Friedman SPSS

7/27/2019 Friedman SPSS

1/3

One-Way Repeated Measures ANOVA on Ranks

Seethis exampleof the traditional analysis, the Friedman ANOVA. Notice that the data areset up such that all the scores for each block are on one data line.

An alternative analysis involvesranking the scores within blocks and thenconducting an ANOVA on the ranks. Herewe have latency data under threeconditions (1 = baseline, 2 = treatment, 3 =post-treatment). To create the ranks,within blocks, click Transform, RankCases.

Notice that each line has data for one cellthatis, one subject in one condition.

Analyze, General Linear Model, Univariate. The ranked latencies are identified as thedependent variable and Block and Condition as the fixed factors.
http://www.or.vcu.edu/help/SPSS/SPSS.Friedman.pdfhttp://www.or.vcu.edu/help/SPSS/SPSS.Friedman.pdfhttp://www.or.vcu.edu/help/SPSS/SPSS.Friedman.pdfhttp://www.or.vcu.edu/help/SPSS/SPSS.Friedman.pdf


2/3

If you were to click OK now, SPSS would conduct a full factorial analysis, with the effects beingBlocks, Conditions, and the Blocks x Conditions interaction. However, there is only one score in eachBlocks x Conditions cell, so the error term would not be defined. For testing the treatment effect, wewant to use the Blocks x Conditions mean square as the error term. Here is how to make thathappen:

Click Model. Select Custom. Scoot into the Model

pane Block and Condition but not the interaction.By not being included in the model, the interactionwill become the error term.

To conduct pairwise comparisons among

the conditions, simply as for post hoc tests.Since we have only three conditions here, wecan use Fishers procedure (which will holdfamilywise alpha at its nominal level, if theomnibus ANOVA is significant.

Tests of Between-Subjects Effects

Dependent Variable: RLatency

Source Type I Sum

of Squares

df Mean

Square

F Sig.

Corrected

Model37.680a 26 1.449 6.741 .000

Intercept 300.000 1 300.000 1395.349 .000

Block .000 24 .000 .000 1.000

Condition 37.680 2 18.840 87.628 .000

Error 10.320 48 .215

Total 348.000 75

Corrected

Total48.000 74

a. R Squared = .785 (Adjusted R Squared = .669)

Conditions has a significant omnibus effect.


3/3

Multiple Comparisons

Dependent Variable: RLatency

LSD

(I)

Condition

(J)

Condition

Mean

Difference (I-

J)

Std.

Error

Sig. 95% Confidence Interval

Lower

Bound

Upper

Bound

1 2 -1.44000

*

.131149 .000 -1.70369 -1.176313 .12000 .131149 .365 -.14369 .38369

21 1.44000* .131149 .000 1.17631 1.70369

3 1.56000* .131149 .000 1.29631 1.82369

31 -.12000 .131149 .365 -.38369 .14369

2 -1.56000* .131149 .000 -1.82369 -1.29631

Based on observed means.

The error term is Mean Square(Error) = .215.

*. The mean difference is significant at the 0.05 level.

Latencies were significantly longer during treatment than at baseline or after treatment, but thebaseline and post-treatment conditions did not differ significant from each other.

Return to Wuenschs SPSS Lessons Page

Karl L. Wuensch,February, 2013.
http://core.ecu.edu/psyc/wuenschk/SPSS/SPSS-Lessons.htmhttp://core.ecu.edu/psyc/wuenschk/SPSS/SPSS-Lessons.htmhttp://core.ecu.edu/psyc/WuenschK/KLW.htmhttp://core.ecu.edu/psyc/WuenschK/KLW.htmhttp://core.ecu.edu/psyc/WuenschK/KLW.htmhttp://core.ecu.edu/psyc/wuenschk/SPSS/SPSS-Lessons.htm

Documents

Friedman SPSS