What is the method used to test when there is no homogeneity in the SPSS?

1.Non-parametric tests.

2.In the analysis of variance dialog box, there are several options in the bottom column specifically for testing when the variance is uneven.

You can use the uneven variance in ANOVA, or the nonparametric rank-sum test.

Methods for performing the homogeneity test of variance at SPSS:

Open a copy of the data for which you want to perform the variance homogeneity test, and then click the Analyze - Descriptive Statistics - Explore <>

1. Open the Explore dialog box, then select the variable to be tested for variance homogeneity in the list of dependent variables, and select the variable to be divided into two groups in the factor list.

2. Then open the [Statistics] subdialog box, and then check the [Descriptive] <>

3. Then open the [Drawing] subdialog, then check [Normal Diagram with Inspection], and then select [Unconverted] in the stretch and level Levene test

4. Then open the [Options] subdialog box, and then check the <> [Exclude Cases by List].

5. Click OK to see the variance homogeneity test, and then you can check whether the variance of the two independent samples is homogeneous according to different data and different situations.

Summary. Theoretically, one-way ANOVA should first satisfy the homogeneity of variance, but in the actual research process, more data have variance irregularity, and the classification data X can be recombined by selecting "Data Encoding" in the "Data Processing" option of SPSSAU to recombine the data.

What to do if the variance of the SPSS ANOVA is uneven.

Theoretically, one-way ANOVA should first satisfy the homogeneity of variance, but in the actual research process, more data have variance irregularity, and the classification data X can be recombined by selecting "Data Encoding" in the "Data Processing" option of SPSSAU to recombine the data.

Or take the logarithm or open the root number of the dependent variable y. The specific method is as follows: select "Data Encoding" - "Generate Variable", drag the variable (question item) to the right, and click Generate New Variable.

In addition, if the categorical data of the study are of two categories, you can consider using an independent sample t-test to represent the ANOVA to avoid the embarrassment that the variance is uneven and cannot be analyzed.

Whether it is a one-way ANOVA or a multi-factor ANOVA, you can open the ANOVA dialog box first, and then click the "Options" button, and there is an option for the homogeneity test of variance in the pop-up dialog box, just check it. The homogeneity of variance can be seen by the results of the homogeneity of variance F test, if the sig value is greater than that, it is homogeneity, which is suitable for ANOVA analysis. It is said that the heterogeneity cannot be followed by ANOVA, because in the mean test (including ANOVA, t-test, etc.), the effect of each experimental treatment is considered to be a fixed effect, and the effect is the same for everyone, that is, the effect of the treatment is to add the same constant to the original level of each person, so that what is the original variance of each subject group, and what is the variance after the experimental treatment, then, if the variance of different subject groups is uneven, that is, the ratio of variance is significantly not equal to 1, If it means that there is a large difference between the subjects, then our ANOVA will not be able to reach an accurate conclusion, and we do not know whether the experimental treatment caused the differences between different groups of subjects, or whether it also confuses individual differences.

In principle, ANOVA cannot be carried out due to the unevenness of variance, but the ANOVA in SPSS is done under the framework of least squares, which is different from the analysis method of ANOVA introduced in educational and psychostatistical textbooks. In SPSS, homogeneity is not a necessary condition for ANOVA. It's just that the textbook is to introduce you to the general principle, and I don't know much about the performance of the latest software, so I have to be homogeneous.

Moreover, the homogeneity test is generally not reported in the ANOVA. So you just go straight to ANOVA. If you are still not sure, you can do some data conversion to make it close to homogeneity, such as box-cox conversion, logarithmic conversion, etc.

The results of the homogeneity test for variance** are based on the mean, based on the median. and so on, and the corresponding significance level, i.e., SIG value, was obtained. t usually stipulates that the sig < rejects the variance homogeneity assumption and supports it instead.

This is the direct purpose of doing the sameness test of variance, so it is the most useful.

The previous one is df, which is the degree of freedom, which can be simply understood as a category or a statistic.

Degrees of freedomWhen the parameters of a population are estimated by the statistics of the sample, the number of independent or freely variable independent variables in the sample is called the degrees of freedom of the statistic.

It mainly depends on whether p is greater than that.