Variance Flashcards
Standard Deviation
(s) average value that individual numbers vary from mean
Deviation
score-mean, error or riesidual
sum of deviation scores
0
sum of squares
(SS); squares of deviations
Variance
(s²) SS/(n-1); average of squared deviation from the mean
effect size
between group variance, assess for experimental effects on DV, measure of the impact of your manipulation
error
within-group variance, comparison of scores within a group, differences are seen as individual defferences
What does ANOVA look at?
ratio of between-group :: within group variance
Why would you throw out an outlier?
based on the premise that they do not represent your population
Sample distribution
frequency distribution from a sample, not affected by sample size, a nonbiased predictor of a population parameter
s
SD of sample distribution
Population distribution
frequency distribution of a population
σ (sigma)
SD of population distribution
Sampling distribution
frequency distribution of data from multiple samples
standard error (SE)
SD of sampling distribution, difference of a sample mean from the true population mean, measure of how well a sample represents the population
How can you increase the reliability of the sample mean as a predictor of the true population mean?
increase sample population size
What does large SE indicate?
significant variability between means of different samples, such that one or more of the samples may not represent the population well
How many scores do you need to generally assure an approximately normal distribution?
N ≥ 30
What is variance equal to?
variance = SD² = MSwithin = SS/df
Which has less error, the full model or the restricted model?
The full model always has less error than the restricted model
Which is a better predictor of the data, the full model or the restricted model?
The full model (using group means) is always a better predictor of the data than the restricted means (group mean)
What do all test statistics test?
the likelihood of obtaining a particular test statistic if there were no effect (null hypothesis)
Unsystematic variation
happens by chance/cannot be explained by the manipulation/model/hypothesis we fit to the data
