Selection Bias in Startup Research: Read It Honestly.
Selection bias in startup research is the worry that the overlap advantage reflects who chooses it. Raffiee and Feng found risk-averse people more likely to pick the hybrid path. Also, the twelve weeks come from practice, not from any tested study.
Twelve weeks is practice, not evidence.
Strategyzer runs twelve experiments in twelve weeks, and venture studios use sprints of four to twelve weeks. Also, Ash Maurya argues ninety days beats a fortnight for measuring a business model. However, no study compares twelve weeks with eight, sixteen or twenty-four. So is 90 days enough to validate an idea? Nobody has tested it, nor thresholds fixed in advance.
Selection bias in startup research and the cautious founder.
Think about who picks the staged route: risk averse entrepreneurs, heavy preparers, people unsure they are brilliant. Then ask whether they last because they overlapped or because they rarely close businesses. Jörn Block and Christian Landgraf also showed hybrids differ in growth ambitions and motives. As a result, nothing yet separates the treatment from the selection.
Weighing the objection honestly.
The selection objection is probably half right, because cautious people do choose overlap. However, it cannot explain why prior experience sharpens the effect. Also, no failed replication of the thirty-three per cent turned up. In short, overlap is associated with lasting longer, and even the sceptical reading does not recommend leaping.
Test selection bias in startup research against yourself.
Write down whether you are choosing overlap from caution or from a plan to learn.
Note one part of your trial you will treat as practice, not proven fact.
From the book.
Twelve weeks is not an evidence-based number. It comes from practice rather than from research. Strategyzer runs twelve experiments in twelve weeks. Venture studios use validation sprints of four to twelve weeks before deciding whether to continue, and Ash Maurya argues that ninety days is a better rhythm for measuring a business model than a fortnight. That's consensus among people who do this for a living. It's worth something. It is not the same as evidence.
I Want to Start a Business but I Can't Quit My Job, chapter 9: Do not leap. overlap.
Questions about selection bias in startup research.
What does selection bias in startup research mean here?
It is the chance that overlappers last longer because of their personality, not because they overlapped.
Is the twelve week trial evidence-based?
No, it comes from practice among people who do this for a living. Nobody has run the comparison study.
Should I still run the trial?
Yes, exactly as written, knowing you follow a well-reasoned practice rather than a proven result.
Read the whole chapter.
This lesson covers one section of chapter 9, “Do not leap. overlap.”. The book goes further, with the full evidence, the stories and every exercise.
Lessons in this chapter.
- 01 How Many Hours for a Side Business: The 96 Hour Trial.
- 02 Hybrid Entrepreneurship Study: Overlap Lasts Longer.
- 03 12 Week Business Experiment: One Number, One Stop.
- 04 Selection Bias in Startup Research: Read It Honestly.
- 05 Downsides of a Part Time Business: The Salary Shield.
- 06 Quit Job After Product Launch: Order Your Decisions.
- 07 Quit Job to Launch a Startup: Two Routes Compared.
- 08 No Time for a Side Hustle: Fixing a Trial That Slips.