Steps for data analysis:
- Rename my variables into something that is easily recognizable.
- Use the frequencies command on my categorical variables to get an idea of how my data set looks – this will be important for methods section to write about stuff such as how many men vs women you sampled.
- Use the compute command to create a Mean or a Sum score for participants’ ratings of YouTube, using only those who responded to every question on my survey.
- Next you need to filter out people who didn’t meet the criteria for my analysis, such as those who didn’t attend the University of Hail.
- If you run the descriptives command for my rating variable, you should have an N (sample size) of 742. This is also where you’ll find the mean/standard deviation.
- Next you need to run the following analysis: a Factorial ANOVA.
- Now, run the same analysis, but remove any nonsignificant interaction terms. Make sure to include post-hoc tests for predictors with more than two levels that are significant. You need also use the plot function to graph anything you feel you need to. Also you have to talk about specific means (such as men vs women), report estimated marginal means. You also want to include measures of effect size. You would also run homogeneity tests to see if my data fits the assumptions of the test.
- If you ran the first analysis according to these steps, you should have 742 under the “df” column in the “total” row of the between-subjects table. For the second analysis, this should be 804. These are degrees of freedom, and they correspond to my sample size.
you should end up with 742 sample sizes and if you did get there then you did what I did exactly and you’re on the right track and good to go.
Steps for writing results:
- You essentially you need to start by specifying exactly what type of test you ran, such as: “A 2 x 2 x 4 mixed factorial analysis of variance was conducted.” This is determined by the number of levels your variables have, and whether they are within or between subjects.
- You then need to state what your outcome variable was, and what your independent variables were, along with the levels of those variables.
- Next, you should talk about significant interactions if they are there. If not, mention that there were no significant interactions and talk about main effects (report significant and nonsignificant effects). Here are examples of the format for how to report F statistics from previous work I have done:
There was no main effect of biological sex, F(1,603)= .095,p>.05,η ̂^2= .002.
There was also a main effect of perception source, with victims reporting greater relationship violence than perpetrators, as well as a main effect of relationship duration, F(1,603)=25.95,p< .001,η ̂^2= .04 and F(4,603)=7.64,p< .001,η ̂^2= .05, respectively.
- Here’s what the different pieces are/mean:
F(1,603) – this is denoting that you ran an F test, with the degrees of freedom for the variable on the left, and the total degrees of freedom on the right
= 25.95 – this is the value of the F test
p < .001 – this is the significance of the F test. Generally you report one of the following numbers: > .05 if the result was nonsignificant, or < .05, < .01, < .001, whichever is closest to your result.
η ̂^2= .05 – This is called “eta squared”, it is a measure of effect size. Basically, how much your variable influences the outcome.
- If you have a significant variable with more than two levels, you should talk about those differences using post-hoc tests of pairwise comparisons. Make sure to specify which post-hoc tests you ran and how you controlled for familywise Type I error (not all tests do this automatically). Here’s an example of how to write about these from previous work:
Among perpetrators, males reported less average relationship violence than females, p < .001.
- For the violations of the test assumptions, consider mentioning them somewhere.
Ibecause I want to know more about what these violations are/how to identify and remedy them, a search for “ANOVA test assumptions” should help.
Here’s an example of an APA format graph – you can modify the labels and font, etc, in SPSS:
Figure 1. Estimated mean ratings of perceived violence in relationships for victims by biological sex and perceived relationship duration (N =61