Example:

Betweensubjects IV = Gender (Levels: male, female).

Withinsubjects IV = Treatment (Levels: before, after)

DV = Stress level

In the Variable View, under the Values column, "male" was assigned a value of "1", and "female" was assigned a value of "2"
Step 1:
Analyze > General Linear Model > Repeated Measures...
Mixed ANOVAs
ANOVA stands for Analysis of Variance. It is used to determine whether there is a significant difference between the means of three or more groups. It is an omnibus test, which means that it cannot tell you where the significant difference lies, such as whether group A is significantly different from group B or group C. You determine which groups were significantly different from each other by carrying out posthoc tests, which will also be covered here.
There are three types:

A betweensubjects/independent samples ANOVA

A withinsubjects/repeated measures ANOVA

A mixed ANOVA
We will deal with the third type here. A mixed ANOVA compares the mean differences between groups where at least one factor is a "withinsubjects" Independent Variable (IV) and at least one other is a "betweensubjects" IV. The purpose of a mixed ANOVA is to understand if there is an interaction between these IVs on the dependent variable (DV).
Assumptions:

The DV should be scale data (either interval or ratio)

There should be at least one withinsuvject IV and one betweensubjects IV

The IVs should consist of categorical groups

There should be no obvious outliers

The data should be roughly normally distributed for each group of the IVs
2x2 Mixed ANOVA(example)
Step 4
Click on Plots... and make one IV the Horizontal Axis and the other the Separate Lines. Click Add.
Click Continue
Because there are only 2 levels in each IV, there is no need for posthoc tests, and so you can simply click OK to get the output
Step 3
Move each of the levels of the withinsubjects IV on the left hand side of the new dialogue box called Repeated Measures into the WithinSubjects Variable (Stress) section on the right and the betweensubjects IV (Gender) into the BetweenSubjects Factor(s) section on the right.
Step 5: Look at the Output
The means and the upper and lower bounds in the "Gender * Stress" table can be used in the writeup of the descriptive statistics, including the
Confidence Intervals.
N.B. In this example, there is no need to calculate individual effect sizes by hand because there are only 2 levels in each IV.
You also do not need to look at the "Mauchly's Test of Sphericity" table because there are only 2 levels in repeatedmeasures IV.
For the inferential statistics, the main tables you need to look at are the "Test of WithinSubjects Effects" table and the "Test of BetweenSubjects Effects" table.
Step 2
In the Repeated Measures Define Factor(s) dialogue box, type the name of the IV into the WithinSubject Factor Name box and enter the number of levels into the Number of Levels box. Then click Add, followed by Define
Step 8: Write up
Figure 1. Mean Stress Score and Confidence Intervals for males and females before and after treatement
Both males and females had lower stress scores after treatment. The mean stress score for males before treatment was 37.1 compared to 18.85 after treatement; the mean stress score for females before treatment was 37.45 compared to 24.15 after treatement. The confidence intervals show that the means are reasonably close to the population mean. *Report upper and lower bounds of the confidence intervals*
A 2 x 2 mixed ANOVA revealed a significant main effect of Treatement at the 5% level f(1, 38) = 1910.851, p < 0.001(twotailed). The Global Effect size using Partial Eta Squared was 0.981, which is a large effect. The observed power was 1, which strong. The main effect of Gender was nonsignificant, f(1, 38) = 0.485, p = 0.49 (twotailed).
The interaction between Treatment and Gender was significant at the 5% level f(1, 38) = 47.0.37, p< 0.001). The Global Effect size using Partial Eta2 was 0.553, which is a medium effect. The observed power was 1, which is strong.
Figure 2. Interaction Plot for Treatment and Gender
Both males and females had a similar level of stress before the treatment, which reduced after the treatment. Stress levels for males after treatment was greaters that stress levels for females after treatment.
To investigate the significant interaction, 4 followup ttests were conducted and a Bonferroni correction was applied. There was a nonsignificant difference between males and females before treatment, t(38) = 0.086 p > 0.013 (twotailed), and after treatment, t(38) = 1.302, p > 0.013 (twotailed). There was a significant difference between the before and after treatment conditions for males, t(19) = 29.603, p < 0.001 (twotailed), and for females, t(19) = 21.59, p < 0.001 (twotailed). The effect size was 1.42 for males, which is very large, and 1.03 for females, which is also very large.
Step 6: Followup the Significant Interaction
Followup tests are conducted using paired and independent ttests as appropriate. Follow the links to remind yourself how this is done
There will be four altogether: two paired sample ttests for each level of the Gender IV, and two independent ttests for each level of Treatment IV.
Do the independent ttests first, following the guidelines already given in this website, but put both the Treatment levels in Test Variables box. You'll get the same tables as a normal independent ttest (the "Group Statistics" table and the "Independent Samples Test" table), but they will each have two rows instead of one, one for each treatment level.
For the paired samples ttest, you first need to go to Data > Split File..., then check "Organize output by groups" and put Gender in the "Groups based on" box. Click "OK"
Now when you conduct the paired samples ttests, following the guidelines already given in this website, you will get separate outputs for males and females.
N.B. You will have to apply a Bonferroni correction on the pvalues when reporting the results. This means that the cut off for significance will be 0.0125 instead of 0.05, which is 0.05 divided by 4 (the number of tests)
Step 7: Calculate the Effect Sizes
To calculate the effect size (Cohen's d), you need the means (x) and standard deviations (SD).
d = (x1  x2) ÷ ((SD1 + SD2) ÷ 2)
Very Small < 0.3
SmallMedium > 0.3; < 0.5
MediumLarge > 0.5; < 0.8
Large > 0.8
3x5 Mixed ANOVA(example)
Example:

Between subjects IV = Login Method (Levels: Password, PIN, Fingerprint, Iris Recognition, USB key).

Withinsubjects IV = Delay (Levels: during training, shortly after training, a long time after training)

DV = Login Time (in seconds)

In the Variable View, under the Values column, "Password" was assigned a value of "1", "PIN" was assigned a value of "2", "Fingerprint" was assigned a value of "3", "Iris Recognition" was assigned a value of "4", and "USB key" was assigned a value of "5"
Step 1:
Analyze > General Linear Model > Repeated Measures...
Step 2
In the Repeated Measures Define Factor(s) dialogue box, type the name of the IV into the WithinSubject Factor Name box and enter the number of levels into the Number of Levels box. Then click Add, followed by Define
Step 3
Move each of the levels of the withinsubjects IV on the left hand side of the new dialogue box called Repeated Measures into the WithinSubjects Variable (Delay) section on the right and the betweensubjects IV (Gender) into the BetweenSubjects Factor(s) section on the right.
Step 4
Click on Plots... and make one IV the Horizontal Axis and the other the Separate Lines. Click Add.
Click Continue
Step 5
For the betweensubjects IV, click on Post Hoc... move LoginMethod to the Post Hoc Tests For: box, and then check Scheffe. Click Continue.
Now click on Options... move the withinsubjects IV (Delay) and the interaction to the Display Means For: box, and then check Descriptive statistics, Estimates of effect size, and Observed power (and Parameter estimates if you want).
To get posthoc comparisons for the withinsubjects IV, you also need to click Compare main effects, and select Bonferroni from the dropdown menu.
Click Continue and then OK to get the output.
Step 6: Look at the Output
The means and the upper and lower bounds for the withinsubjects IV in the "Parameter Estimates" table and for the betweensubjects IV in the "Estimates" table can be used in the writeup of the descriptive statistics, including the Confidence Intervals. The effect sizes can be calculated based on the numbers in the Descriptive Statistics table.
For the inferential statistics, the main tables you need to look at are the "Mauchly's Test of Sphericity" table, "Test of WithinSubjects Effects" table and the "Test of BetweenSubjects Effects" table.
Mauchly's test of sphericity tells you whether the variances of the differences between the conditions are equal. The "Mauchly's Test of Sphericity" table displays this. If Mauchly's test statistic is significant, it means that that there are significant differences, and that the assumption of sphericity has not been met. In this case, you read from the GreenhouseGeisser row in the "Tests of WithinSubjects Effects" table when reporting the results of the ANOVA. If Machly's test statistical is not significant, you read from the first row of the "Tests of WithinSubjects Effects" table. In this example, Mauchly's test statistic is significant (< 0.05), meaning that we need to read from the GreenhouseGeisser row when reporting the outcome of the ANOVA.
The "Tests of BetweenSubjects Effects" table displays the main effect of the betweensubjects IV.
The "Pairwise Comparisons" table displays the results of the posthoc tests for the withinsubjects IV, telling us which groups differed from each other. In this case, there are significant differences between all but the short and long time delays.
The "Multiple Comparisons" table displays the results of the posthoc tests for the betweensubjects IV, telling us which groups differed from each other. In this example, the significant differences lie between the Password login method and the remaining four methods, and between the USB key method and the PIN and Fingerprint login methods.
Step 7: Followup the Significant Interaction
Followup tests are conducted using oneway ANOVAs as appropriate. Follow the links to remind yourself how this is done
There will be 8 oneway ANOVAs altogether: 3 oneway betweensubjects ANOVAs (one for each level of the withinseubjects factor, which has 3 levels) and 5 oneway withinsubjects ANOVAs (one for each level of the betweensubjects factor, which has 5 levels) were conducted.
Do the betweensubjects ANOVAs first, following the guidelines already given in this website, for each level of the withinsubjects IV. You'll get the same tables as a normal oneway betweensubjects ANOVA for each test you run.
For the withinsubjects ANOVAs, you first need to go to Data > Split File..., then check "Organize output by groups" and put LoginMethod in the "Groups based on" box. Click "OK". Now when you conduct the withinsubjects ANOVAs, following the guidelines already given in this website, you will get 5 outputs, one for each level of the betweensubjects IV.
Step 8: Calculate the Effect Sizes
To calculate the effect size (Cohen's d), you need the means (x) and standard deviations (SD).
d = (x1  x2) ÷ ((SD1 + SD2) ÷ 2)
Very Small < 0.3
SmallMedium > 0.3; < 0.5
MediumLarge > 0.5; < 0.8
Large > 0.8
Step 9: Write up
Figure 1. Mean Login Time (seconds) and Confidence Intervals During Training, Shortly after Training, and Longer after Training for Each of the Login Methods
Passwords resulted in the slowest login times at each login time. Iris recognition had the shortest login time at during training, and this did not differ much after a short or a long delay. The USB key had the shortest login time shortly after training and longer after training. For the password condition, the mean login time during training was 40.5 seconds, compared to 26.6 seconds at the short delay, and 34.5 seconds at the long delay. For the PIN condition, the mean login time during training was 15.05 seconds, compared to 16.4 seconds at the short delay, and 11.85 seconds at the long delay. For the fingerprint condition, the mean login time during training was 16.4 seconds, compared to 12.15 seconds at the short delay, and 14.15 seconds at the long delay. For the iris recognition, the mean login time during training was 12.35 seconds, compared to 12.65 seconds at the short delay, and 11.85 seconds at the long delay. Finally, for the USB condition, the mean login time during training was 14.5 seconds, compared to 9.15 seconds at the short delay, and 4.85 seconds at the long delay. *Report upper and lower bounds of the confidence intervals*
A 3 x 5 mixed ANOVA revealed a significant main effect of Time Delay at the 5% level f(2, 190) = 16.096, p < 0.001. The Global Effect size using Partial Eta Squared was 0.145, which is a small effect. The observed power was 1, which strong. The main effect of Login Method was also significant, f(4, 95) =2280.684, p < 0.001. The Global Effect size using Partial Eta Squared was 0.864, which is a large effect. The observed power was 1, which strong.
The interaction between Time Delay and Login Method was significant at the 5% level f(8, 190) = 253.263, p < 0.001). The Global Effect size using Partial Eta Squared was 0.214, which is a small effect. The observed power was 1, which is strong.
Figure 2. Interaction Plot for Login Method and Time Delay
To investigate the significant interaction, 8 followup oneway ANOVAs were conducted. There was a significant difference between the login methods during training, f((4, 95) = 2730.985, p < 0.001, after the short delay, f(4, 95) = 23.98 p < 0.001, and after the long delay, f(4, 95) = 71.15, p < 0.001. Sheffé posthoc tests revealed that significant differences lie between the Password condition and all other login method conditions at all levels of the Time Delay condition (p < 0.001 for all). After a short delay, there was also a significant difference between the PIN and the USB Key conditions (p = 0.011). Additionally, after a long delay there were significant differences between the USB Key condition and all other login method conditions (p = 0.011 for PIN and for Iris Recognition, p < 0.001 for Fingerprint, pvalue already reported with the Password results). All other posthoc conditions were nonsignificant (p > 0.05). *report cohen's d for each significant posthoc comparison* There were also significant differences between the time delays for the Password condition, f(2, 38) = 24.508, p < 0.001, and the USB Key condition, using the GreenhouseGeisser correction, f(1.164, 22.114) = 29.067, p < 0.001. Bonferroni posthoc tests revealed that for both the Password and the USB Key conditions, significant differences lay between all the levels of the Time Delay IV. Specifically, for the Password condition, p < 0.001 for the difference in login times between training and after a short delay, p = 0.003 for the difference in login times between a training and after a long delay, and p = 0.009 for for the difference in login times between a short delay and a long delay; for the USB Key condition, p = 0.002 for the difference in login times between training and after a short delay, p < 0.001 for the difference in login times between a training and after a long delay, and p < 0.001 for for the difference in login times between a short delay and a long delay. *report cohen's d for each significant posthoc*