Calculate statistics with SPSS. Gerd_Altmann / Pixelio
SPSS is a commonly used at universities and in business software, can be made with different data calculations. Also variance analyzes are possible. Here, it is investigated whether there are significant differences between two groups of people or objects with respect to a characteristic. How should you go about doing this is explained below by way of example.
Check conditions of the analysis of variance
- An analysis of variance should only be performed when you present metric data. This means that the data should consist of numerical measurements (z. B. Size Dimensions, indicators).
- Data from questionnaires and scales can be used if the distances between the scales are the same points.
- Furthermore, the readings should be able to accept more than two categories. For example, if only be yes or no answers, then you should prefer to use the chi-squared test. Even this can be performed in SPSS.
- In addition, the data should be normally distributed, ie, resembling in its distribution of a bell curve, are in the middle forms most frequently. For this purpose the Kolmogornov-Smirnov test has to be calculated.
Normality Test in SPSS
- In SPSS you find this normal distribution test in the "Analyze" in sub-item "descriptive statistics". Select "exploratory data analysis" and unplug the dependent variable (the measure) in the right selection box.
- If you examine about whether men and women differ in their income, then the income would be your dependent variable.
- Under "charts" you should select "normal distribution diagram with test" so that the Kolmogornov-Smirnov test is calculated as a condition of analysis of variance. Then confirm your entries with OK.
- Now SPSS calculates the results and are more tables, the decisive factor is the overview "Normality Test". Here are the results sought in the column "Kolmogornov-Smirnov test."
- There, the significance is given, so the probability of error. If this is greater than 0.05, so the result is not significant, then an analysis of variance can be calculated.
Calculate group differences in SPSS
- In "Analyze" menu contains the tab "General Linear Model". Select "Univariate analysis of variance" and add the dependent variable (z. B. income) and fixed factors (eg. As gender) a.
- You can "Options" choose to save also the descriptive statistics (means, standard deviations) should be calculated in the sub-item.
- In the following edition SPSS has summarized all the tables of analysis of variance. Here is the table "test for between-subjects effects".
- Decisive is the first line (corrected model). The value entered under the heading "significance" should be less than 0.05.
- This would mean that the differences between the groups are larger than the measured value fluctuations within the groups (z. B. that men and women differ more clearly than many women from each other).
- The direction in which this difference is (if z. B. Men and women have a higher income), can be identified by the listed groups for both averages in the descriptive statistics, the SPSS has also calculated.
If your data is not normally distributed, they can take the analysis of variance also use the Welch test, this is also in the selection menu "General Linear Model" and is asymptotic data, ie data that follows no bell distribution suitable.