week 7 Flashcards
count data categories must be
mutually exclusive and exhaustive
mutually exclusive = can not be in more than one category
exhuastive = categories represent all options seen within data
four requirements of binomial experiments
- have a set number of identical trials
- have only two possible outcomes for each trial (eg yes/no)
- have mutually exclusive and exhaustive outcomes for each trial
- the probabilities of occurrence for the two categories must be constant across all the trials (expected values)
the centre of the distribution of sample proportion is
p-hat
the SD of the distribution of sample proportion is
square root of (p-hat)(q-hat)
variability is maximized when p-hat is
0.5
the further from 0.5 the smaller the
standard deviation
cohens D
difference between the sample proportion and the population proportion, all divided by the standard deviation of the population
the width of CI is
twice the bound of CI
the largest variance of a binomial variable is
0.25
variability is maximized when p hat =
0.5
what is often a better choice for the presentation of proportional data
count data
chi square is a test of
model fit
what is the z distribution
the standard normal distribution
how can a test of two independent proportions be conducted
z-test
2x2chi square test of independence
what does count data refer to
categorical random variables
what is the simplist form of categorical data
dichotomous variable
how many trials is a bernoulli trial
a single trial
what is the mean and SD for bernoulli distributions
SD = square root of pq
mean= p
what happens to SD as p-hat moves further away from 0.5
gets smaller
the width is twice the
bound
as variance increases the width
increases
two advantages of presenting proportional data as frequencies
- reminds the researcher to present total n size
- allows comparison of more than two proportions
what statistic will we use for count data analysis
chi-square calculations