Lecture 2 Flashcards
what are statistically methods useful for
studying, analyzing, and learning about populations of experimental units
define experimental or observational unit
an object about which we collect data
person, thing, event, tree
thing we measure
define population
set of all units - all possible measurements we could take
set of all unit that we want top study
can be infinite
define variable
what we observe
characteristic or property of an individual experimental or observational unit in population
like age, weight, counts
define experiment
measure outcome of some physical action we take
define sample
a subset of units in population
Collection of observations we make
represents pop
describe random sample
assume some things about how data was collected
define statistical inference
estimate, prediction or some generalization about population based on info contained in sample
does evidence in data support physical law
Characteristic
describe sample mean
sample mean = sum of observations/# of obs
describe measure of reliability
statement
usually quantitive
about a degree of uncertainty associated with a statistical inference = using limited number of obs= wont learn everything
based on data of sample
give examples of measure of reliability
flipping coin = heads all 3 times= chance of heads is 1
but experiment limited
based on data of sample = size mainly
how many things do we need to determine in study
7 things
name all 7 things we need to determine in study
1 - Objectives of study
2 - experimental units
3 - population under study
4 - characteristic(s) of interest
5 - sample
6 - inference(s) of interest
7 - measure(s) of reliability
describe objectives of study - determine in study
what questions we want to answer in study
important for ethics approval if using humans
describe experimental units - determine in study
people
machines
transactions
what we could possibly measure
describe population under study - determine in study
set of experimental units
(all first year undergrads, all transactions at Walmart in a day)
describe characteristic of interest - determine in study
variable(s) measured on th experimental unit
can measure multiple but mainly we focus on one
describe sample - determine in study
subset of pop under study
if we could = take census (whole population) but if cannot= take sample
describe inferences of interest- determine in study
about pop based on sample
describe measure of reliability- determine in study
to account for limited sample
describe infinite vs finite pop
if finite = sample = all the obs units = population = census
but infinite = # of possible experimental units we could get is infinitely large
describe sub cutaneous fat necrosis in newborns - generally
Some newborns suffer asphyxia during birth
A treatment for perinatal asphyxia is called therapeutic hypothermia= involves lowering the child’s body temperature for 72 hours, then the child’s body temperature is slowly raised to normal
Some children develop the rare condition sub-Cutaneous fat necrosis (SCFN)= this usually disappears within a few weeks but some children develop more serious complications
describe sub cutaneous fat necrosis in newborns - population
all full term newborns who may suffer from perinatal asphyxia that will get hypothermia as treatment (units of interest)
describe sub cutaneous fat necrosis in newborns - variable of interest
if child developed scfn (yes/no)
birthweight (sample mean)
other risk factors
could measure all or specific one
describe sub cutaneous fat necrosis in newborns - sample
term asphyxiated newborns admitted to montreal childrens hospital NICU from aug 2008 –> march 2015
describe sub cutaneous fat necrosis in newborns - inference
one estimation of birthweight for those who developed scfn
report sample mean and try to extrapolate and reflect pop
describe inference more detailed
drawing conclusions from data about pop
also use word to pose question and answer it
describe quantitative data
Measurements recorded on naturally occurring numerical scale
Numerical measurement
ordinal
ex = height, blood pressure, temperature, counts
describe qualitative data
Measurements cannot be made in a natural numerical scale
can only ascertained and classified into one of several groups or categories
nominal
no natural order
ex= gender, presence of disease, religion, ethnicity, coin toss, faculties, colours
describe the methods of obtaining data
published sources = literature, books, government, publications
designed experiments
observational studies
describe designed experiments
researcher designs experiment = has control over experimental units sampled = can intervene and change what is going on
usually 2 groups = group of experimental units assigned to a treatment and another untreated
may involve more than 2 groups
like clinical trial, best for science
can have heterogeneity = can control so much, less variability
give ex of fertilizer study for designed experiments
claim = organic fertilizer increases yield and size of tomatoes compared to other brand
Experiment = tomato plants assigned to 2 groups = organic and reg fertilizer
size and weight of each tomato recorded and total yield
RESEARCHER has control over what plant gets which fertilizer
describe observational studies
Sometimes unethical or impossible to assign treatment to a group (money, cannot deny standard of care)
researcher can only observe - indirectly or directly the experimental units and record variables of interest
cheaper but can be hidden sources of variability bc don’t have control over what groups are doing specifically
ex of obs study = smoking and lung cancer
cannot assign smoking and non smoking groups = cannot be randomized bc unethical, also we know relationship now = health hazard
possible approach = select cancer and non cancer patients and determine whether they were smokers or not from records
RESEARCHER can only observe directly or indirectly the experimental units and records variables of interest = smoking status
describe iron man competition
swim, bike, marathon, no break
some athletes develop hyponatremia during race = abnormally low levels of sodium in blood
can lead to health issues and death
mild case can be treated with iv saline solution
describe what investigators interested in for iron man comp
determining possible causes of hyponatremia
testing whether oral saline solution as effective as iv saline for treating mild case
explain objective 1 of iron man comp
causes of hyponatremia
cannot indue or provoke in subjects
can only wait for it to occur in some althltes during race
then can make comparisons between who got it and who didn’t
OBSERVATIONAL STUDY
explain objective 2 of iron man comp
testing treatments
consider only those who suffered hyponatremia during race
make 2 groups by assigning at random some to oral and some get iv
have control over who gets what
Compare effectiveness of treatments between groups
DESIGNED EXPERIMENT
why would a researcher restrict population
for homogeneity of pop and effect we look at
like for age = common restriction
