Chapter 11: Processing and cleaning the data

Question 1

When cleaning the data, what 3 files should you keep of the survey?

Answer 1

1. Original raw dataset
2. Cleaned eataset
3. Enciphered (keyed) dataset

Question 2

How to check for multivariate outliers?

Answer 2

1. Conduct a regression analysis and use Mahalanobis distance

Question 3

what are the 3 different missingness mechanisms?

Answer 3

1. Missing completely at random (MCAR)
2. Missing at random (MAR)
3. Missing not at random (MNAR)

Question 4

What is missing completely at random? (3)

Answer 4

1. Probability of missing is the same for all cases
2. Missingness does not depend on any data
3. Missingness does not follow a pattern

Question 5

What is missing at random? (3)

Answer 5

1. Probability of missing is not the same for all cases and may depend on observed information
2. Missingness depends on the data, but data does not depend on missing data (thus; missing data is irrelevant)
3. Missing data follow a known pattern

Question 6

What is missing not at random (2)?

Answer 6

1. Probability is not the same and may depend on the missing information
2. Missingness has a pattern, but pattern is not observed or unknown

Question 7

With what types of missing data can you use imputations?

Answer 7

1. MCAR or MAR

Question 8

What are ways to treat outliers?

Answer 8

1. Minimize influence by changing raw score to something less extreme (mean/mode/median)
2. Delete the extreme cases from the analysis

Question 9

What are 2 different imputation strategies?

Answer 9

1. Impute the variable mean to the cases with missing results
2. Use a regression analysis for imputation value

Question 10

How can you check for strange patterns in categorical variables?

Answer 10

1. Frequency distribution of 2 variables at the same time. Combination of value X on Y can be strange

Question 11

How can you check for strange patterns in continuous variables? (3)

Answer 11

1. Bivariate outliers, Mahalonobis distance
2. Min-max, means and standard deviation
3. Mediam or mode

Question 12

What should be included in the codebook of a dataset (10)?

Answer 12

1. Respondent / panel recruitment
2. ID numbers
3. Randomization variables
4. Instruction text
5. Vraiables
6. Question text
7. value and variable labels
8. Routing
9. Descriptives
10. Screenshots of questionnaire

Question 13

What is a bivariate outlier?

Answer 13

1. An outliers that is an outlier because of the combination of values on both x- and y-space. The values in itself are not outliers, but their combination is

Question 14

Name 3 different types of paradata and why it is useful to add those to your dataset?

Answer 14

1. Duration of the survey - survey seriousness and understanding
2. Duration of the question - question seriousness and understanding
3. Brower or device - for detection of errors across different devices / operating systems