I got a question when estimating sample size. After getting the theoretical sample size, it needs to consider the response rate and invalid response rate when collecting data in the study based on an online survey. So, how to determine the invalid response rate before collecting data?
Hello Qichu,
Here are some options:
1. Run a modest-sized pilot study, using participants from your target population.
2. Examine extant studies using similar measures and methods, and determine what sort of non-completion rates were noted therein. In some cases, you might have to contact author/s directly, if such information were absent from the study report.
3. Look for meta-analytic studies of response legitimacy rates for on-line surveys.
In most instances, you're safer off to over-sample rather than under-sample.
Good luck with your work.
The % of responses that should be thrown out will vary by a lot of factors including the sample (the filters you set on MTurk or whatever), the actually system (e.g., Prolific v MTurk, Article Data quality in online human-subjects research: Comparisons ...
), etc.). Also it depends on how you decide what are invalid responses (e.g., Catcha, fast response times, questions designed for this).
You can check this out : https://stats4edu.com/services/power-analysis-and-sample-size-computation/
- Badges
- Science topic