# One way ANOVA v. Univariate v. Multivariate analyses?

Discussion in 'Psychology [Psy.D. / Ph.D.]' started by danzgymn86, 11.30.07.

1. SDN is made possible through member donations, sponsorships, and our volunteers. Learn about SDN's nonprofit mission.
1. ### danzgymn86 2+ Year Member

Joined:
03.18.07
Messages:
132
Okay so this is probably a dumb question but I don't know the answer!

What is the difference between a One-Way ANOVA and a Univariate Analysis? I have all my data in SPSS and was running preliminary tests. I am getting WONDERFUL results with the One-Way ANOVA (like .001, .000, .002) but not so much with the Univariate test (.030, and some others not significant). What exactly is the difference between the two tests?? (My confusion lies in the fact that I thought an ANOVA WAS a univariate analysis?)

I'm also getting a few results with the Multivariate but nothing is giving me anywhere near the results of the ANOVA.

3. ### MarkpClinical Psychologist 7+ Year Member

Joined:
11.19.07
Messages:
2,263
Status:
Psychologist
The GLM univariate analysis and anova should come out the same if you are doing it correctly. However the two analysis are different. The GLM univariate analysis in SPSS is actually a regression.

regression using univariate model:

Tests of Between-Subjects Effects
Dependent Variable: degree
Source Type III Sum of Squares df Mean Square F Sig. Noncent. Parameter Observed Power(a)
Corrected Model .233(b) 1 .233 .264 .616 .264 .076
Intercept 67.433 1 67.433 76.229 .000 76.229 1.000
gender .233 1 .233 .264 .616 .264 .076
Error 11.500 13 .885
Total 80.000 15
Corrected Total 11.733 14
a Computed using alpha = .05
b R Squared = .020 (Adjusted R Squared = -.056)

ANOVA:

ANOVA
degree
Sum of Squares df Mean Square F Sig.
Between Groups .233 1 .233 .264 .616
Within Groups 11.500 13 .885
Total 11.733 14

**************************************

Ok, the output is ugly but notice that both the ANOVA and the Univariate have P= 0.616 and F= 0.264. The Univariate gives you alot more information like Power. Sum of squares and error variance are the same though.

Mark

4. ### danzgymn86 2+ Year Member

Joined:
03.18.07
Messages:
132
Hmmm maybe I am remembering incorrectly. I don't have access to SPSS until Monday, but maybe it's the multivariate that is way different.

So it's good to know that the univariate and 1 way ANOVA should be the same. So now, if they are the same results, which is the better one to report in an academic paper?

5. ### MarkpClinical Psychologist 7+ Year Member

Joined:
11.19.07
Messages:
2,263
Status:
Psychologist
It really depends on the information you need. If you only need the F and P values, then stick with the simple ANOVA. I would suggest that if you don't understand the statistics you are using then stick with the simpler analysis, as you can explain it. If you use statistics you don't understand you are asking for trouble.

Mark

Joined:
10.06.06
Messages:
20,597
Location:
The Beach
Status:
Psychologist
Verified
Faculty

-t

7. ### Therapist4ChngeNeuropsych Ninja FacultyModerator Emeritus 10+ Year Member

Joined:
10.06.06
Messages:
20,597
Location:
The Beach
Status:
Psychologist
Verified
Faculty
This was too funny not to post....

--

Wow...1:17am, and I need to get up in <6 hours for work. If you can try, don't ever work early on Saturdays, it just isn't worth it.

-t

8. ### paramour 7+ Year Member

Joined:
01.16.07
Messages:
1,979
I absolutely lurved this one. I've done the exact same thing entirely too many times over the past years.

9. ### Therapist4ChngeNeuropsych Ninja FacultyModerator Emeritus 10+ Year Member

Joined:
10.06.06
Messages:
20,597
Location:
The Beach
Status:
Psychologist
Verified
Faculty
My last stats class.....in the lab.....with a few of my classmates around. Yup, still remember doing pretty much that exact thing. The rest of the people in the lab were like, "Yeah....great, now stop interrupting my MySpace/Facebook/Gmail time....I've EARNED THIS!!!" I also had this happen in my head, while going over some stuff with a prof. I was really frustrated because I knew WHAT to do, but I was having a problem with explaining the WHY, in addition to not being sure if my post-hoc adjustment was right......then it happened, I got it! Now, I was under no illusion that my prof. knew the exact answer, but when I was able to walk through the entire process and explain it all....it was like I just won Jeopardy!, of course...then I had to go write it up. Though I could see he was happy I got it, he probably was also happy that he could get back to his stuff.

-t

10. ### hants 2+ Year Member

Joined:
08.12.08
Messages:
57
Location:
San Jose
Status:
Pre-Medical
i know this thread is super old, but just in case anybody every looks.

a one-way and a univariate are the same thing. both are based off of a single independent variable.

a multivariate analyses, such as a repeated measures anova (possible GLM) or 2x2 factorial design for instance is when you have multiple independent variables, or dependent variables as with the case with some factorial designs like a 2x2 or 2x4.

11. ### Therapist4ChngeNeuropsych Ninja FacultyModerator Emeritus 10+ Year Member

Joined:
10.06.06
Messages:
20,597
Location:
The Beach
Status:
Psychologist
Verified
Faculty
Thanks for the memories. I still remember feeling awesome that I got through my last stats class. I do miss the prof, as he doesn't teach as much anymore, and the cohorts coming through don't know what they are missing.

12. ### EquineScience

Joined:
08.22.12
Messages:
1
Status:
Psychology Student
Hello Everyone,

This is a very informative topic but I would also like to know what would be the non-parametric equivalent of univariate and multivariate, and how to execute it in SPSS.

Many thanks!
Aspiring Equine Psychologist

13. ### AcronymAllergyNeuropsychologist SDN Moderator 5+ Year Member

Joined:
01.07.10
Messages:
6,573
Status:
Psychologist
Verified
Psychologist
Kruskal-Wallis and Mann-Whitney U are the two that spring immediately to mind. As for how to run them, I believe the SPSS help file and/or a good SPSS book would be your best bets there.