Collecting Data

population

entire group of individuals we want information about

census

collects data from every individual in a population

sample

subset of individuals in the population from which we collect data

random sampling

using chance process to determine which members of a population are included in a sample

simple random sample

sample of size n is chosen in such a way that every group of n has an equal chance to be selected as the sample (label, randomize, select)

sampling without replacement

individual only selected once

sampling with replacement

individual can be selected more than once

bias

study shows bias if it is very likely to under/over estimate the value you want to know. result of bad design, not bad luck. will consistently miss truth

convenience samples

selects individuals from population who are easy to reach

voluntary response bias

people choose to be in the sample by responding to general invitation

undercoverage bias

when members of a population are less likely to be chosen/cannot be chosen in a sample

nonresponse bias

when individual is chosen for sample can't be contacted or refuses to participate

response bias

systematic pattern/conditions in a survey when someone influences your response

stratified random sample

split population into groups of similarity (strata) then take SRS of each group. those chosen from each SRS are selected for final sample [homogenous grouping]

cluster

split population into groups (clusters) based on location then randomly select clusters (1 SRS). all individuals within cluster are selected for sample.

systemic random sample

randomly select a starting point and every nth individual thereafter will be selected from sample

observational study

observes individuals and measures variables of interest but doesn't try to influence responses (don't do anything to subjects)

explanatory variable

may help explain or predict changes in a response variable

response variable

measures an outcome of a study

confounding

two variables are confounding if it is impossible to determine which one is causing a change in the response variable. beware the influence of other variables

lurking variable

hidden variable

experiment

deliberately imposes treatments/conditions on individuals to measure their response

placebo

treatment with no active ingredient but is otherwise like other treatments

placebo effect

when some subjects in an experiment respond favorably to any treatment, even an inactive one

treatment

specific condition applied to individuals in an experiment

experimental unit

object to which a treatment is randomly assigned

subjects

when experimental units are human

control group

used to provide a baseline for comparing the effects of other treatments. depending on purpose, control group might be given an active treatment, a placebo, or no treatment at all

double blind

neither subjects nor those interacting with subjects know which treatment a subject is receiving

random assignment

experimental units assigned to treatments using a chance process

control

keeping other variables constant for all experimental units

replication

giving each treatment enough experimental units so that a difference in the effects of the difference can be distinguished from change variation due to random assignment. repeatability!

completely randomized design

experimental units are assigned to the treatments completely at random

block

group of experimental units that are known before the experiment to be similar in some ways that is expected to affect the results of the treatment [use randomized block design]

matched pairs design

common experimental design for comparing two treatments that use 2 blocks of 1 each [2 very similar experimental units are paired and are randomly assigned to 2 treatments]