Two Way ANOVA In SPSS
-
Upload
sister-edith-bogue -
Category
Technology
-
view
29.913 -
download
17
Transcript of Two Way ANOVA In SPSS
Two-Way ANOVA in SPSS
• Plan your analysis ahead of time
• Select the same dependent variable– Select two factors– Each factor is a nominal (grouping) variable– If both factors have many levels (values), your
analysis will be complex
• SPSS will generate a lot of output.
• Knowing which parts to use is important
Set options to show Variable Names in Alphabetical Order
Edit → Options
Use a boxplot to preview data
Dependent variable
Factor A
Factor B
Edited boxplot. Background changed to white.Text changed to white on outliers (doesn’t show)
Notice that the labels for the variable names cut off in mid-sentence. If you edit the labels in Variable View, you can generate useful labels.
Factorial ANOVAAnalyze→General Linear Model→Univariate
Several steps to prepare:First: choose the variables
DEPENDENT
FACTOR A
FACTOR B
Notice that the buttons for options are vertically
arranged along the side.
Second, choose Options
In the Options box …
• Request Means for Factors
•Request Means for Interaction
•Request descriptive statistics
•Request estimates of effect size (This will generate an estimate of η2)
•Click CONTINUE
•In main box, Click PLOTS
Get a Plot of the Means of Groups•Put Factor with the fewest categories into the box titled “Separate Lines”
•Put other Factor into “Horizontal Axis”
•Click on ADD to move variable names into the Plots listing box.
•Click CONTINUE to return to main dialogue box.
•Click POST HOC
Choose Post-Hoc Tests•Post-hoc tests are only needed if a Factor has a significant effect. You can request them now and decide later whether to use them.
•Post-hoc tests are only needed if a factor has a significant effect AND if it has 3 or more groups
•In this data, pricegas has 3 levels.Move pricegas to “Post hoc tests for”
•Choose Scheffé test because groups are not equal size.
•Note that Bonferroni (conservative) is also available.
•Click CONTINUE for main dialogue box
Click OK in the lower left corner
Means of all groupsTotal = row or column means
Descriptive Statistics
Dependent Variable: econgood On another subject, would you describe the state of the nation'seconomy these days as excellent, good, not so good or p
3.10 .718 195
3.14 .739 109
3.12 .724 304
2.81 .751 206
2.83 .830 107
2.82 .778 313
2.47 .713 249
2.57 .762 133
2.51 .731 382
2.77 .771 650
2.83 .809 349
2.79 .785 999
employed Are youyourself employedoutside the home, or not?1 Yes
2 No
Total
1 Yes
2 No
Total
1 Yes
2 No
Total
1 Yes
2 No
Total
pricegas Have recentprice increases ingasoline caused anyfinancial hardship foryou or others in yourhousehold, or not? [If y
1 Yes, serious
2 Yes, not serious
3 No
Total
Mean Std. Deviation N
Hypothesis Test Results
Tests of Between-Subjects Effects
Dependent Variable: econgood On another subject, would you describe the state of the nation's economythese days as excellent, good, not so good or p
64.366a 5 12.873 23.239 .000 .105
7153.781 1 7153.781 12914.219 .000 .929
55.839 2 27.920 50.402 .000 .092
.621 1 .621 1.121 .290 .001
.296 2 .148 .267 .765 .001
550.068 993 .554
8384.000 999
614.434 998
SourceCorrected Model
Intercept
pricegas
employed
pricegas * employed
Error
Total
Corrected Total
Type III Sumof Squares df Mean Square F Sig.
Partial EtaSquared
R Squared = .105 (Adjusted R Squared = .100)a.
Factor A
Factor B
Interaction
p-value η2
Post-hoc TestsAll groups of pricegas differ from each other significantly
Multiple Comparisons
Dependent Variable: econgood On another subject, would you describe the state of the nation's economy these days asexcellent, good, not so good or p
Scheffe
.30* .060 .000 .15 .44
.61* .057 .000 .47 .75
-.30* .060 .000 -.44 -.15
.31* .057 .000 .17 .45
-.61* .057 .000 -.75 -.47
-.31* .057 .000 -.45 -.17
(J) Have recent priceincreases in gasolinecaused any financialhardship for you orothers in yourhousehold, or not? [If y2 Yes, not serious
3 No
1 Yes, serious
3 No
1 Yes, serious
2 Yes, not serious
(I) Have recent priceincreases in gasolinecaused any financialhardship for you orothers in yourhousehold, or not? [If y1 Yes, serious
2 Yes, not serious
3 No
MeanDifference
(I-J) Std. Error Sig. Lower Bound Upper Bound
95% Confidence Interval
Based on observed means.
The mean difference is significant at the .05 level.*.
Means Plot, Edited
•Background made white
•Title on horizontal axis edited for clarity
•Chart title deleted (APA format has Figure title below the chart)
•Legend moved inside the box for clarity
Write-up• Sandwich format• Report three hypothesis tests with F and p-value
for each:– Factor A– Factor B– Interaction
• For any significant element, report means and effect size
• For any significant element with 3 or more groups, report post-hoc tests
• Close with a summary statement• Refer to boxplot or means plot as Figure 3 (or
whichever number) in the write-up.