Академический Документы
Профессиональный Документы
Культура Документы
Ease of implementation
variables It is also the most abuses/misused analysis Arbitrary categorization of variables leads to lower correlations and in general missed/inaccurate effects Nonexperimental designs can make it difficult to interpret In nonexperimental design it is typically an oversimplification of the research problem
Allows for analysis of several groups at once Allows analysis of interactions of multiple grouping
In the past anyway, now one can do much more complex and computationally extensive analyses with the same amount of effort
As before, if the assumptions of the test are not met we may have problems in its interpretation The usual suspects:
Independence of observations Normally distributed variables of measure1 Homogeneity of Variance
H0: 1= 2 = k H1: not H0 ANOVA will tell us that the means are different in some way
As the assumptions specify that the shape and
dispersion should be equal, the only way left to differ is in terms of means
ANOVA allows for an easy way to look at interactions Probability of type one error goes up with multiple tests
For independent contrasts Probability .95 of not making type 1:
.95*.95*.95 = .857 1-.857 = .143 probability of making type 1 error1 So probability of type I error = 1 - (1-)c
Note: some note that each analysis could be taken as separate and independent of all others
With an independent samples t-test we looked to see if two groups had different means
t= X1 X 2 s2 p n1 + s2 p n2
With this formula we found the difference between the groups and divided by the variability within the groups by calculation of their respective variances which we then pooled together.
In this sense with our t-statistic we have a ratio of the difference between groups to the variability within the groups (individual scores from group means) Total variability comes from: Total Variance
Differences between groups Differences within groups
Variance due to Group Membership Variance due to non-systematic factors
difference between means t= difference due to individual differences/chance variance between sample means F= variance due to individual differences/chance
Note as we have before that the t-test is just a special case (2-group) of Analysis of Variance, i.e. t2 = F
The F-statistic is essentially this ratio, but after allowance for n (number of respondents, number of groups) has been made The greater the effect of the group differences, the larger F will be all else being equal The p-value is the probability that the F-statisic obtained (or more extreme) occurred due to sampling error assuming the null hypothesis is true
As always p(D|H0)
Sums of squares1 Treatment/Model/Regression/Between Groups/Effect etc. Error/Residual/Within groups etc. Total (Treatment + Error) SSTotal = sums of squared deviations of scores about the grand mean SSGroup = sums of squares of the deviations of the means of each group from the grand mean (with consideration of group N) SSerror = sums of squared deviations of the scores about their group mean
SSTotal =
what they are, and one can see again that we are still in regression land
Variance in the outcome
SSB/t groups =
n( X j X .. ) 2
SSW/in groups = ( X ij X j ) 2
SStotal
SSBetween Groups
SSWithin Groups
The more the sample means differ, the larger will be the between-samples variation
The graph here helps us to actually see the F statistic as a ratio of variances The example looks at length of sentence given the defendant by the mock-juror subject for defendants rated as physically attractive, average, or unattractive
Red triangles = Group means Green dot = grand mean Black = individual data points (with some jitter) Line represents a plot of the means against their difference from the grand mean (bottom axis) Blue box = MSw/in Red box = MSb/t
Ratings for a reality tv show involving former WWF stars1, people randomly abducted from the street, and a couple orangutans 1) 18-25 group2 2) 25-45 group 3) 45+ group 7 4 6 8 6 6 2 9 Mean = 6 SD = 2.2 5 5 3 4 4 7 2 2 Mean = 4 SD = 1.7 2 3 2 1 2 1 3 2 Mean = 2 SD = .76
SSGroup = 8(6-4)2 + 8(4-4)2 + 8(2-4)2 = 64 SSTotal = (7-4)2 + (4-4)2 + (6-4)2 + (3-4)2 + (2-4)2 = 122 SSerror = SSTotal SSGroup = 58 For SSerror we could have also added variances 2.22+1.72+.762 and multiplied by n-1 (7).
Well now we just need df and were set to go. SSGroup = k 1 where k is the number of groups (each group mean deviating from the grand mean) SSerror = N - k (loss of degree of freedom for each group mean) SStotal = N - 1 (loss of degree of freedom from using the grand mean in the calculation) or just add the other two.
Source
Group Error Total
df 2 21 23
SS 64 58 122
MS ? ?
F ?
p ?
Calling this ANOVA rather than regression doesnt change how the table is constructed. MS refers to the Mean Squares which are found by dividing the SS by their respective df. Since both of the SS values are summed values they are influenced by the number of scores that were summed (for example, SStreat used the sum of only 3 different values (the group means) compared to SSerror, which used the sum of 24 different values). To eliminate this bias we can calculate the average sum of squares (known as the mean squares, MS). As we have noted before our F ratio (or F statistic) is the ratio of the two MS values.
Source
Group Error Total
df 2 21 23
SS 64 58 122
MS 32 2.76
F 11.57
p .0004
There is some statistically significant difference among the group means. Measure of the ratio of the variation explained by the model and the variation not explained If < 1 then it must represent a non-significant effect.
The reason why is because if the F-ratio is less than 1 it
means that MSGroup i.e. our model variance, is just another estimate of the residual variance This is why you will sometimes see just F < 1 reported
this point. All that can be said is that there is some difference among the means of some kind. The details require further analyses which will be covered later.
Want equal ns if at all possible If not we will have to adjust our formula for SStreat, though as it was presented before (and below) holds for both scenarios.
n( X
X .. )
The more discrepant they are, the more we may have trouble generalizing the results, especially if there are violations of our assumptions. Minor differences (you know what those are right?) are not going to be a big deal.
When we violate our assumption of homogeneity of variance other options become available. Levenes is the standard test of this for ANOVA as well though there are others.
Levenes is considered to be conservative,
and so even if close to p = .05 you should probably go with a corrective measure. Especially be concerned with unequal n
Options:
See e.g. Tomarkin & Serlin 1986 for a comparison of these measures that can be used when HoV assumption is violated Welchs F and B-F
When you report the df, add round up and add a
squiggly (~) Welchs is probably more powerful in most heterogeneity of variance situations It depends on the situation, but Welchs tends to perform better generally
violation of HoV
Regular F
When normality is a concern, we can use nonparametric techniques (e.g. bootstrapped estimates) The Kruskal-Wallis is a non-parametric one-way on the ranked values of the dependent variable
With ANOVA one must run planned comparisons or post hoc analyses (next time) to get to the good stuff as far as interpretation Turn to robust options in the face of yucky data and/or violations of assumptions
E.g. using trimmed means
more groups One would report similar info: Effect sizes, confidence intervals, graphs of means etc.
nk wk = 2 sk
.
w X X = w
k k
X .) 2 k 1 F = w 2(k 2) 1 1+ 2 1 k n 1 k 1 k wk
k k
w (X
df =
k 2 1 1 wk 3 nk 1 wk
2
n (X F* = (1 n
k k k
X .. ) 2 / N )Sk 2
df = (k 1), f f = nk 1 ck 2
k
(1 nk / N ) Sk 2 c= (1 nk / N ) Sk 2
k