RLST 4900
These instructions are being placed in the Samdahl Outbox on the computers in the Ramsey computer lab, and also online via the Samdahl home page. However, the discussion below will make most sense if you read it while at a computer that has SPSS so you can follow along and practice these commands while you read. The instructions are written for SPSS 11.0.1 which is the version in the Ramsey computer lab.
The examples below use a different data file than the one used for class. Therefore, the screen images shown here do not exactly match what you see when you follow these procedures with the class file. However, the steps for doing analyses are the same. Details of that assignment can be found on the class web page.
If you find the SPSS program confusing, you might want to re-read pages 6-8 in the online link to Instructions for SPSS Data Entry (which opens a file called "Introduction to SPSS").
The following discussion uses the point-and-click menu system in SPSS to design and execute some basic analyses. To use this menu system, open the class data file and have the data file (looks like an Excel spreadsheet) visible on your monitor.
File / Open / Data
The toolbars and menus described below are visible only when you are looking at this Data Window. When you execute an analysis you are automatically moved to the Output Window which had different toolbars across the top. To return to the Data Window, look for shortcut link at the bottom of your screen, or go to the top toolbar and select Window (the actual procedure differs depending on which version of SPSS and Windows you are using).
The Output Window is divided into two frames. On the left is an outline of all procedures you have run, and on the right are the results of each of those analyses. You can easily move up and down in your output file by clicking on the appropriate spot in the outline on the left. Or, you could temporarily close any section of your output by clicking on the box (+/-) near that procedure in the outline. You can also delete any analysis you no longer want by highlighting and then deleting those items from the outline in the left window.
If you run several analyses the output file can get quite large. At the end of this handout are instructions for printing or copying only that portion of the output that you want to print.
I. DESCRIPTIVE STATISTICS
There are two ways to get basic descriptive statistics on a variable in the SPSS file. You can use the procedure Frequencies, or the procedure Descriptives. Both are described below.
Analysis using the Command Frequencies. With the data window visible on screen, select Analyze / Descriptive / Frequencies. You'll get a screen like the one below. All variables are listed in the left of this box. To select the variable(s) you want to see frequencies for, click on that variable name on the left then click the arrow that points to the right. The variable you had highlighted will move into the box on the right. To move it back, click on a variable from the list on the right then click on the arrow that points to the left.
The example below shows that the variable Leisure is boring [var2] has been selected for analysis (it was moved from the left into the box on the right).
Basic dialog menu box for Frequencies
Clicking on OK executes your command and tells SPSS to run frequencies on this variable. If you are running frequencies on a lot of variables, you can move all of them into the box on the right before clicking OK.
The output of Frequencies is shown below. The first box shows you how many cases were analyzed ("valid" cases) and how many were not included because they had missing values. The second box shows each value for that variable and how many cases in the data set have that value. In the box below, 132 people answered "Strongly Disagree," 128 answered "Agree," etc.
After running your first Frequencies take a moment to study the frame at the left that has the outline in it. Understanding how to jump around using this outline might be useful later.
The statistics menu for Frequencies. In this example, I've requested
to see the mean and the standard deviation for the selected variable.
I exited from the above screen by clicking on Continue. Then I executed this analysis by clicking on OK. The screen below illustrates output I got. Because it was still computing Frequencies a portion of this output is identical to what I had before. However, now the statistics I requested are shown above the actual frequency display.
Output Window for Frequencies, with Statistics
Analysis using the Command Descriptives. If you want more elaborate descriptive statistics on a variable and you don't need to see the frequency table, you can get this using the commands Analyze / Descriptive / Descriptives. The procedure is very similar to that described above. The first menu box lists all possible variables and you select the one(s) you want to analyze by highlighting them in the box on the left then clicking the arrow to move them into the box on the right. In the example below I've selected the variable I decide with whom I do recreation to analyze. Remember, if you select the wrong variable simply move it back to the left by highlighting it then selecting the arrow that points left.
Descriptives dialog box
Descriptives Options
Output from Descriptives
The above output shows the total number of cases, the minimum and maximum scores, the mean, and the standard deviation for the variable I selected.
Crosstab tables have been discussed in class and you should review that material. In particular, if you think one variable might be causally influencing the other that first variable (the independent variable) must go along the top to define the columns in the table. The dependent variable goes along the side to define the rows. We also discussed the importance of looking at percentages in addition to the actual numbers in the table. The procedure below shows you how to request a crosstab table from SPSS.
Enter into the Crosstabs menu with the commands Analyze / Descriptive / Crosstabs. You'll get a menu box like the one shown below, very similar to the menu boxes you saw for Frequencies or Descriptives. Notice, though, that you select which variable will define the rows and which will define the columns. The independent variable goes at the top of the table so that should be the variable you use to define the columns.
In the example below I've asked for a table that has gender across the top and the variable Social Interaction along the sides. I'm looking to see if gender influences people's desire for social interaction in leisure.
Main dialog box for Crosstabs
When working in the Crosstabs dialog box you can change your selections by clicking on any variable on the right and then clicking the arrow to move that variable back to the left. That clears the box so you can different variables for your analysis.
Once you select the variables for your table you need to tell SPSS what statistics you want and what information you want to appear in each cell. You do this using the boxes at the bottom of the Crosstabs menu.
The example below shows the dialog box for Crosstab Statistics. In this example I've asked SPSS to compute Chi-Square and nothing else.
Dialog box for Crosstabs Statistics
I clicked on Continue from the above screen and returned to the main Crosstabs menu. Then I selected Cells from the bottom of that screen and got the menu box shown below. Here I've asked SPSS to display the actual count for each cell (how many people were in that cell) as well as the column percentages. I asked for column percentages because of the rule "Percentage down, compare across" which helps me analyze the data in the table more easily.
Dialog box for Crosstabs Cells
I clicked on Continue from the above screen, then clicked on OK from the main Crosstabs menu to execute this analysis. The results are shown below. Though this table might look a bit different than those we worked on in class, it has labels for each column and row and the cells show the actual count plus the percentages just like the class examples.
Output Window showing the Crosstabs table
The statistics from a crosstab analysis are shown below the crosstab table. SPSS computes Chi square three different ways. Don't let that confuse you; simply use the top one (Pearson's Chi-square). SPSS also computes the p-value (the "significance") for each Chi-square. We usually jump right to the p-value because it's hard to interpret the Chi-Square number. In the example below, the p-value for Pearson's Chi-Square is <0.001. This tells me that I can be fairly confident that there really is a relationship between these two variables (with a 1 in a 1000 chance of being wrong when I say that).
Output for Crosstabs Statistics
One other common statistic you might be interested in is a correlation. The true correlation (called Pearson's correlation coefficient) must be calculated on interval level data (it cannot be used for nominal or ordinal variables). Researchers often assume that Likert-like scales are interval level measurement so their data can be analyzed through correlations.
To get to the menu for correlations, select Analyze / Correlate / Bivariate. The "bivariate" simply means that you want the correlation between two variables (the other options let you compute complex correlations between three or more variables). The menu box is similar to the others you've seen in SPSS analyses. You select the variables from the left and move them to the right to identify which variables you want correlated. If you select more than two variables, SPSS computes the correlation between all possible 2-way combinations of those variables. This main menu also lets you select the type of correlation you want (typically you want Pearson's correlation) and choose how the p-values should be calculated (typically you want two-tailed tests). SPSS will also flag the significant correlations with asterix if you ask for it (the box on the bottom left of the menu).
Here is an example of the correlations menu. I've asked for the correlation between answers to My leisure is boring and I'm pretty happy with myself. I was wondering if people who are happy with themselves probably are less likely to be bored in their leisure.
Dialog box for Correlations
From the above box I clicked on OK to execute the command. The results are below.
Output Window for Correlations
The output for correlations lists each of the variables in the correlation and tells you how many valid cases there were with data on both of these variables (in the above example, N=340). A correlation matrix always has 1.0 running along the diagonal (top left to bottom right) because a variable correlates with itself at 1.0. SPSS shows both the top right and the bottom left halves of this matrix; notice that they are identical and you only need to look at the bottom left half of the matrix. If you've only included two variables in your analysis, the correlation you are looking for is that bottom left cell in the output.
In the example above, these two variables correlated at -0.427. This number is negative, like I expected, which means that people who are more happy with themselves are less likely to find leisure boring. On a scale of zero to one, a value of -0.427 indicates a fairly strong correlation between these two variables. The item marked Sig 2-tailed is the p-value testing how confident you can be that this correlation isn't a result of sampling error, and the asterix is used to mark all p-values that are lower than 0.05. In the example above, SPSS has determined that I can be fairly confident that this correlation would exist in the population too, not just in my sample. (This calculation stems from complex probability theory and is determined, in part, by how strong that correlation is and how many cases I had in my analysis).
When you've done an analysis that you want to print out, don't print the entire Output Window. Instead, go to the outline on the left side of the Output Window and highlight that analysis by clicking on the analysis you want. When you click on the main word Crosstabs or Correlations all output associated with that particular analysis will become highlighted in the outline and the screen on the right jumps to that point in your output. You can use File / Print / Selection to print only that particular analysis. Alternately you can use Edit / Copy Objects to copy and paste this material into a Word file (if you use simple copy you'll lose all the formatting for the tables).
|
