Randomization Procedures Sample Clauses

Randomization Procedures. The goal is to have a randomly assigned study sample of approximately 3,000 participants in five groups balanced within a 5% range. Since chance may not lead to balance, we developed a type of blocking that preserves random assignment while assuring a balanced sample through the duration of the study. Using the server clock, the decisecond (0,1,2,3,4,5,6,7,8 or 9) when a participant clicks their mouse to go to the “branch” page of the survey is captured. This “draw” is mapped to the 5 branch groups (a draw of 5-10 is “re-drawn” until a 0-4 comes up), giving us a random draw of 5 numbers. Each half-second, a participant has a 1 in 5 chance of being assigned to any given branch. When 100 participants have been through the draw, we begin an evaluation procedure to determine the balance among the groups that has been achieved by chance. This process: Counts the number assigned to each group Determines each group’s percentage of the whole Designates each group as: “high,” meaning it contains more than 22.5% of the participants “low,” meaning it contains less than 17.5% of the participants Determines if the block (100 records at this point) is “balanced” or “not balanced” within the objective 5% range If a group (or groups) is “high,” that number is taken out of the “draw” until that group no longer represents more than 22.5% of the whole. This means that during this condition, participants are being randomly assigned to the remaining groups. If one group is “high,” participants have a 1 in 4 chance of being assigned to each of the other groups. If a group (or groups) is “low,” that number is “emphasized,” meaning an additional number which results in an assignment to that group is added to the “draw” (from the unused 5-9), effectively increasing the chances of assignment to that group from 1 in 5 to 1 in 3. If two groups are emphasized, the chances increase to 1 in 3.5. Again, assignments continue to be made to all groups randomly. (No more than two groups would ever be emphasized at the same time.) This “nudging” up of low groups and down of high groups continues until the block becomes “balanced” with each group having between 17.5%-22.5% of the whole. Once “balance” is achieved, the next block begins, and the procedure starts again when the next 100 participants have been assigned. The key point is that random assignment continues through the balancing process. This is preferable to fully preventing assignment to full groups while waiting for the others to “...