Sampling is frequently used by researchers. What is a sample in research terms? A sample is a representative subset of the population from which generalizations are made about the population.
What is a ‘population’? The ‘population is the group of people who you will conduct your research on.
Why one should sample:
1. Advocated by positivists as it contributes to the scientific narrative about the way in which society works.
2. All members of a population may not be available.
4. Less time consuming
What are probability samples?
Probability samples are samples where randomness is the bases for sample selection and insures that the sample is representative of the population. Probability samples: generalizations from sample to population are possible because sample is representative of the population.
What are Non-Probability Samples?
Randomness is not the basis for selecting the sample. With non-probability samples generalizations are not possible because the sample is not representative of population. Who advocates non-probability samples? Anti-positivists.
Limitations of non-probability samples:
Why is it not possible to generalize from sample to general population because:
Types of qualitative sampling procedures = Intensity sampling, Homogeneous sampling, Criterion sampling, Snowball sampling and Random purposive sampling.
Sampling in qualitative studies:
Qualitative sampling procedures are based on non-random processes.
Qualitative samples are typically small.
These are the conditions that maximize the likelihood of sampling variation and sampling bias.
Drawing inferences about a population from such samples is not logically defensible.
What is probability sampling? It is an “Equal and independent” approach to research. Every member of a population has an equal chance of being selected. Selection of one individual has no influence on the selection of the next individual. Humans cannot generate random numbers; a mechanism (such as a random number table) must be used. EG: Simple Random Sampling and Stratified Sampling
Simple Random Sampling: Every member of a population has an equal chance of being selected. Least chance of sample bias
Stratified Sampling: Proportion of subgroups in sample represents proportion of subgroups (strata) in population. Every member within the subgroup has an equal chance of being selected. Used when size of population subgroups is discrepant.
Systematic Sampling: Example: Select every tenth student from a randomly ordered school register. Principle of independence is violated, for selection of first student determines selection all others.
Sampling considerations: Random procedures do not guarantee that the sample is representative, but they do increase the probability. Sampling variation – Random differences between sample and population. Decreased by increasing sample size. Sampling bias – Non-random difference due to flawed procedures.
How large should the sample be? Too small a sample increases the likelihood of sampling error. Too large a sample reduces efficiency.
Poor sampling: What can happen with improper sampling? Incorrect conclusions can be drawn, such as the 1936 US Presidential election. Literary Digest poll incorrectly predicted Alf Landon the winner because the sample (people with telephones) was not representative of voters. This is an example of sampling bias.
A recent example of sampling – what will Scots do on the day of the royal wedding?
A new poll by Ipsos MORI has revealed a range of ways that Scots will mark this year’s royal wedding, while highlighting that significant numbers will ignore the whole event.
Around 6 in 10 will participate in some way, most commonly through watching the event on TV live (37%) and/or recorded highlights (38%). Fewer than one in 10 (9%) will be having a party with friends and family, while just 4% will buy souvenirs of the occasion. At the same time, 4 in 10 Scots will do nothing (5%) or will be trying to ignore the event (35%).
Women and those in older age groups are most likely to participate in the event. Around half of women intend to watch the occasion live (53%) and/or as highlights (48%). On the other hand, half of all men (51%) report that they will ignore the wedding. These findings show little significant change from when Ipsos MORI asked the same question in November 2010; it does however reveal that more Scots will be attending a party while fewer will be watching on TV.
Mori interviewed 1,002 adults aged 18 or over. Interviews were conducted by telephone between 14th and 17th April 2011. The data was weighted by age, sex, working status using census data, tenure using SHS 2007-2008 data and public-private sector employment by Scottish Government Quarterly Public Sector Employment series data.
Gallup found that World residents are more likely to blame human activities than nature for the rise in temperatures associated with climate change. Thirty-five per cent of adults in 111 countries in 2010 say global warming results from human activities, while less than half as many (14%) blame nature. Thirteen percent fault both.
People nearly everywhere, including majorities in developed Asia and Latin America, are more likely to attribute global warming to human activities rather than natural causes. The U.S. is the exception, with nearly half (47%) -- and the largest percentage in the world -- attributing global warming to natural causes.
Results are based on face-to-face and telephone interviews conducted in 2010 with approximately 1,000 adults, aged 15 and older, in 111 countries. For results based on the total sample in each country, one can say with 95% confidence that the maximum margin of sampling error ranges from ±1.7 percentage points to ±5.7 percentage points. The margin of error reflects the influence of data weighting. In addition to sampling error, question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of public opinion polls.
Courtesy of Lee Bryant, Director of Sixth Form, Anglo-European School, Ingatestone, Essex
"Sampling". HistoryLearningSite.co.uk. 2014. Web.