(for more on SPSS)

There are three segments of this assignment. Turn them in early to get input as to how they can be improved to insure an A.

PART ONE: Select eight variables from the list of variables in any database available to SPSS. Treat three of these as dependent, i.e. variables that will be explained by the five remaining variables which will be treated as independent. Be sure to include nominal, ordinal, and interval levels of measurement. Present the frequency distribution for each variable. This is done by choosing Statistics, then summarize, then frequencies in the SPSS data window. Interpret the results. If the interval variable has numerous categories (such as age or education), collapse the data by using as appropriate spread (recode into a new variable using ranges ). See below for detailed information on how to do this.

PART TWO: State 5 hypotheses (statements denoting an expected correlation between variables). Wherever possible state the direction of the hypothesis (e.g. Ha: As people get older they become more conservative. Ho: Age has no effect on how conservative people are.) Choose the appropriate measures of association to test the above hypotheses. Interpret the results. It is OK to state the hypotheses after examining the correlations obtained.

Run the appropriate inferential statistic to identify whether or not the results can be generalized . Identify the probabilities of being wrong.

PART THREE: Select 1 of the 5 hypotheses for further analysis. Select 3 control variables (usually these will be the other independent variables) for the hypothesis and test to see if the original bivariate association changes. Interpret the results. Which mode of elaboration is suggested by the results (spuriousness, suppression, specification/interaction, or no effect)?


After opening SPSS, click on file (upper left) then open, select data. Change to the appropriate drive (e.g. f:) and open the folder called Common. Choose any database with the .sav extension (the last three letters of a file name). This will place SPSS into a data window with all the variables across the top.

Select Utilities and choose variables. Highlight and choose variables using the paste function. When at least five independent and three dependent variables have been chosen close the dialog box. SPSS will automatically go to the syntax window and the variables will be typed in. Enter before and after the variables the following text until the syntax box resembles the text below (it is critical that the typing be accurate).

Save out="a:newfile.sav"/keep= age sex or whatever variables were pasted, then a period.


Both periods are important, newfile can be any name (yours) but the extension must be .sav.

Highlight the above two lines and click on the icon > off the toolbar just below and to the right of Help. This will run the syntax command.

Go to file and open the newfile (change to drive a:). If new variables are desired, add them by reopening the original database, click on the desired variable (it will be highlighted), go to edit and choose cut or copy, then reopen the newfile and click on an unused variable column. Go to edit and choose paste.

Once again the variables will be across the top and SPSS will be in the data window. Examine the variables by double clicking on the name of each variable. A dialog box will open.

Click on Labels to see what the numeric values the data was coded as actually stand for. Several values may have no meaning for the research being done (e.g. no response, refused etc.). Click on missing values and type in the appropriate values or range of values to declare as missing and therefore not to be used in computing statistics involving this variable.

In some instances, the research team may want to examine a population which is controlled for a single variable. For instance, if the research was just about women, the first step would be to transform Sex into a new variable female (see below on how to recode). Then under data, choose select cases and click on filter. Place (by using the radio button and clicking) the variable female into the filter box and choose OK. From now on, all output will be data only on females. Make sure to do this before computing any descriptives each time SPSS is run.

Occasionally variables will have been coded without even spreads or with an inverted ordinal order. When this happens it will be necessary to recode the data. To do this select Transform then recode then into another variable. A dialog box will open.




Select (by clicking and using the transfer arrow) the variable to be recoded and type in its new name (in the example age is being changed to ageord). Type in the label to understand the difference between age and ageord. Then click on Old and New Values.

A dialog box will open.


In this example, age is being transformed from interval data to ordinal data with just 3 categories: old, middle, and young. The process involved choosing a range of ages and recoding them as a single value. Remember to make the spreads even. When all the desired ages have been changed choose All other values and make them System-missing. This will automatically declare the unchosen ages as missing values. Note that the oldest ages received the highest value. It is the same as ranking in increasing order all ordinal categories.

Some variables may have a range of values that go from one extreme to the other. For example, political affiliation might go from strong Democrat to strong Republican. Variables that do this must be recoded into two separate variables (e.g. Democrat and Republican). Remember to make the strong Democrat a higher value than a middle or weak Democrat. Once the variables have been recoded (they must be recoded not just relabeled), they will need to have their value labels named. Do this by double clicking on the variable and opening the label box.

In the previous example, the Define Labels would be for ageord not age and the Variable label would be ordinal values for age. Simply fill in the Value and then the Value Label and click Add. When done click Continue and then OK.

Remember to save the recoded data.

The next step is to provide output. The central tendencies and dispersion of each variable can be computed by choosing Statistics then summarize then frequencies. This will open up the following dialog box.

Place all the appropriate variables into the box on the right, then make sure Display frequency tables is checked and choose Statistics. Choose all the statistics that have been covered in this course as displayed below.

Choose Continue and then OK. SPSS will compute all the chosen information and make Output the active window. The Output window can be edited by simply typing in any desired information. Place name, class, and project goal at the top. Remember to save this Output to the a: drive. If the Output window is active, just save and it will automatically give the file name an extension of .lst. It is a good idea to save the data and output onto two separate discs to protect from virus or loss.


Once the descriptive output has been obtained, run statistics to test association and inference. This is done by choosing Statistics then crosstabs. This opens the following dialog box.


Place the independent variables in one box and the dependent variables in the other. Normally the independent variables are in the columns and the dependent variables are in the rows, this doesn’t really matter as the tables will be suppressed and the statistics will be computed in either format. Make sure the Suppress tables box is checked (the statistics themselves are to be interpreted). Then click on Statistics, which will open the following dialog box.

Check these statistics, then Continue and OK. SPSS will compute all the associations and make Output the active window. Examine the associations. Remember to use the blue chart as a guideline for which statistic to interpret the association. Level of Measurement is very important.

Find a relationship which looks interesting and return to the crosstabs dialog box. Place the interesting variables into their respective column and row, then choose an independent variable and place it into the bottom box. This will cause a control situation similar to partial gamma. Do this for three different control variables and examine the output. Did the relationship change? Why or why not?

The interpretation of the Output is the assignment. When all the unnecessary page breaks and information has been edited out, save the output and then print it. To save paper, print the output file only when all the information is correct and the file is ready to be turned in. (Remember the staff center will not print out delayed material unless it is signed for at the front desk. )

Examine the printed material. Highlight the appropriate statistics for the data relating to the 5 hypotheses. Interpret the computed values, the association and inference statistics are interpreted in the same manner as hand calculated statistics. Getting these right is the main assignment.

table of contents