week2-nonresponse-in-various-survey-modes

Reading Notes:

web surveys have lower response rate compared with other modes (11%)
pros of web-surveys:
- positive data quiality (increased level of reporting of sensitive info)
- time-sensitive aspects
- cheaper and faster
cons of web-surveys:
- question skipping, speeding
- responsive inconsistency, satisficing
- representativeness issues
representativeness issues is a big challenge:
- for populations with high internet coverage, nonresponse bias is low
- but in general population, internet users and non-users are not randomly distributed
Results
- comparing web surveys to other modes (mail and telephone), web surveys show significantly lower response rates
- survey design:
  - simpler and shorter surveys reduce the response rate gap between web and others
- target population:
  - among younger and digitally literate populations, the disadvantage of web surveys diminishes

declining trend in response rates (long term):
- clear and consistent long-term decline in survey response rates, especially in Western democracies
- response rates vary by mode of data collection:
  - face-to-face > telephone > web and mail surveys
Face-to-face:
- direct contact to researcher
- official clothing, researchers with official ID’s create trust
- it is possible to touch groups that are unreachable online or via phone
(especially in web surveys) financial and non-financial incentives and follow-ups help, but their effects are limited
distinguishing between refusal to participate and failure to contact is important
- methods to compete these are different
Conclusions
- authors recommend methodological pluralism: combining different data collection strategies, adapting to new technologies, and increasing transparency in reporting response metrics.

Difference between survey and marketing research:
- There is a legal difference between the two
  - Marketing research collects “personalized data”, written permission is needed
  - In survey research, data is not personalized
Sampling:
- Random: We know that the selection probabilities of our target group, and try to make it equal to all parts of the target population
- SRS — Simple Random Sampling
- Clustered Sampling, etc.
We are interested in target population, but because it is too costly, we select a sample, based on the answers that are from the sample group, we estimate the results for the target population
Purpose of the Survey: We want to estimate the parameters of the target population
- Mean, median, standart deviation
Total Survey Error:

- Variance: - Random error - Precision of estimate - Reliability - Bias: - Systematic Error - Accuracy of Estimate - Validity

What the reasons of non-response error?
- Mode of the survey
  - face-to-face, phone, web
Relevance of the survey to the target population
- making a survey about “language learning” among language learners gives more successful responses
It’s not only in the country level we see the difference in non-response, but also the study level — some topics have more respondandts
It is hard to measure whether non-response is a refusal or non-contact in terms of mail and e-mailing
Why response non contact rates are getting higher?
- leisure activities and work times are higher than the past, it is harder to find people available to survey
- lower number of household members than ever
  - less people are living together, modernity
  - it is more difficult to reach the people who could answer the phone
What about refusal?
- data protection sensitivity
- not interested in the topic, no connection to the topic
- they have less time, see it as a waste of time
Various modes of survey:
- telephone interview
- face-to-face(CAPI)
- face-to-face(PAPI)
- web survey
- mobile web survey
- classroom survey
- mail survey (address based sampling) — getting popular recently and the response rates are pretty good
Effect of different modes:
- Presence or the absence of the interviewer has a great effect on measurement error and non-response error
  - Sensitive questions are a great example.
  - Feeling anonymous