant, next look at whether the strength of the
association is reduced by including the mediator (M) by checking the c’ path
o If the c’ path is not statistically significant, this is evidence of full mediation (i.e.,
the association between X and Y is accounted for by M)
• If the c’ path remains significant but the coefficient appears reduced, consider whether
you have evidence of partial mediation