math unit 5 -- data collection

types of studies

cross sectional and longitudinal

cross sectional

a study that considers individuals from different groups at the same time

longitudinal

study that considers individuals over a long period of time

quantitative variable

number values (continuous, discrete)

qualitative

places an individual into one of several groups or categories (ordinal- naturally ordered, nominal- no order)

simple random sampling

each member of the pop is EQUALLY likely to be chosen and members are chosen independently of each other

systematic random sampling

• randomly choose a staring point from an ordered list

  • select every nth element in the population

stratified random sampling

• population is divided into strata

• do a simple random sample of each stratum

• the size sample is proportionate to the stratum’s size (10%)

cluster random sampling

• divide population into clusters

• randomly select a few of the groups, but all members from selected groups

multi- stage random sapling

• population is organized into groups

• simple random sample the groups

• simple random sample of the people within the groups

convenience sampling

• choose without any random mechanism

• individual who are easy to reach

• often leads to unrepresentative data

voluntary response sampling

• people who choose themselves by responding to a general invite

• people who feel strongly about the topic

• leads to bias

principles of survey design

1. survey should be designed around specific needs of a relevant topic

2. interviewer must create questions in advance, identify relevant variables, craft questions; interviewee must be truthful

3. interviewer at question design stage must try to help interviewee to be honest

closed question types

information, checklist, rating, ranking

what does a good question have

simple, specific, relevant, reliable

what does a bad question have

jargon, abbreviation, negatives, leading respondents, insensitive

sampling bias

chosen sample does not accurately represent population

household bias

one type of respondent is over represented B/C groupings of different sizes are polled equally instead of proportionately

non-response bias

individual chosen cannot be reached or refuses to participate

response/measurement bias

anything in the survey design that influences the responses (ex: wording)

4 principles of experimental design

1. Comparison

2. Random assignment

3. Control

4. Replication