unit 8/9 things to remember Flashcards
x2 assumptions
- SRS
- 10% rule where n<10% of population
- large number count - all expected counts are greater >5
when to use GOF chi square?
when there is one categorical variable with on population, comparing one population to a baseline
how to calculate expected value and degree of freedom for GOF chi square test
expected value - (percent of outcomes)x(total population)
degree of freedom - # of subcategories - 1
mechanics to know for GOF
P(x2>5/5) –> always greater than
use second vars #8 to find p value
when to use homogeneity x2 test
1 categorical variable with multiple populations - comparing multiple groups of samples
again, MULTIPLE populations
how to calculate expected value and degree of freedom for homogeneity chi square test
expected value - (marginal value of column)x(marginal value of row) / total population
df - (number of rows - 1)(number of columns -1)
mechanics to know for homogeneity + hypothesis
again, p( x2> jrngo) – GREATER THAN
normal x2 test
hypotheses: the number of (subject) will be spread evenly between the groups.
use a matrix :)
when to use an independence x2 test
when there are 2 categorical variables and 1 population and the word association or something like that is used
how to calculate expected value and degrees freedom for an independence test
expected value - (marginal value of column)x(marginal value of row) / total population
df - (number of rows - 1)(number of columns -1)
r
correlation coefficient - describes the strength and direction of the linear association of the scatter plot equation
LSRL - least squares residual line
y hat = a + bx
y hat = predicted y given x
a = y intercept
b = slope
se
standard deviation of residual
“the actual y variable is typically about s away from the predicted y variable”
Sb
standard error of slope. average distance that observed values deviate from the regression line. the smaller the closer to the line
“the slope of the sample LSRL typically varies from the slope of the. population LSRL by about Sb”
one sample t interval for slope conditions
- SRS
- 10% condition - independence
- Nearly normal - “The distribution of sample residuals is free of strong skewness and outliers” –> when n > 30
- nearly linear:
a) make a scatterplot to identify data/ look at it
b) check residuals
“The scatter plot appears to be fairly linear and residual plot appears to not have a pattern” - Equal SD: “Scatter plots show similar variability for each x value.”
one sample t test for slope mechanics
everything same as the interval except mechanics:
t= b-beta(0) / SEb
Hypotheses and df for one sample t interval for slope
null: beta = O
alternative: beta doesn’t equal 0/ is greater/less than 0
df : n-2
