Stat Computing > SPSS > FAQ

SPSS FAQ
How can I analyze my data by categories?

Sometimes you may want to analyze your data based on categories or a grouping variable. One way that you could do this is to split the data file into different data files and conduct the same analyses on the two (or more) data sets. However, that is cumbersome and error prone. Several commands in SPSS will allow you to do separate analyses by category, and we will consider them below.

Let's use the example data set below. You will notice that one of the independent variables, iv1, is a string variable. We will use this variable as our grouping variable to demonstrate how to use a string variable as the grouping variable. All of the techniques that will be shown can be used with a numeric categorical variable as well.

data list list / sub * iv1 (A)  iv2 * dv1 dv2.
begin data
1 "1" 1 48 25
2 "1" 1 49 37
3 "1" 1 50 55
4 "2" 1 17 19
5 "2" 1 20 38
6 "2" 2 23 48
7 "2" 2 28 44
8 "3" 2 28 68
9 "3" 2 30 30
10 "3" 2 32 37
end data.

To begin with, suppose we wanted to find the mean and standard deviation for dv1 for groups one, two and three in iv1. We can use the means command to obtain simple descriptive statistics.

means tables= dv1 by iv1.

Case Processing Summary

	Cases
	Included		Excluded		Total
	N	Percent	N	Percent	N	Percent
DV1 * IV1	10	100.0%	0	.0%	10	100.0%

Report
DV1

IV1	Mean	N	Std. Deviation
1	49.0000	3	1.00000
2	22.0000	4	4.69042
3	30.0000	3	2.00000
Total	32.5000	10	12.25878

You could also use the examine command, as shown below. We will use the /plot = none statement to suppress the stem-and-leaf and boxplots.

examine dv1 by iv1
 /plot = none.

Case Processing Summary

	Cases
	Valid		Missing		Total
	N	Percent	N	Percent	N	Percent
DV1	10	100.0%	0	.0%	10	100.0%

Descriptives

			Statistic	Std. Error
DV1	Mean		32.5000	3.87657
	95% Confidence Interval for Mean	Lower Bound	23.7306
		Upper Bound	41.2694
	5% Trimmed Mean		32.3889
	Median		29.0000
	Variance		150.278
	Std. Deviation		12.25878
	Minimum		17.00
	Maximum		50.00
	Range		33.00
	Interquartile Range		26.0000
	Skewness		.516	.687
	Kurtosis		-1.278	1.334

Case Processing Summary

		Cases
		Valid		Missing		Total
	IV1	N	Percent	N	Percent	N	Percent
DV1	1	3	100.0%	0	.0%	3	100.0%
	2	4	100.0%	0	.0%	4	100.0%
	3	3	100.0%	0	.0%	3	100.0%

Descriptives

	IV1			Statistic	Std. Error
DV1	1	Mean		49.0000	.57735
		95% Confidence Interval for Mean	Lower Bound	46.5159
			Upper Bound	51.4841
		5% Trimmed Mean		.
		Median		49.0000
		Variance		1.000
		Std. Deviation		1.00000
		Minimum		48.00
		Maximum		50.00
		Range		2.00
		Interquartile Range		.
		Skewness		.000	1.225
		Kurtosis		.	.
	2	Mean		22.0000	2.34521
		95% Confidence Interval for Mean	Lower Bound	14.5365
			Upper Bound	29.4635
		5% Trimmed Mean		21.9444
		Median		21.5000
		Variance		22.000
		Std. Deviation		4.69042
		Minimum		17.00
		Maximum		28.00
		Range		11.00
		Interquartile Range		9.0000
		Skewness		.543	1.014
		Kurtosis		-.153	2.619
	3	Mean		30.0000	1.15470
		95% Confidence Interval for Mean	Lower Bound	25.0317
			Upper Bound	34.9683
		5% Trimmed Mean		.
		Median		30.0000
		Variance		4.000
		Std. Deviation		2.00000
		Minimum		28.00
		Maximum		32.00
		Range		4.00
		Interquartile Range		.
		Skewness		.000	1.225
		Kurtosis		.	.

Now let's a technique that is more general and that can be used with any type of analysis. First, we need to sort the data by by our grouping variable, in this case, iv1. Then we split the file by the same variable. The split file command temporarily splits the file by the variable specified. All analyses will be grouped by this variable until the split file off command is issued, or until the data are resorted. Note that the split file command can be used with numeric, short and long string variables. (Many SPSS commands will not work with long string variables, but split file will.) Next, list the commands for the analyses that you would like. Finally, issue the split file off command.

sort cases by iv1.
split file by iv1.
correlations var = dv1 with dv2. 

Correlations

IV1			DV2
1	DV1	Pearson Correlation	.993
		Sig. (2-tailed)	.073
		N	3
2	DV1	Pearson Correlation	.780
		Sig. (2-tailed)	.220
		N	4
3	DV1	Pearson Correlation	-.766
		Sig. (2-tailed)	.444
		N	3

split file off.

Note that you can use more than one variable to categorize your analysis. To do so, list all of the variables by which you want the analysis categorized in the sort cases command and in the split file command.

sort cases by iv1 iv2.
split file by iv1 iv2.
correlations var = dv1 with dv2. 

Correlations

IV1	IV2			DV2
1	1.00	DV1	Pearson Correlation	.993
			Sig. (2-tailed)	.073
			N	3
2	1.00	DV1	Pearson Correlation	1.000
			Sig. (2-tailed)	.
			N	2
	2.00	DV1	Pearson Correlation	-1.000
			Sig. (2-tailed)	.
			N	2
3	2.00	DV1	Pearson Correlation	-.766
			Sig. (2-tailed)	.444
			N	3

split file off.

UCLA Researchers are invited to our Statistical Consulting Services
We recommend others to our list of Other Resources for Statistical Computing Help
These pages are Copyrighted (c) by UCLA Academic Technology Services