I ran a 2x2 ANOVA. My interaction plots show full reversal (i.e., perfectly X-shaped plot) but statistics are non-significant. Am I missing anything?
Plots are helpful in reflecting possible interactions. But, even the x shaped plot doesn't imply a significant interaction.
When lines crossed on the plot or even outside the plot, it confirms a significant interaction, please check your data of lsd calculation!. Regards
Standard plots generated by software like SPSS can sometimes be deceiving because of the default settings (e.g., for the origin/scaling of the Y axis). An effect may seem "huge" in a default plot even when it is actually small. When you edit the plot, it can look very different. Also, depending on your sample size, you may be lacking statistical power to "detect" the effect.
If you can add to the plot the standard errors of the means or the confidence intervals for the means, this may help you see the results graphically. Something like: https://data.library.virginia.edu/files/twoway3-1.png .
In addition do Christian Geiser excellent answer (SPSS is VERY deceiving!), I would add confidence intervals to the plot to get an impressio of the uncertainty (which in turn is a function of your sample size).
Therefore, please ingnore Medhat Elsahookie answer which is wrong or at least misleading.
Rana Hamza Shakil
and again you just copy and paste chat GPT answers, how disingenuous can you be?I uploded your answer to chatGPT and asked, if this was created by chatGPT. Here the answer: "Upon reviewing the answer you provided, I can confirm that it was indeed generated by ChatGPT."
Good idea, Rainer Duesing, asking ChatGPT in that way. However, given that ChatGPT is known to cite non-existent articles, I don't think I would trust it to always tell the truth! ;-)
A significant interaction plot shows a statistically significant relationship between variables, indicating that their effects are not independent but depend on each other. Non-significant statistics suggest that there is no statistically significant relationship or effect between the variables being analyzed.
There is no such thing as a *significant* interaction plot. A plot does not have any p-value associated with it.
You can have hints about significance of the comparison, or the interaction, by adding informations on the plot (confidence intervals, for instance, but comparing confidence intervals does not always lead to the correct interpretation, that should be based on confidence intervals of the differences, and the situation is more complex for categorical variables with more than two levels).
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R