significance and regression coeffeicent Flashcards
what tests would you use for significance and regression coeffeincent
- F test
- likelhood ratio test
- wald test
what has wider limits
prediction limits (unaffected by sample size)
what has narrower limits
confiednce intervals (affected by the sample size)
what is the general form for linear regression
predict y for a given value of x
what is the simple linear regressiobn model
y=a+bx+e
what is ‘a’ within a line with equation y= a +bx
= intercept or constant
what is ‘b’ within a line with equation y= a +bx
slope
what is the aim of regression anyalsis
find the line for the data in hand
what is residual
(random error)
it is teh difference between each obsereved data points
what is homosedasity
constant variance
what is there at each fixed value of x
a distint distribution of y that have differnet means but the same shape
what is a regression line
best fit straight line taht joins the values of y
what is the best way to look at normality
histograms and/or normal probabilty plots
what is the best way to look at linerailty
scatterplot
what is the main advantage of regression framework
capabilty of fitting multiple predictors
what does it mean if r=0
there is no linear relatiojnship between the variables
what does r>0 mean
positive slope
what does r<0 mean
negative slope
out of correlation and regression which one provides more infromative
regression
what happens to the confidence interval as the sample size increases
gets narrower
what happens to the prediction intervals when the population size increases
it is unaffected
what typoe of varaible is y
dependant vairble
what type of variavble is x
independant varibale
what is a residual
the difference between each obserevd data and is predicted by the fitted line
they are assumed to be normal
what is pearson r applicable to
continous varibales
what are the values for r between
-1 and +1
what is teh r value for a positive slope
if biggeer than 0 = positive
what is teh r value for a negative slope
if less than 0 then negative
what does it mean when r =0
no linear relation between the variblaes
what does the p value depend on
r and sample size
what are the disadvantages of correlation
only extreme values are easily interpreatble
what test would you use for agreement
bland altman
what is good agreement in bland altman
small bias, narrow limits of agreement and most data falloing within the limits
what is the limit of agreement in bland altman
biaas +or - 1.96X sd