How many participants do you need?

This planner is for experiments where each participant only sees part of a large stimulus set, such as similarity ratings, Multiple Arrangement or triplet tasks. Meadows can give every participant a different subset, chosen so that each stimulus or pair gets observed about equally often (balanced sampling). The planner estimates how many participants that takes, including dropout, by simulating your study many times with the sampling code Meadows uses.

The numbers assume balanced sampling as in Meadows. Picking subsets at random needs more participants for the same result.

Stimuli

Individual stimuli Most other tasks, such as ratings and key responses

Pairs of stimuli Multiple Arrangement, Triplet Choice, Drag & Rate, Discriminability and other tasks that sample pairs

Trials of three stimuli Each participant gets a number of trials of three stimuli, chosen so that every pair comes up equally often

Goal

Every pair

At least 99% of pairs

At least 95% of pairs

80% of simulated studies

90% of simulated studies

95% of simulated studies

Recruitment

Pilot, then top-up batches

Fixed batch size

Fixed number of batches

In Meadows you recruit in batches, each with a number of slots: one per participant. Stimuli are assigned when you create a batch, and stimuli assigned to people who drop out only get assigned again in the next batch, so several smaller batches use your participants better than one big one.

Fewer rounds of recruitment

Fewer participants in total

Calculate

Fill in your design and select Calculate. The plan appears here: how many participant slots to create, how many people will complete, and how to split them into batches.