8 - sampling in quantitative research

steps in conducting a social survey - only sampling

population + sample design

sample frame + sample size

follow up non respondence

sample

smaller + (ideally) representative part of bigger whole / pop

sampling frame

listing of all units in the pop from which sample will be selected

population

overall units from which sample is selected

sampling bias

distortion in representation

sampling error

error in findings due to dif between sample + pop from which it is selected

non sampling error

deficiencies in the sampling approach

census

all items of one population

2 potential sources of bias

non probability sampling used

sampling frame as inadequate

weighing

statistical adjustement that allows a study with a biased sample to be representative

inferential statistics

tools for making inferences about pop based on the analysis of data from a random sample of pop

Types of probability sample

simple random sample

systematic sample

stratified random sampling

multistage cluster sample

simple random sample

selected by chance

each same chance of getting selected

systematic sample

variation on the random sample

select units directly from the sampling frame - withoug using a table of random numbers

pick random number as starting point and progress at interval

stratified random sample

random from a pop that has been categorised

Multi stage random sampling

primery units as ‘clusters’

randomly selected from the clusters

margin of error

amount of random sampling error in results

max expected dif between sample statistic and true pop

Types of non-probability sampling

convenience sample, snowball sample, quota sample

Convenience sample

avalible to researcher due to accessibility

Snowball sample

researcher makes contact with small group

uses to get in contact with others

quota sampling

aims to produce sample of relative proportion of dif categories

absolute and relative size

absolute > relative size

sample size

large sample cannot guarantee precision

increases the likely precision of a sample

larger sample = lower error

unit non responce

non-sampling error

members of a sample refuse to cooperate

item non responce

survey responcets failute to responf to an item in the survey resulting in missing data

heterogenity of the population

higher heterogenity = need for larger sample

kind of analysis

more comparisons / subgroups = larger sample

sampling issues of online surveys

ppl with + 1 email adress = may answear survey + 1 time

household devices might be used by multiple ppl - ensure right person responds

internet users - specific demographics

few sampling frames of the general online pop + expensive

boost responce rates in online surveys

contact potential responders before q

follow ip

incentives

limits to generaliztion

generalize findings to the pop from which sample was taken

sampling error

unlikely that any sample will truly represent pop

data collection error

design on the data-collection instruments + poor worded questions

data-processing error

issues of the way data is managed

inputted / coaded