Hypothesis Testing Flashcards
(19 cards)
type I error
rejecting a true null hypothesis (false positive)
type II error
accepting a false null hypothesis (false negative)
Probability of making a type I error?
α (e.g., for α = 0.05, type I error would occur 5% of the time)
H0 and HA for a one sample T test?
H0: this sample’s mean is equal to the hypothesised value
HA: this sample’s mean is different to the hypothesised value
H0 and HA for a two sample T test?
H0: the means of the two samples are the same
HA: the means of the two samples are different
two groups, s.d.’s differ by 3-fold. Which T test is appropriate?
two sample T test. If s.d.s differed >= 4-fold then use Welch’s T test (does not assume equal variances)
one group. Which T test is appropriate?
one sample T test
two groups, s.d.’s differ by 4-fold. Which T test is appropriate?
Welch’s T test. If s.d.s differed <= 3-fold, use two sample T test
H0 and Ha for analysis of variance (ANOVA)?
H0: all of the means of these groups are the same
HA: at least one of these groups has a different mean to the others
Test statistic for ANOVA?
F = treatment mean square (signal) / error mean square (noise)
grand mean versus group mean?
grand mean is the overall mean of a data set. Group means are the individual means of two (or more) groups within a data set
total sum of squares?
squared deviations from grand mean
treatment sum of squares?
squared deviations from grand mean to group means
error sum of squares?
squared deviations from group means to data points
how are total sum of squares, treatment sum of squares, and error sum of squares related?
total sum of squares = treatment sum of squares + error sum of squares
how are total degrees freedom, treatment degrees freedom, and error degrees freedom related?
total degrees freedom = treatment degrees freedom + error degrees freedom
how are mean squares constructed?
divide sums of squares by degrees of freedom
construct treatment mean square
treatment sum of squares/ treatment degrees freedom
construct error mean square
error sum of squares / error degrees of freedom
