One approach is to include pre-treatment variables as aggressors, or if you have complete control over assignment (as we do in online settings), you can preemptively avoid the pre-treatment differences problem by using a blocking design, where you stratify on factors that correlate strongly with outcomes.
Articles in this section
- What is the best way to ensure my study is completed quickly and accurately?
- Where can I find information on the demographics of Amazon Mechanical Turk workers?
- Can I measure reaction time?
- What is Amazon Mechanical Turk (AMT) and how does it work?
- Do I have to manage participants from AMT?
- Who protects the personal information of AMT workers?
- Can I discover who my participants are?
- How can I be sure my data is secure? Where is the data stored?
- Can I trust the quality of data from AMT?
- I'm satisfied with the way I do research. Why would I use AMT?