Response Issues Flashcards
What are the two most common causes of monotones no response
Attrition
Break offs
Why do we care about the missing data?
Smaller effective N Findings may not be representative Inferences may be invalid Estimates may be unstable, biased Errors may be large Suggests problems with data quality in general
Reasons for item nonresponse
Chooses not to answer
Does not know how to answer
Accidentally skips a question
Breaks off an interview
How will you deal with item nonresponse?
Treat as valid response option
Non-completed if too many items are missing
Drop units
Impute values where missing
Reasons for unit 👓nonresponse
Non location Non contact Ineligible Refusal Incomplete participation
How will calculate response rate?
Percentage of all the rates e.g location rate, contact rate, eligibility rate, cooperation rate, completion rate
What are general characteristics of no respondents?
Males Low SES groups Unemployed Urban residence AA and other minority race Disabled Low health status
What are general strategies for reducing nonresponse?
Multiple contacts Mode switch Use of incentives Minimize privacy/confidentiality concerns Survey design Burden Sponsorship Relevance
Methods of assessing nonresponse bias?
Follow up interviews with non-response
Comparing early vs late respondents
Comparisons to similar estimates from other sources
Contrasting various post-survey adjustment for non respondents
How would you correct for nonresponse?
Imputation
Weighting eg population proportion of females 0.5, sample proportion 0.6, p/s is 0.8333
