Chapter 12: Weighting the survey data

Question 1

In what 2 scenarios do you need to weight your data?

Answer 1

1. If you want to make inferences to the entire population
2. If your weighting variable has a significant effect on your outcome variable. otherwise it provides no additional information

Question 2

What is post-stratification weighting?

Answer 2

1. Weighting after the data collection has been completed.
2. you compare your gender scores vs gender score of the population.
3. Formula = divide population by sample score / ratio

Question 3

What is Propensity score adjustment?

Answer 3

1. A correction for if someone had the possibility to participate in a study or not or belongs to a certain subgroup or not
2. These propensity scores (possibility scores) are used as weights.

Question 4

What are auxiliary variables in statistics?

Answer 4

1. Variables about which information is available prior to data collection and this data is known for the full population

Question 5

What is generalized regression estimation?

Answer 5

1. Use a linear regression model to explain a target variable from one or more auxiliary vraiables

Question 6

What is raking ratio estimation?

Answer 6

1. Auxiliary information is used to create survey weights raking on known marginal totals

Question 7

What is reference survey estimation?

Answer 7

1. Do s small probability-based sample and your final research sample is weighted proportionally to the data from the reference survey

Question 8

What are the 6 reasons for applying weights?

Answer 8

1. Consequences of random sampling
2. Over- or undersampling
3. Self-selection bias correction
4. Under-coverage
5. Non-response or low response rates
6. Correlation between target variable and participation

Question 9

What is the difference between design weights and adjustment weights?

Answer 9

1. design weights - Weights that account for different inclusion probabilities due to design choices
2. Adjusment weights - weights used in order to achieve a representative sample of the population

Question 10

Why are sociodemographic variables often used as auxiliary information? (3)

Answer 10

1. Variables are often related to an outcome variable
2. Practical to use because information is known
3. Knowledge of the population distribution, thus good for potentially improving generalization

Question 11

What are the downsides of using a reference survey?

Answer 11

1. It is often a relatively small sample, thus decreasing precision of population distribution
2. There is less bias, but more variation in a population estimate