Glad it helped!

Thanks, I'm glad it helped!

Hi, thanks for your question. In APA style, it would be p = .707 (without a 0 before the decimal point because p can't exceed 1, and with an italic p)

@@DavidRobinsonPhD Thank you so much! It helped a lot ❤

Hi, thanks for your question. The median is just the middle value when the values are ordered, so it may be that more than half of your values are 0. E.g., the median would be 0 in the following as the 4th value (i.e., the value in the middle) is 0: 0, 0, 0, 0, 1, 2, 3.

@ thank you for the clarification! In this case, would it be ok for me to compare means instead even though I’m using a non-parametric test

@@sitech-x3r The means might not provide a good reflection of the data if, e.g., you don't have normal distributions. However, you could consider comparing groups in terms of their mean ranks.

@ thank you!

Hi, thanks for your question. I don't think there's a way of telling Jamovi that you have a one-tailed hypothesis. However, you can convert a two-tailed p value into a one-tailed p value by dividing it by 2.

@ thanks for the clarification! I’m assuming the p-value calculated by jamovi is already the two-tailed one?

@@sitech-x3r Yes, that's right.

@ thanks for the clarification!

Hi Inayah, thanks for your question. There isn't a universally accepted minimum sample size. Technically, the test could be completed with just four observations, though larger samples are preferable as you increase the likelihood of detecting differences.

@DavidRobinsonPhD thank you heaps!

Thanks!

Hi, analyses based on questionnaire data usually require that you calculate sum or mean scores. I.e., rather than doing analyses on specific items, it's more common to do an analysis on a variable that you create by adding up all of the scores for the individual items or by finding the mean of the scores for the individual items. Therefore, if you needed to remove an item, you could just not include it when creating this variable.

Hi Leon, glad you found the video helpful and I hope the book is too if you check it out! Perhaps you could create a new binary variable for hours (e.g., 0 = 0-2 and 1 = 3-5) and then run the analysis again to see whether it has a significant effect on recovery.

Thanks, I'll add that to my list!

@@DavidRobinsonPhD 🤩😍🤩

Cohen's d is usually used with parametric tests (e.g., t-tests) rather than with non-parametric tests (e.g., the Wilcoxon test) as it's based on means, which may not be informative when data are not normally distributed.

Thanks, I'm glad it helped!

Hi, thanks for your question. Typically, we interpret the odds of the "event" occurring (like "completing a course") because that's often the behaviour or outcome of primary interest. However, you can interpret the odds of the opposite event (e.g., "not completing a course") if that aligns better with your research focus or if it's more meaningful for the context. The choice depends on which outcome provides clearer insights for your audience or is most relevant to your study. For example, if you're studying factors that contribute to course dropout rates, it might make more sense to interpret the odds of not completing the course. Conversely, if you're looking at predictors of successful completion, you’d focus on the odds of completing it.

@@DavidRobinsonPhD Thanks for prompt response, highly appreciated. I got it but is there any additional step on spss to make the result represent the likehood of "not"?

@@adlesal24 You could consider recoding your dependent variable (e.g., 0 becomes 1 and 1 becomes 0).

@@DavidRobinsonPhD thank you, i got it.

Thank you for watching the video and for your question! When SPSS shows a dot (.) in the "Sig." value, it usually means there’s insufficient data in that specific group to calculate the significance level, often due to a small sample size. This means the normality test for that group couldn’t be performed, so the result is indeterminate rather than a confirmed outcome. As for the lower bound of true significance, this can depend on the sample size and the test used. Generally, for tests of normality (like Shapiro-Wilk or Kolmogorov-Smirnov), a "Sig." (p-value) of 0.200 or greater suggests that the data do not significantly deviate from a normal distribution, indicating normality. So, yes, if a group has a significance value of 0.200 or higher, it’s usually considered approximately normally distributed.

Thanks!

Thanks, glad it helped!