Term
|
Definition
whole population is surveyed.
cons:
-time
-expensive
-need list of everyone
- need to get hold of everyone
-need everyone to volunteer/participate |
|
|
Term
|
Definition
| portion of population is surveyed |
|
|
Term
| descriptive study vrs. analytical study |
|
Definition
descriptive: describe characteristics of a population
analytical: assess specific associations between factors (exposures) and disease (outcomes) |
|
|
Term
| 3 main stages of sampling |
|
Definition
1. determine who/what to sample
2. determine how you're going to choose subjects
3. determine how many you'll need to be confident in findings |
|
|
Term
| validity of results are determined by what step of sampling |
|
Definition
determining how you choose subjects
-if subjects not truly representative of population of interest, conclusions may be wrong |
|
|
Term
| what two criterias should be established BEFORE you start sampling? |
|
Definition
| inclusions criteria and exclusion criteria |
|
|
Term
| name the three types of populations |
|
Definition
target population
- (pop we want to study)
source population
- (pop we draw sample from)
study population
- sample participants |
|
|
Term
| describe and discuss relationship between external and internal validity |
|
Definition
external validity: how well can study results be extrapolated to target population
internal validity: how well does study relate to source population?
*if we don't have right answer for internal validity, extrapolation to external validity is pointless |
|
|
Term
| what determines the nature of any extrapolations made from sample to population? |
|
Definition
|
|
Term
| briefly list 3 sampling strategies (and sub groups if there are) |
|
Definition
non-probability sampling
probability sampling
- simple random sampling
- systematic random sampling
- stratified random sampling
other:
-cluster
-multistage |
|
|
Term
|
Definition
| sampling units are chosen because they're easy to get |
|
|
Term
|
Definition
| investigator chooses what s/he deems to be units that are representative of population |
|
|
Term
|
Definition
| sampling units are chosen on purpose because of their exposure or disease status (in analytic study) |
|
|
Term
| list pros and cons of non-probability sampling |
|
Definition
pros:
- relatively easy to do
- cheaper if you choose subjects based on convenience
- appropriate for a homogenous population
cons:
- biased results if subjects don't represent target pop.
- can limit how far you can extrapolate results |
|
|
Term
|
Definition
fixed percentage of source population is chosen using random process with all individs having equal chance of being chosen.
sampling frame must be known. |
|
|
Term
| systematic random sampling |
|
Definition
good when you don't have complete list of individuals in pop to be smapled.
inidvids must be sequentaially availabe ( ie cattle running through chute)
sampling interval (j) is computed
j = study pop size/required sample size
starting point randomally selceted |
|
|
Term
| stratified random sampling |
|
Definition
| before participants chosen, sampling frame broken down into strata based on some factor likely to influence level of characteristics being measured. random sampling done within each strata. percentage sampled within each strata does not have to be the same in al groups. |
|
|
Term
|
Definition
sampling unit is a group of individuals with things in common. the unit of concern is individuals within the group. ALL individuals in sampling units are tested. (can be either nonprobability or probability )
pros:
- easier to get list of all clusters (farms) in area than to get list of every single animal.
- cheaper/ easier (example: going to test all animals in 20 heards than driving around to test 5 animals on 200 dif farms) |
|
|
Term
|
Definition
sampling takes place at both cluster/group level AND individ level.
convient when indivuals in cluster are so alike that measuring just a few will provide sufficent info.
- if random sampling done, can be more cost-efetive than other propability sampling methods. |
|
|
Term
|
Definition
how tight your confidence interval is around your study mean/proportion
dependent on:
- sample size
-variability of characteristics
-sampling strategy you've used |
|
|
Term
| steps to choosing sample size to estimate simple, descriptive characteristics |
|
Definition
1. estimate expected variation of interest from either EXPECTED PROPORTION (p) if you are estimating proportion or EXPECTED POP VARIANCE (σ2) if measuring mean
2. select level of confidence that your estimate will include the true value in the pop (eg 95% CI)
3. specify the desired precision (total width) or the 95% CI (Eg within 6% of true level 95% CI)
4. use appropriate formula to calculate sample size |
|
|
Term
contrast variance for proportion and means
- give symbols and meanings |
|
Definition
variance proportions
- p = proportion of individs that have attriubte
- q = proprotion of individs that don't have attribute
variance means:
- x = individual value, -x- = mean value
- variacne = standard deviation ^2 |
|
|
Term
| descibe level of confidence (Za) in descriptive studies and analytic studies |
|
Definition
descriptive studies
- desired confidence htat our estimate is close to the true pop value
analytic studies:
- our certainity that observed differences are real and not due to chance alone |
|
|
Term
|
Definition
| "allowable error" or margin or error |
|
|
Term
| n = Za^2 * σ2/L^2 is used to calculate sample size for what? |
|
Definition
|
|
Term
| n = Za^2 *p*q/L^2 is used to calculate sample size for what? |
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
| estimate of pop proportion |
|
|
Term
| steps for choosing sample size to test for difference between 2 groups with respect to a factor |
|
Definition
1. state null hypothesis and either 1- or 2- sided alternative hypothesis
2. determine what it is you're comparing (means or proportions)
3. determine how much of a difference between the groups you want to detect (and expected σ2 if necessary)
4. set alpha and beta (confidence and power)
5. use appropriate formula or table to estimate sample size |
|
|
Term
|
Definition
| ability of study to detect differences between groups when real differences exists |
|
|
Term
| type I error and Type II error |
|
Definition
Type I: rejuect null hypoth when it is true
- false claim
Type II: accepting Ho when it is false
- missed opportunity |
|
|
Term
[image]
this equation is used to calculate sample size when? |
|
Definition
| when comparining 2 proportions |
|
|
Term
[image]
this equation is used to calculate sample size when? |
|
Definition
|
|
Term
| what happens to number of subjects needed per group if we want more power to be able to detect statisically significant difference between two groups |
|
Definition
| with increased power, there will be decreased B value, and thus number of subjects needed in each group will increase |
|
|
Term
| what happens to number of subjects needed per group if we think that the prevalence of one group is raised? ie there is a bigger difference between the two groups? |
|
Definition
| fewer group members are needed |
|
|
Term
| contrast detectability of disease when there is a large or small association |
|
Definition
large association = easier to detect
small association = harder to detect |
|
|
Term
| name four times you need to increase sample size |
|
Definition
-size difference between 2 means of proportion decreases
- level of power to detect a difference between 2 groups increases
- number of confounders controlling increases
- number of hypotheses tested increase |
|
|
