HELP ME, please... How to introduce a set of data and perform Kruskal-Wallis
carista maldeño
statisics. Now thinks changed and I am fighting with a sample of data
and cannot find the solucio... Could somebody tell me how to perform
the Kruskal–Wallis test with the following set of data. I even cannot
introduce them into SPSS, e.g. instead of S10, S11, etc. appear S1...
I hope somebody can help me... I would like to find out if there is
any significant difference witin th variables an between them. The
answers are on a Likert scale:
0 - incorrect answer, 1 - "less incorrect" answer, 0 - I don´t know,
1- "less correct" answer, 2 - corret answer
ID Group Item Condition Answer
S1 G1 P1 C1 4
S1 G1 P2 C1 4
S1 G1 P3 C2 4
S1 G1 P4 C2 3
S1 G1 P5 C2 0
S1 G1 P6 C3 0
S1 G1 P7 C3 1
S1 G1 P8 C3 0
S1 G1 P9 C4 0
S1 G1 P10 C4 0
S1 G1 P11 C5 0
S1 G1 P12 C5 0
S2 G1 P1 C1 4
S2 G1 P2 C1 4
S2 G1 P3 C2 4
S2 G1 P4 C2 4
S2 G1 P5 C2 1
S2 G1 P6 C3 0
S2 G1 P7 C3 1
S2 G1 P8 C3 0
S2 G1 P9 C4 0
S2 G1 P10 C4 1
S2 G1 P11 C5 0
S2 G1 P12 C5 1
S3 G1 P1 C1 4
S3 G1 P2 C1 3
S3 G1 P3 C2 4
S3 G1 P4 C2 4
S3 G1 P5 C2 4
S3 G1 P6 C3 1
S3 G1 P7 C3 0
S3 G1 P8 C3 0
S3 G1 P9 C4 1
S3 G1 P10 C4 1
S3 G1 P11 C5 4
S3 G1 P12 C5 3
S4 G1 P1 C1 4
S4 G1 P2 C1 3
S4 G1 P3 C2 4
S4 G1 P4 C2 3
S4 G1 P5 C2 4
S4 G1 P6 C3 0
S4 G1 P7 C3 1
S4 G1 P8 C3 1
S4 G1 P9 C4 0
S4 G1 P10 C4 1
S4 G1 P11 C5 0
S4 G1 P12 C5 1
S5 G1 P1 C1 3
S5 G1 P2 C1 4
S5 G1 P3 C2 1
S5 G1 P4 C2 3
S5 G1 P5 C2 3
S5 G1 P6 C3 0
S5 G1 P7 C3 0
S5 G1 P8 C3 0
S5 G1 P9 C4 0
S5 G1 P10 C4 0
S5 G1 P11 C5 3
S5 G1 P12 C5 1
S6 G1 P1 C1 3
S6 G1 P2 C1 4
S6 G1 P3 C2 4
S6 G1 P4 C2 4
S6 G1 P5 C2 3
S6 G1 P6 C3 0
S6 G1 P7 C3 0
S6 G1 P8 C3 1
S6 G1 P9 C4 1
S6 G1 P10 C4 1
S6 G1 P11 C5 0
S6 G1 P12 C5 1
S7 G1 P1 C1 4
S7 G1 P2 C1 4
S7 G1 P3 C2 3
S7 G1 P4 C2 1
S7 G1 P5 C2 2
S7 G1 P6 C3 0
S7 G1 P7 C3 0
S7 G1 P8 C3 0
S7 G1 P9 C4 0
S7 G1 P10 C4 1
S7 G1 P11 C5 1
S7 G1 P12 C5 0
S8 G1 P1 C1 4
S8 G1 P2 C1 3
S8 G1 P3 C2 4
S8 G1 P4 C2 4
S8 G1 P5 C2 4
S8 G1 P6 C3 1
S8 G1 P7 C3 0
S8 G1 P8 C3 0
S8 G1 P9 C4 3
S8 G1 P10 C4 4
S8 G1 P11 C5 0
S8 G1 P12 C5 4
S9 G1 P1 C1 4
S9 G1 P2 C1 4
S9 G1 P3 C2 4
S9 G1 P4 C2 3
S9 G1 P5 C2 3
S9 G1 P6 C3 3
S9 G1 P7 C3 0
S9 G1 P8 C3 1
S9 G1 P9 C4 1
S9 G1 P10 C4 0
S9 G1 P11 C5 1
S9 G1 P12 C5 0
S10 G1 P1 C1 4
S10 G1 P2 C1 4
S10 G1 P3 C2 3
S10 G1 P4 C2 4
S10 G1 P5 C2 3
S10 G1 P6 C3 0
S10 G1 P7 C3 0
S10 G1 P8 C3 0
S10 G1 P9 C4 0
S10 G1 P10 C4 1
S10 G1 P11 C5 4
S10 G1 P12 C5 4
S11 G2 P1 C1 4
S11 G2 P2 C1 4
S11 G2 P3 C2 4
S11 G2 P4 C2 3
S11 G2 P5 C2 1
S11 G2 P6 C3 2
S11 G2 P7 C3 0
S11 G2 P8 C3 0
S11 G2 P9 C4 0
S11 G2 P10 C4 1
S11 G2 P11 C5 3
S11 G2 P12 C5 0
S12 G2 P1 C1 4
S12 G2 P2 C1 4
S12 G2 P3 C2 4
S12 G2 P4 C2 4
S12 G2 P5 C2 3
S12 G2 P6 C3 3
S12 G2 P7 C3 4
S12 G2 P8 C3 3
S12 G2 P9 C4 1
S12 G2 P10 C4 0
S12 G2 P11 C5 4
S12 G2 P12 C5 3
S13 G2 P1 C1 4
S13 G2 P2 C1 4
S13 G2 P3 C2 4
S13 G2 P4 C2 4
S13 G2 P5 C2 3
S13 G2 P6 C3 0
S13 G2 P7 C3 4
S13 G2 P8 C3 3
S13 G2 P9 C4 0
S13 G2 P10 C4 3
S13 G2 P11 C5 0
S13 G2 P12 C5 0
S14 G2 P1 C1 4
S14 G2 P2 C1 3
S14 G2 P3 C2 2
S14 G2 P4 C2 3
S14 G2 P5 C2 3
S14 G2 P6 C3 0
S14 G2 P7 C3 1
S14 G2 P8 C3 0
S14 G2 P9 C4 3
S14 G2 P10 C4 4
S14 G2 P11 C5 0
S14 G2 P12 C5 1
S15 G2 P1 C1 4
S15 G2 P2 C1 4
S15 G2 P3 C2 3
S15 G2 P4 C2 4
S15 G2 P5 C2 3
S15 G2 P6 C3 0
S15 G2 P7 C3 0
S15 G2 P8 C3 0
S15 G2 P9 C4 0
S15 G2 P10 C4 3
S15 G2 P11 C5 1
S15 G2 P12 C5 0
S16 G2 P1 C1 4
S16 G2 P2 C1 4
S16 G2 P3 C2 3
S16 G2 P4 C2 2
S16 G2 P5 C2 3
S16 G2 P6 C3 0
S16 G2 P7 C3 0
S16 G2 P8 C3 1
S16 G2 P9 C4 4
S16 G2 P10 C4 3
S16 G2 P11 C5 4
S16 G2 P12 C5 4
S17 G2 P1 C1 4
S17 G2 P2 C1 4
S17 G2 P3 C2 4
S17 G2 P4 C2 3
S17 G2 P5 C2 4
S17 G2 P6 C3 0
S17 G2 P7 C3 1
S17 G2 P8 C3 1
S17 G2 P9 C4 3
S17 G2 P10 C4 1
S17 G2 P11 C5 3
S17 G2 P12 C5 1
S18 G2 P1 C1 4
S18 G2 P2 C1 3
S18 G2 P3 C2 3
S18 G2 P4 C2 3
S18 G2 P5 C2 1
S18 G2 P6 C3 0
S18 G2 P7 C3 0
S18 G2 P8 C3 0
S18 G2 P9 C4 1
S18 G2 P10 C4 3
S18 G2 P11 C5 4
S18 G2 P12 C5 4
S19 G2 P1 C1 4
S19 G2 P2 C1 4
S19 G2 P3 C2 4
S19 G2 P4 C2 4
S19 G2 P5 C2 4
S19 G2 P6 C3 0
S19 G2 P7 C3 0
S19 G2 P8 C3 0
S19 G2 P9 C4 0
S19 G2 P10 C4 1
S19 G2 P11 C5 4
S19 G2 P12 C5 4
S20 G2 P1 C1 4
S20 G2 P2 C1 0
S20 G2 P3 C2 4
S20 G2 P4 C2 4
S20 G2 P5 C2 4
S20 G2 P6 C3 1
S20 G2 P7 C3 0
S20 G2 P8 C3 1
S20 G2 P9 C4 0
S20 G2 P10 C4 3
S20 G2 P11 C5 2
S20 G2 P12 C5 4
S21 G3 P1 C1 4
S21 G3 P2 C1 4
S21 G3 P3 C2 4
S21 G3 P4 C2 4
S21 G3 P5 C2 3
S21 G3 P6 C3 1
S21 G3 P7 C3 0
S21 G3 P8 C3 0
S21 G3 P9 C4 4
S21 G3 P10 C4 4
S21 G3 P11 C5 0
S21 G3 P12 C5 0
S22 G3 P1 C1 4
S22 G3 P2 C1 4
S22 G3 P3 C2 4
S22 G3 P4 C2 1
S22 G3 P5 C2 0
S22 G3 P6 C3 1
S22 G3 P7 C3 0
S22 G3 P8 C3 0
S22 G3 P9 C4 1
S22 G3 P10 C4 2
S22 G3 P11 C5 4
S22 G3 P12 C5 4
S23 G3 P1 C1 4
S23 G3 P2 C1 4
S23 G3 P3 C2 4
S23 G3 P4 C2 4
S23 G3 P5 C2 4
S23 G3 P6 C3 0
S23 G3 P7 C3 0
S23 G3 P8 C3 1
S23 G3 P9 C4 1
S23 G3 P10 C4 0
S23 G3 P11 C5 1
S23 G3 P12 C5 1
S24 G3 P1 C1 4
S24 G3 P2 C1 4
S24 G3 P3 C2 3
S24 G3 P4 C2 4
S24 G3 P5 C2 1
S24 G3 P6 C3 1
S24 G3 P7 C3 0
S24 G3 P8 C3 0
S24 G3 P9 C4 3
S24 G3 P10 C4 4
S24 G3 P11 C5 3
S24 G3 P12 C5 3
S25 G3 P1 C1 4
S25 G3 P2 C1 4
S25 G3 P3 C2 4
S25 G3 P4 C2 4
S25 G3 P5 C2 4
S25 G3 P6 C3 0
S25 G3 P7 C3 0
S25 G3 P8 C3 0
S25 G3 P9 C4 4
S25 G3 P10 C4 3
S25 G3 P11 C5 4
S25 G3 P12 C5 0
S26 G3 P1 C1 4
S26 G3 P2 C1 4
S26 G3 P3 C2 3
S26 G3 P4 C2 4
S26 G3 P5 C2 4
S26 G3 P6 C3 1
S26 G3 P7 C3 0
S26 G3 P8 C3 1
S26 G3 P9 C4 0
S26 G3 P10 C4 2
S26 G3 P11 C5 4
S26 G3 P12 C5 4
S27 G3 P1 C1 4
S27 G3 P2 C1 4
S27 G3 P3 C2 4
S27 G3 P4 C2 3
S27 G3 P5 C2 4
S27 G3 P6 C3 0
S27 G3 P7 C3 0
S27 G3 P8 C3 1
S27 G3 P9 C4 3
S27 G3 P10 C4 3
S27 G3 P11 C5 4
S27 G3 P12 C5 3
S28 G3 P1 C1 4
S28 G3 P2 C1 4
S28 G3 P3 C2 3
S28 G3 P4 C2 4
S28 G3 P5 C2 4
S28 G3 P6 C3 0
S28 G3 P7 C3 0
S28 G3 P8 C3 0
S28 G3 P9 C4 0
S28 G3 P10 C4 0
S28 G3 P11 C5 3
S28 G3 P12 C5 4
S29 G3 P1 C1 4
S29 G3 P2 C1 4
S29 G3 P3 C2 4
S29 G3 P4 C2 3
S29 G3 P5 C2 4
S29 G3 P6 C3 0
S29 G3 P7 C3 4
S29 G3 P8 C3 3
S29 G3 P9 C4 0
S29 G3 P10 C4 4
S29 G3 P11 C5 4
S29 G3 P12 C5 4
S30 G3 P1 C1 4
S30 G3 P2 C1 4
S30 G3 P3 C2 4
S30 G3 P4 C2 3
S30 G3 P5 C2 3
S30 G3 P6 C3 0
S30 G3 P7 C3 0
S30 G3 P8 C3 1
S30 G3 P9 C4 0
S30 G3 P10 C4 3
S30 G3 P11 C5 4
S30 G3 P12 C5 4
S31 G4 P1 C1 4
S31 G4 P2 C1 4
S31 G4 P3 C2 4
S31 G4 P4 C2 4
S31 G4 P5 C2 4
S31 G4 P6 C3 4
S31 G4 P7 C3 4
S31 G4 P8 C3 3
S31 G4 P9 C4 4
S31 G4 P10 C4 4
S31 G4 P11 C5 4
S31 G4 P12 C5 4
S32 G4 P1 C1 4
S32 G4 P2 C1 4
S32 G4 P3 C2 3
S32 G4 P4 C2 4
S32 G4 P5 C2 4
S32 G4 P6 C3 4
S32 G4 P7 C3 4
S32 G4 P8 C3 3
S32 G4 P9 C4 3
S32 G4 P10 C4 4
S32 G4 P11 C5 4
S32 G4 P12 C5 4
S33 G4 P1 C1 4
S33 G4 P2 C1 4
S33 G4 P3 C2 4
S33 G4 P4 C2 4
S33 G4 P5 C2 4
S33 G4 P6 C3 4
S33 G4 P7 C3 4
S33 G4 P8 C3 4
S33 G4 P9 C4 3
S33 G4 P10 C4 1
S33 G4 P11 C5 4
S33 G4 P12 C5 4
S34 G4 P1 C1 4
S34 G4 P2 C1 4
S34 G4 P3 C2 3
S34 G4 P4 C2 4
S34 G4 P5 C2 4
S34 G4 P6 C3 4
S34 G4 P7 C3 4
S34 G4 P8 C3 4
S34 G4 P9 C4 3
S34 G4 P10 C4 4
S34 G4 P11 C5 4
S34 G4 P12 C5 4
S35 G4 P1 C1 4
S35 G4 P2 C1 4
S35 G4 P3 C2 4
S35 G4 P4 C2 4
S35 G4 P5 C2 4
S35 G4 P6 C3 3
S35 G4 P7 C3 3
S35 G4 P8 C3 4
S35 G4 P9 C4 3
S35 G4 P10 C4 4
S35 G4 P11 C5 4
S35 G4 P12 C5 4
S36 G4 P1 C1 4
S36 G4 P2 C1 4
S36 G4 P3 C2 4
S36 G4 P4 C2 4
S36 G4 P5 C2 4
S36 G4 P6 C3 4
S36 G4 P7 C3 4
S36 G4 P8 C3 4
S36 G4 P9 C4 3
S36 G4 P10 C4 4
S36 G4 P11 C5 4
S36 G4 P12 C5 4
S37 G4 P1 C1 4
S37 G4 P2 C1 4
S37 G4 P3 C2 4
S37 G4 P4 C2 4
S37 G4 P5 C2 4
S37 G4 P6 C3 4
S37 G4 P7 C3 4
S37 G4 P8 C3 3
S37 G4 P9 C4 3
S37 G4 P10 C4 4
S37 G4 P11 C5 4
S37 G4 P12 C5 4
S38 G4 P1 C1 4
S38 G4 P2 C1 4
S38 G4 P3 C2 4
S38 G4 P4 C2 4
S38 G4 P5 C2 4
S38 G4 P6 C3 3
S38 G4 P7 C3 3
S38 G4 P8 C3 4
S38 G4 P9 C4 4
S38 G4 P10 C4 3
S38 G4 P11 C5 4
S38 G4 P12 C5 4
S39 G4 P1 C1 4
S39 G4 P2 C1 4
S39 G4 P3 C2 4
S39 G4 P4 C2 4
S39 G4 P5 C2 4
3S9 G4 P6 C3 4
S39 G4 P7 C3 4
S39 G4 P8 C3 4
S39 G4 P9 C4 4
S39 G4 P10 C4 2
S39 G4 P11 C5 4
S39 G4 P12 C5 4
S40 G4 P1 C1 4
S40 G4 P2 C1 4
S40 G4 P3 C2 4
S40 G4 P4 C2 4
S40 G4 P5 C2 3
S40 G4 P6 C3 4
S40 G4 P7 C3 4
S40 G4 P8 C3 4
S40 G4 P9 C4 4
S40 G4 P10 C4 3
S40 G4 P11 C5 4
S40 G4 P12 C5 4
Ray Koopman
> Hi! I have never used SPSS before because I didn´t have to make
> statisics. Now thinks changed and I am fighting with a sample of
> data and cannot find the solucio... Could somebody tell me how to
> perform the Kruskal–Wallis test with the following set of data.
> I even cannot introduce them into SPSS, e.g. instead of S10, S11,
> etc. appear S1... I hope somebody can help me... I would like to
> find out if there is any significant difference witin th variables
> an between them. The answers are on a Likert scale:
> 0 - incorrect answer, 1 - "less incorrect" answer, 0 - I don't know,
> 1- "less correct" answer, 2 - corret answer
Do you mean: 0 - incorrect answer, 1 - "less incorrect" answer,
2 - I don't know, 3 - "less correct" answer, 4 - correct answer ?
The Kruskal-Wallis test is for data collected in a simple one-way
design, but your design is more complicated. You have two crossed
nests: subjects within groups, crossed with items within conditions;
a further complication is that different conditions have different
numbers of items (just as different groups could, but do not, have
different numbers of subjects). An analysis of variance is possible,
but I do not know how to tell SPSS what the design is.
[... data deleted; see the original post ...]
Bruce Weaver
Nesting can be indicated in two ways in SPSS:
1. ID within group
item within condition
2. ID(group)
item(condition)
So I think you could do something like the following:
UNIANOVA Answer BY Group Item Condition ID
/RANDOM=ID
/EMMEANS=TABLES(Group)
/EMMEANS=TABLES(Condition)
/EMMEANS=TABLES(Group*Condition)
/DESIGN=Group ID within group
condition item within condition
group*condition group*item within condition .
Or equivalently:
UNIANOVA Answer BY Group Item Condition ID
/RANDOM=ID
/EMMEANS=TABLES(Group)
/EMMEANS=TABLES(Condition)
/EMMEANS=TABLES(Group*Condition)
/DESIGN=Group ID(group)
condition item(condition)
group*condition group*item(condition) .
Am I missing any terms, Ray? Bear in mind that one has to omit the
last error term, because SPSS will try to compute a residual, so if
all terms (including the last error term) are included, the residual
will equal 0, and the model won't run.
There's probably a way to run this model via the MIXED procedure too
(i.e., as a multilevel model).
GLM Repeated Measures might work too, but the data would have to be
restructured to have one row per person. However, I think the
different number of items per condition might make that approach
problematic.
--
Bruce Weaver
bwe...@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/Home
"When all else fails, RTFM."
Ray Koopman
> "When all else fails, RTFM."
Not having used SPSS since the maroon-manual days, I'm in no position
to say anything about the code.
I organize the sources of variance in crossed-nest designs by making a
table with one nest pair as row labels and the other as column labels.
The interactions are at the intersections.
In the present case the nest pairs are G, S|G and C, I|C,
where G = group, S = subject, C = condition, I = item.
It's arbitrary which pair goes down the side and which goes across
the top. (For completeness, I put the grand mean µ in the corner.)
µ C I|C
G GC GI|C
S|G SC|G SI|GC
I assume G and C are fixed, and S is random. If I is fixed then each
term involving S is the error terms for those above it in the table.
If I is random then things are messier. For instance, one form of the
test of G would be F' = MS(G)/[MS(S|G) + MS(GI|C) - MS(SI|GC)].
Another would be F" = [MS(G) + MS(SI|GC)]/[MS(S|G) + MS(GI|C)].
In both cases the degrees of freedom would be found using the Welch-
Satterthwaite approximation. See your favorite text that covers
combined mean squares and quasi-Fs.
Bruce Weaver
OK, so I omitted SC|G term, or in SPSS syntax, condition*ID(group).
Here's the syntax with that term included:
UNIANOVA Answer BY Group Item Condition ID
/RANDOM=ID
/EMMEANS=TABLES(Group)
/EMMEANS=TABLES(Condition)
/EMMEANS=TABLES(Group*Condition)
/DESIGN=Group ID(group)
condition item(condition) condition*ID(group)
group*condition group*item(condition) .
The SI|GC term is omitted, and shows up as the residual.
--
Bruce Weaver
bwe...@lakeheadu.ca