![spss code for test of normality spss code for test of normality](https://ezspss.com/wp-content/uploads/2019/11/normal5.png)
Yes, Likert scale data are confusing! It seems to be intuitive and easy method for people taking the survey. If they’re different people, using the 2-sample t-test is a good choice. But, it just affects the analysis you use. But, you’d also have to have the same numbers before and after! Usually when I see a pretest/posttest analysis, it uses the same subject in both groups. You’re using pre and post surveys, but I’m gathering from the differences in numbers that it’s not the same group of people taking the survey? I’m asking because if they’re the same people, then you’d use a paired t-test. Hope you have time to answer my question. => which one makes most sense? Or do I need another type of analysis? When entering proportion satisfied cohort 1 0.50 and cohort 2 0.75, and again power 80%, aplha 0.05, two tailed, I get 58 patients to be included in each cohort Assuming satisfaction first two answer options=satisfied, other three answer options=not satisfied).
![spss code for test of normality spss code for test of normality](https://spss-tutorials.com/img/spss-shapiro-wilk-test-explore-dialog.png)
I tried:Ī) MW-U with expected means (most frequent response, a bit out of the blue but for example purpose) of 3 and 2 in cohort 1 and 2 respectively (SD 0.5, power 80%, aplha 0.05, two tailed), this resulted in 6 patients to be included in each cohortī) z-test, proportion, difference between two independent proportions (chi-2?). I have a question concerning type of test and calculation of a priori statistical power.ġ) power calculation: how many patients to include in the two cohorts? I generally use Gpower. Satisfaction is measured on a 5 point scale. We will assess satisfaction in a before and after cohort. I hope you are doing fine! We are planning a study to increase satisfaction through systematic/better pre-operative information. In that article, I refer to them as interaction effects, but that’s the same thing as a moderating variable. And, learn more about using moderators in a regression model. First, read about Spearman’s correlation and when you want to use it. Here are a couple of posts you might helpful. By including the IV and your moderator, the model estimates the effect of each one while while controlling for the other. You could use partial correlation, but really I’d recommend putting them into a regression model and assessing their effects and significance that way. However, to really know, graph the two variables in a scatterplot and see if the relationship follows a straight line.
![spss code for test of normality spss code for test of normality](https://d2mvzyuse3lwjc.cloudfront.net/doc/en/UserGuide/images/Normality_Test/450px-Normal_PPPlot_2.png)
In that case, you might be able to use regular Pearson’s correlation. If they were individual Likert items, I’d recommend Spearman’s correlation but it sounds like you’re either summing or averaging multiple items together. It gets more tricky if you’re just using individual Likert items for each variable. It sounds like that is what you’re doing. If you’re summing multiple Likert items for each variable, you can often treat them as continuous variables. This error rate should equal the significance level. The test results are statistically significant but, unbeknownst to the investigator, the null hypothesis is actually true. A type I error rate is essentially a false positive.
![spss code for test of normality spss code for test of normality](https://miro.medium.com/max/962/0*dXd1m8Ylozw6tzZa.jpg)
Comparing Error Rates and Power When Analyzing Likert Scale DataĪfter analyzing all pairs of distributions, the results indicate that both types of analyses produce type I error rates that are nearly equal to the target value. The project also looked at different sample sizes to see if that made a difference. Their goal is to calculate the error rates and statistical power of both tests to determine whether one of the analyses is better for Likert data. The study statistically analyzed each pair of samples with both the 2-sample t-test and the Mann-Whitney test. The computer simulation generated independent pairs of random samples that contained all possible combinations of the 14 distributions. The study produced 10,000 random samples for each of the 98 combinations of distributions. The investigators assessed a group of 14 distributions of Likert data that cover the gamut.