Analysis of Covariance (ANCOVA) - SPSS (part 2)
Vložit
- čas přidán 28. 08. 2024
- I demonstrate how to perform an analysis of covariance (ANCOVA) in SPSS. The first part of the series is relevant to the ANCOVA tested through the conventional approach to doing so by getting SPSS to estimate adjusted means through the GLM univariate utility. In the second part of the series, I demonstrate the exact correspondence between ANCOVA and multiple regression.
NB: The results of the analysis in this series found that males appear to have larger cranial capacities than females, even after controlling for the effects of body size. However, it should be important to emphasize that research has found that there are little to no general mean differences in IQ between males and females. Furthermore, there is neuroanatomical research to suggest that female brains appear to have more neurons per cubic cm than male brains. Thus, the difference in cranial capacity/brain size between the sexes may be counteracted by the differences in neuronal density.
Have I told you lately that I LOVE YOU! Thanks for these wonderful tutorials. Very clear and easy to understand. What a great refresher.
Statistics in plain English! Thank you for explaining so clearly rather than using your video as a showcase for your statistical knowledge as some other stats videos do!
Can we please have closed captioning on the video enabled? The automatic transcript option is not showing up for some reason....
Can we get the real data so that we can do the analysis after watching the videos (parts1-4)?
Thanks for so clear explanations!
Why did you assess the homogeneity of variance? I thought you could only use it when comparing three or more groups?
Definitely not the case. Homogeneity of variance can and should be assessed even with just two groups. In fact, SPSS tests homogeneity of variance automatically in the independent t-test case.
@@how2stats What would you do if this assumption is not met and the Levene's test p < .000? I'm trying to assess some differences between smokers and non-smokers with big N differences (65 smokers and 586 non-smokers).
Statistics is the language of love.
I mean not sure if you have interpreted eta partial squared correctly in this video. Otherwise, the video is very helpful. Thanks
sorry, it is ok, because in a case of one-way ANOVA partial or full eta squared would be the same! :-)
When would I not want to create a composite score? I tried to use the dimension reduction technique to create a composite score for three demographic variables. Is that wrong?
It is probably wrong. You should consider creating a composite score when the variables are inter-correlated positively; ideally, they are inter-correlated positively sufficiently to yield a coefficient (Cronbach's) alpha of .70 or greater.
@@how2stats Thank you so much for your help!
my teacher, a little help here. and if my levene test is p < 0,05 ?
Are your sample sizes equal? If not, which sample has the biggest variance?
Thank you so much for this tutorial sir. Where can we access the dataset for our practice. thank you.
I only make available the data files for my textbook available here: www.how2statsbook.com
you're great
If I want to know if the respondent's sexuality has an effect on the decline of brand attitude between a pre-test and a post-test, would ANCOVA be a good option? I did a repeated measures ANOVA to see if my between-subject factor had an effect on the difference between my within-subject factors but now I still need to see if variables like degree and sexuality can be the cause of this.
Not sure if this is clear but I thought I'd give it a go.
Sounds like a mediation analysis. I don't think I have a video on that, yet, but it will be in my upcoming textbook.
Thank you for your answer. Is there any video on this analysis you can suggest?
Also, I now put 'degree', 'sexual orientation' and 'attraction to model' in the 'covariate' box to see if these variables covariate with the changing brand attitude. Do you think this is a plausible solution for my question? I can now see in the output that for example 'attraction to model' is a factor that has an influence on the relation between the dependent and independent variable.
what lets do and PCA? This is getting advance.