Dale Story

[email protected]

SPSS Help

Sorry, no pretty pictures here. You can find those in the Help Command of SPSS, books, and
other internet sites. Included here are the basic steps in a very abbreviated style.

• SPSS Available on Campus

o OIT maintains a list of sites on campus where SPSS is available:
http://www.uta.edu/oit/cs/computerlabs/software/index.php.

• Help Command in SPSS

o I prefer the Contents option (under Topics). There is also a Search option (also
under Topics). And they provide a Tutorial.

• SPSS Book
o If you purchased the Student Version of SPSS, the book that came with the
software is an excellent resource.

• SPSS Help on the Internet

o Use any search engine to find web site (many at Universities) that offer SPSS
Help.

• Open Excel File in SPSS

o File—Open—Data. Change Files/Type to “Excel.” Find and select the Excel
file. If the first row includes variable names, select “read variable names from the
first row…” If the first row does not include variable names, de-select “read
variable names from the first row…” Finally, save in SPSS as an SPSS file
{SPSS (*.sav)}.

• Dichotomize a continuous variable

o You will be dividing a continuous variable (example: GDP in million $) into just
two categories: low (units below the median) and high (units above the median).
o Find the median value in SPSS. Analyze, Descriptive Statistics, Frequencies—
Turn off “Display frequency tables”--Select the variable—Statistics, Median,
Continue, OK.
o Recode the variable in SPSS. Transform, Recode, Into Different Variables.
Select variable, Output Variable Name (“New Name”), Label (“New Variable
Label”), Old and New Values, Range—Lowest through put in median value,
Value = 1, Add; AND Range—put in median value + increment through Highest,
Value – 2, Add. Continue. Change. OK.

• Recode a variable
o Example: You want to record a discrete variable with 7 categories into just 2
categories with equal units in each category.
 o Generate the frequencies for the variable: Analyze, Descriptive Statistics, Select
Variable, OK.
o Determine which categories need to be collapsed. Example: say category I had 2
values, category II had 3 values, category III had 2 values, category IV had 4
values, and category V had 3 values. You would want to collapse categories I, II,
and III (to produce a single “low” category with 7 values) and categories IV and
V (to produce a single “high” category also with 7 values).
o Transform, Recode, Into Different Variable. Select this variable, Output Variable
Name (“New Name”), Label (“New Variable Label”), Old and New Values,
Range—1 through 3, Value = 1, Add; AND Range 4 through 5, Value = 2, Add.
Continue. Change. OK.

• Missing Value
o A missing value is a value you want to “ignore” in your analysis. Example: no
response, no data, refuse to answer, do not know, etc.
o Variable View, Column “Missing” and Row “this variable.” Select the small,
shaded box in this cell. Discrete Missing Values—put each missing value in one
of the three boxes. OK.

• Variable Label
o Variable Labels allow you to add a lengthier variable name to an abbreviated
variable name. Example: variable name is MEPCGNP1989 and you add a
variable label that is “Military Expenditures as a Percent of GNP in 1989.:
o Variable View, Column “Label” and Row “this variable.” Simply type the
variable label in this cell.

• Value Label
o Value Labels provide the English-language labels for the numeric values. For
example, the variable Gender will have values of 1 and 2. Value Labels will
provide the information about which value is male and which is female.
o Variable View, Column “Values” and Row “this variable.” Select the small,
shaded box in this cell. Value = whichever value and Value Label = the value
label (example, Value = 1 and Value Label = male). Add. Than repeat for all
Value Labels. OK.

• Frequencies
o Analyze, Descriptive Statistics, Frequencies
o Select variable and move it to the “variable” box using the right arrow.
o For statistics (mean, median, mode), select statistics.

• Cross-Tabs and Chi-Square

o Chi-Square is a test of significance for cross-tabs. Use smallest value under
“Asymp. Sig. (2-sided).” The criteria is 0.05 (5 percent probability that the null
hypothesis is true.
o Analyze, Descriptive Statistics, Crosstabs.
o Select your independent variable and move it to the “column” box.
 o Select your dependent variable and move it to the “row” box.
o Statistics—select Chi-Square—Continue.
o Cells—turn-off “Observed” and Select “Column Percentages”—Continue.
o OK.

• Lambda
o Lambda is a measure of association for nominal level data (only categories). The
range is 0 to 1. 0 would be a perfect “non-relationship” while 1 would be a
perfect relationship. It must be used in comparison.
o Analyze—Descriptive Statistics—Crosstabs.
o Move the X variable to the “column” box and the Y variable to the “row” box.
o Statistics—Lambda (under Nominal).
o Continue.
o OK.
o Output—Use Lambda, Symmetric. The “Value” column shows the value for
Lambda.

• Gamma
o Gamma is a measure of association for ordinal level data (rank-ordered; > or <).
The range is -1 to +1. The sign only indicates inverse versus directly proportional
relationships. 0 would be a perfect “non-relationship” while an absolute value of
1 would be a perfect relationship. It must be used in comparision.
o Very simply, identical to Lambda above; except after Statistics, select Gamma
(under Ordinal).

• Correlation
o The correlation coefficient is a measure of association for interval level data
(know the intervals; + or -). The sign only indicates inverse versus directly
proportional relationships. 0 would be a perfect “non-relationship” while an
absolute value of 1 would be a perfect relationship. It must be used in
comparision.
o Analyze—Correlate—Bivariate.
o Select your two variables and move them to the “Variables” box.
o Leave all default selections “as-is” (Pearson, Two-tailed, and Flag significant
correlations should be “turned-on”).
o OK
o Your output will be a 2x2 table, but only one cell is relevant. Use either cell that
shows the relation between your two variables (there are two such identical cells).
The first value in that cell is the correlation coefficient.

• Partial Correlation
o Essentially a “multi-variate” correlation coefficient—or a correlation between two
variables holding a third variable constant.
o Analyze—Correlate—Partial.
o Two “original” variables in “Variables” box.
o Third variable in “Controlling For” box.
• Regression
o Produces the linear equation: Y = a + bX; or Y = mX + b
o Analyze—Regression—Linear.
o Put Y variable in the “Dependent” box.
o Put X variable in the “Independent” box.
o OK.
o Output contains 4 tables.
o The 4th (last) table contains the coefficients for the regression equation (Y=a+bX).
o The coefficients are in the column “Unstandardized Coefficients—B.”
o The (Constant) row contains “a” and the independent variable row contains “b.”

• Multiple Regression
o Essentially a multi-variable regression analysis. One dependent variable with two
(or more) independent variables.
o Analyze—Regression—Linear.
o Put Y variable in the “Dependent” box.
o Put two X variables in the “Independent(s) box.