comparing several means: one-way anova
DESCRIPTION
Comparing Several Means: One-way ANOVA. Lesson 15. Analysis of Variance. or ANOVA Comparing 2 or more treatments i.e., groups Simultaneously H 0 : m 1 = m 2 = m 3 … H 1 : at least one population different from others ~. Experimentwise Error. Why can’t we just use t tests? - PowerPoint PPT PresentationTRANSCRIPT
Comparing Several Means:
One-way ANOVALesson 15
Analysis of Variance
or ANOVA Comparing 2 or more treatments
i.e., groups Simultaneously
H0: 1 = 2 = 3 … H1: at least one population different
from others ~
Experimentwise Error
Why can’t we just use t tests? Type 1 error: incorrectly rejecting H0 each comparison = .05
Experimentwise probability of type 1 error P (1 or more Type 1 errors) ~
Experimentwise Error
H0: 1 = 2 = 3
Approximate experimentwise errorH0: 1 = 2 = .05H0: 1 = 3 = .05H0: 2 = 3 = .05
experimentwise .15 ANOVA: only one H0
= .05 (or level you select) ~
Analysis of Variance: Terminology Factor
independent variable Single-Factor Design (One-way)
single independent variable with 2 or more levels
levels: values of independent variable ~
Analysis of Variance: Terminology Repeated Measures ANOVA
Same logic as paired t test Factorial Design
More than one independent variable Life is complex: interactions
Mixed Factorial Design At least 1 between-groups & within
groups variable Focus on independent-measures ~
e.g., Effects of caffeine on reaction timeSingle-factor design
with 3 levels
Caffeine dose
0 mg 50 mg 100 mg
1 rt M 2 rt M 3 rt M
3 x 2 Factorial design
Sexmale
female
0 mg 50 mg 100 mg
1 rt M 2 rt M 3 rt M
4 rt M 5 rt M 6 rt M
Test Statistic F ratio
ratio of 2 variances
(error) chanceby expected difference
means samplebetween differencet
(error) chance by expected e)(differenc variancemeans sample between es)(differenc variance
F
same concept as t tests
F = t2 Only 2 groups ~
F ratio MS: mean squared deviations = variance MSB = MS between treatments
Textbook: MSM
Average distance b/n sample means MSW = MS within treatments
Textbook: MSR differences between individuals same as s2
pooled ~
Logic of ANOVA
Differences b/n groups (means) bigger than difference between individuals?
If H0 false then distance between groups should
be larger ~
W
B
MSMSF
Partitioning SS SST = total sums of squares
total variability SSB = between-treatments sums of squares
variability between groups SSW = within-treatments sums of squares
variability between individuals
WBT SSSSSS T
B
SSSSR 2
* % variance explained by IV
Calculating SS
TSS 2)( Grandi XX
BSS
WSS 2)( ki XX
2)( Grandk XX
Calculating MS
dfSSMS
Calculating MSW Same as s2
pooled for > 2 samples
WW df
SSSSSSMS 321 kNdfW
Calculating MSB
B
BB df
SSMS 1kdfB
SPSS One-way ANOVA Menu
Analyze Compare Means One-way ANOVA
Dialog box Dependent List (DV) Factor (IV) Options:
Descriptives, Homogeneity of Variance Post Hoc ~
Interpreting ANOVA Reject H0
at least one sample different from others do not know which one(s)
Must use post hoc tests Post hoc: after the fact ONLY if rejected H0 for ANOVA
Many post hoc tests Differ on how conservative ~
Post Hoc Test: Pairwise comparisons Adjusted levels LSD (Least Significant Difference)
Basically t-test, no adjustment Tukey’s HSD
Similar logic to t – test Scheffe Test
F test with only 2 groups Differ on how conservative
More conservative bigger difference required ~
Detour Learning Task
Prenatal exposure to methamphetamine effects on learning?
FIGURE 1Males
0
50
100
150
200
250
300
350
1 2 3 4
Detour Learning Trial
Mea
n La
tenc
y to
Soc
ial C
onta
ct Strangers
Cagemates