chapter 5: Challenges in Statistics Flashcards
what type of study can we see prevalence used in?
cross sectional
define prevalence
fraction of the group that has the disease AT ONE single point in time
define incidence
rate of NEW cases of disease over a PERIOD of time
a test that compares the observed vs expected numbers of subjects in each category is known as
chi-square goodness of fit test
what is the best test to do for counts/frequency of things occurring within a sample
chi square goodness of fit test
what does it mean when the p value is low in a chi square test
reject the null that things are due by chance/random not related to what was expected
aka something really is going on.. (observed results are different from expected)
some factor other than chance must account for the large number of deaths from CHD while fighting fires
what does it mean when the p value is high in a chi square test
do not reject the null hypothesis that there really is no true difference b/w observed and expected and it is due by chance/random.
in a chi square test what is the null hypothesis in the data
the number expected
t/f cross sectional study can determine incidence
false! incidence are cases over a PERIOD of time vs cross sectional study is in a single point in time
cross sectional are divided into groups based off of _____
risk factors
single samplle of subjects is selected without regard to either the disease or the risk factor test is known as
cross sectional study
the prevalence of disease is compared between two groups can be tested with
cross sectional test
what type of study is a longitudinal study
propetive study
what type of study is one that collectsdata NOW and continues later
prospective study
what study observes over time to determine the incidence rates in the two groups selected
prospective study
what type of study has a single sample of subjects selected and randomly divided into two groups where outcomes or incidence is observed
expermental study
which study is easier to do and why, retrospective or prospective
retrospective; its using data that has already been collected! all you need to do is interpret the results
is a case control study retrospective or prospective and how
retrospective bc looking at risk factors that some people already have i.e. smokers (already smoke; not telling them to start smoking for the purpose of the study)
what does randomized study mean
so you already have the group that youre going to be studying on and THEn randomize them into groups like one will receive the medication and another the placebo
single blind vs double blind study
single: the patient doesnt know what theyre getting
double: the patient NOR the person conducting study (researcher) knows what group the patients are in
if the Confidence interval does not have a range of zero; i can assume that this test result is ____
statistically significant
i.e. the null hypothesis is saying that there is NO difference aka the difference would equal to zero; so if there is a zero in the CI range, this means that the null hypothesis is not rejected and the result is NOT statistically significant
if the Confidence interval does not have a range of zero; i can assume that this test result is ____
statistically significant
i.e. the null hypothesis is saying that there is NO difference aka the difference would equal to zero; so if there is a zero in the CI range, this means that the null hypothesis is not rejected and the result is NOT statistically significant
how can we tell if a study is stat significant without a pp value ?
The CI range, if there is NO zero in the range we can say it is stat significant
the incidence of disease attributed to the risk factor minus the incidence of the disease in persons not exposed to the risk factor is known as
attributable risk
how is the attributable risk determined
subtracting the incidence of disease progressed in group taking medication vs the disease progressed in group taking placebo
what tells you how many patients would have to be treated for the disease to be prevented in one person
number needed to treat (NNT)
attributable risk
what is the risk of disease progression attributable to the influence of (treatment)
what is better, a smaller or a bigger NNT?
less people needed to treat one (which is the goal)
is a higher or lower number needed to harm preferred?
(goal is to not harm people) – aka the more people not harmed per one person the better..
—
in other words, if i had a smaller number that means more people would be harmed with medication!! (not what we want)
what does a relative risk lower than 1.00 mean
statistically significant
- tx is making a difference compared to placebo–> lessening progression of disease
a relative risk of 1.0 means
it is not statistically significant; cannot reject the null (ratio of 1/1)
if a relative risk is greater than 1.0
treatment is making a difference negatively; (ratio of high experimental group compared to controlled group); treatment is progressing disease
is there a direct influence in observational studies?
no, just observing to seeng whats going on, not having a direct intervention
what is the purpose of using a mutliple regression method
- they asses the impact of one variable, while adjusting for others
- to create an equation for making useful predictions
- to understand scientifically how VARIOUS variables might impact an outcome***
looking at a whole bunch of different x values to predict the y that we are interested in is a ____
multivariable regression
what is the independent and dependent variable in the mutliple regression
dependent variable is y= the outcome
independent variable is the x=the predictor.. (what predicts y)
mutltivariable vs mutltivariate
used interchangeably (mainly looking at variable)
variable: the methods used when there is one outcome and several independent variables
- vs variate: looks at SEVERAL outcomes at once
an extraneous variable that correlates (pos or neg) with both the dependent and independent variable
what’s confusing/skewing the results! making it widely different
how do you adjust for confounders
multiple regression to adjust for those multiple variables
which study is stronger, observational studies or randomized control trials, why?
randomized control bc there is an intervention; we can control, it is PROSPECTIVE
vs
observational there is NO intervention and dont prove cause and effect
—-
I.e. the best way to proof cause and effect is to intervene in RCT and inject people with lead to see if x really is causing y
what is the best way to overcome any doubt about cause and effect
experimentation (RCT/intervention)
what regression is used when there are 2 possible outcomes
logistic regression
if odds ratio is near 1, does the independent variable have little or large impact on outcome
little
if odds ratio is further/higher than one, what does it mean about the independent variable
very likely that the outcome will occur
if the odds ratio is much less than one, what can be said about the independent variable
an increase in that independent variable is associated with a decrease likelihood of the event occurring