Mock exam Flashcards
In an even sample size which ones the mean
mean is midway between the two middle values
In an odd sample size which ones is the mean
mean is the middle number
which value is the upper quartile
75%
which value is the lower quartile
25%
where do 50% of values lies between
between the upper and lower quartile
does sample mean equal population mean
no - sample mean varies from sample to sample whereas population mean is a fixed value
what happens to the widths of the confidence intervals as sample size increases
confidence intervals decrease as the estimate gets better/more precise so confidence interval narrows
what is the confidence interval for a mean
estimate of the mean - given in a range rather than a single value
what does standard error measure
how well a sample mean estimates the corresponding population quantity/mean
what does standard deviation measure
measures the spread of the population or sample
if a p-value is less than 0.05 what do you with null hypothesis
reject it
if a p value is above 0.05 what do you do with null hypothesis
accept the null hypothesis as there is no evidence against it, but we cannot say that it is true
is the data normally distributed for an unpaired t test
yes
is the data normally distributed for a paired t test
no, only necessary for the within pairs differences to follow a normal distribution
if a sample size increases by x what happens to the confidence intervals
they are reduced by the square root of x, not x
what do we use to assume normality during a regression analysis
residuals
what is another way to assume normality
inspect the y-values but you will not be taking proper account of the way the mean of the observations changes with x
what does the interquartile range measure
the spread of the data
how do you estimate the standard error of a sample
take standard error of the current sample and divide it by the square root of the factor the sample increases by
a 95% confidence interval for the mean of a large sample is approximately
the sample mean + or - 2 x standard error
what indicates a good agreement using the bland altman method of agreement
small bias, narrow limits of agreement, most data points falling within the limits
how do we test the association between response and predictor variables in regression analysis
use a t test built using the estimated regression coefficient and its standard error
what does an odds ratio of > 1 mean
condition is more likely in exposed subjects
what does a odds ration < 1 mean
less likely to be exposed
what does an odds ratio of 1 mean
equally like to be exposed/unexposed
what do we know about the standard deviation of the variation about the regression line
assumed to be constant, known as homeoscedasticity.
If the assumption is violated then the regression may be invalid
when data is highly positively skewed
the mean is greater than the median
lower tail is shorter than the upper tail
usually one or more extreme values
what does pearson correlation coefficient show when it’s negative and significant
increase in one variable means a decrease in another variable
what does cohen kappa do
calculation that corrects for chances of agreement, range of values between -1 and 1
what does a chi2 test do
it is used to test for associations rather than agreement
what is chi2 test applied to
it is applied to frequencies (actual counts) not percentages
what happens to standard error of mean and standard error of proportion when sample size increases
they both get smaller
what does the logit function do
used in regression
natural logarithmic transformation of the odds, natural log of the odds is the logit of probability of event or condition
what is logistic regression applicable to
a binary response variable with any number of predictors, categorical or continuous