Chapter 7 Sampling

Question 1

Population

Answer 1

The pool of units from which a statistical sample is drawn for a study. For example, if you want to examine Lund University students’ opinions on Allemansrätten, then the population = all the 40,000 students admitted to Lund.

Question 2

Sample

Answer 2

A subset of the population. The number of units that you will actually examine / apply some statistical manipulation to. The sample will often NOT equal the whole Population, because of cost- and time-constraints. To be able to generalize findings in the sample to the population, it is important that the sample has been unbiasedly picked. It may be picked base on either a probability or non-probability approach.

Question 3

Representative Sample

Answer 3

A sample that reflects the population accurately, so that the results gathered from the sample can be generalized to the whole population.

Question 4

Probability Sample

Answer 4

Probability sampling is a technique used to select a sample of individuals from a larger population in a way that each member of the population has a known and equal chance of being selected for the sample. In other words, probability sampling ensures that each individual has an equal opportunity to be selected, and the chances of being selected are determined by the laws of probability.

Question 5

Non-probability Sample

Answer 5

A sample that has not been picked using a random selection method. Implies unequal probability of selection of units across the sample; some units are more likely to be selected than others. Impairs the generalizability of the findings in the sample towards the population, and thus there is a higher risk that a non-probability sample won’t be representative.

Examples:
- Convenience Sampling
- Quota Sampling
- Snowball Sampling

Question 6

Sampling Error

Answer 6

Sampling error is a term used in statistics that refers to the difference between the results obtained from a sample and the results that would have been obtained if the entire population had been sampled.

Despite using a random and representative sample, there may still be some degree of error or variation due to chance. Sampling error is a natural part of statistical analysis and is usually quantified to determine the accuracy and reliability of the results.

Question 7

Non-response

Answer 7

A source of non-sampling error, particularly likely to happen when individuals are being sampled. Occurs whenever some members of the sample refuse to “cooperate” (e.g. you refuse to answer a survey sent to you via mail).

Question 8

Non-sampling error

Answer 8

Non-sampling errors refer to errors in data collection or analysis that are not related to the sampling process itself. These errors can occur at any stage of the research process, from designing the survey instrument to data entry and analysis. Non-sampling errors can be caused by a variety of factors, including interviewer bias, respondent bias, measurement errors, and data processing errors. Unlike sampling errors, non-sampling errors are not related to the size of the sample and can affect both small and large samples. These errors can have a significant impact on the accuracy and validity of research findings, and researchers need to take steps to minimize them as much as possible.

Question 9

Census

Answer 9

The total of an entire population. If data is collected in relation to all units in a population, the data is called “census data”.

Question 10

Biased Sample

Answer 10

A sample that has been picked in a manner where the probability of picking a certain unit of the population has been unequal to the probability of picking another. Resulting in skewed results and weak generalizability. I.e. a NON-representative sample

Question 11

Sources of Bias

Answer 11

3 main sources of bias

1. Using a non-probability sampling method => Possibility that human judgement will affect the selection process, making the probability of picking some units of the population unequal to the probability of picking some others.
2. Inadequate / Inaccurate sampling frame. (difference from 1????)
3. If there are considerable amount of non-responses from the sample.
   Why problem? Because those who agree to participate might differ in various ways from those who do NOT agree to participate in the survey.

Question 12

Types of Probability Sampling

(remember: Probability Sampling is good; generally makes the sample representative aka generalizable)

Answer 12

1. Simple Random Sample = most basic form of probability sampling. Here, each unit of the population has an equal probability of being selected into the sample. E.g. s = 450, P = 9000. If Simple Random Sample is applied, each unit should have a 450/9000 = 5% chance of being picked into the sample.
2. Systematic Sample: A systematic sample is a type of sample where you choose units from a sampling frame by selecting a random starting point and then picking every nth unit until you have the desired sample size.

3.Stratified random sample (dividing population into subgroups (strata) based and organized upon shared characteristics) and then randomly select units from each subgroup.

4. Multistage Cluster Sampling (sample from population, using smaller and smaller samples each time) often used when there is a large geographic spread. E.g. first sample companies, and then sample employees from those companies. A probability sampling method would need to be employed at each stage. E.g. we might randomly sample ten companies from the entire population of 100 largest companies in the UK, thus yielding ten clusters, and we would then interview 500 randomly selected employees at each of the ten companies.

Question 13

Why is Probability Sampling so much better to do than non-probability sampling?

Answer 13

Because it allows us to make inferences from conclusions drawn from a sample to the population from which the sample was selected.

Essentially, it is a mechanism for reducing bias in the selection process.

Remember STAA36 ;)

Question 14

What is Equivalence Sampling?

Answer 14

A method that aims to ensure that findings are equivalent BETWEEN samples. It is NOT a probability sampling method and does not in and of itself make the sample generalizable to the whole population

Question 15

What does a confidence interval of 95 % in a sampling distribution imply?

Answer 15

That we can be 95 % confident that the true population mean lies between the sample mean and + or - 1.96 standard errors of the mean.

Question 16

Does Sample Size matter?

Answer 16

Yes, but there is no definitive threshold for what’s a sufficient sample size. Often it is a matter of time and cost.

Do also note, that it is the absolute sample size, not relative sample size (as in relative to the population) that matters. => a sample of 1000 from the population of China will be as valid and inferable as a sample of 1000 from the population of Sweden, although China has more than 100 times the population of Sweden.

Increasing the absolute sample size will increase the precision of the sample. However, at around n = 1000, the incremental precision-increases become less and less pronounced (even though they don’t plateau)

Another thing to note: the more heterogeneous the population, the larger the sample size needs to be.

Question 17

Convenience Sampling (a type of non-probability sampling)

Answer 17

When you sample simply according to accessibility. Not generalizable to any population, because we cannot then know of what population this sample would be representative.

Can still be useful though, especially for piloting a research instrument (trying it out) before using it in an actual representative survey.

Question 18

Snowball Sampling (a type of non-probability sampling)

Answer 18

A sampling method in which the initial observations bring you to new observations, i.e. the later observations will be dependent on the initial former. (Like rolling a small snowball down a hill, making it successively grow in size as it gathers more and more snow). Used in Qualitative Research (e.g. when doing semi-structured interviews).

Very unlikely that the snowball sample will be representative of any theoretical population you might have thought of, and is therefore not generally very generalizable. However, concern about generalizability and external validity is not as widespread in Qualitative Research, so it might still give some good insights if you’re doing Qualitative Research.

Question 18

Quota Sampling (a type of non-probability sampling)

Answer 18

Sampling where you produce a sample that reflects a population in terms of the relative proportions of people in different categories, that the researchers make up (e.g. gender, ethnicity, age groups, socioeconomic groups etc.)

The difference from a Stratified Sample (the probability-sampling-equivalence to Quota Sampling) is that the sampling of individuals is not done randomly; the final selection of people is left to the interviewer. Because the choice of respondent is left to the interviewer, the proponents of probability sampling argue that a quota sample cannot be representative.

Common in market research and political opinion polling