|
|
| Non-Interference Assumptions |
Impact on A/B Testing on Web and Mobile |
| 1. Stable Unit Treatment Value Assumption (SUTVA) |
Assumes that the treatment assigned to one unit does not affect the outcome of another unit. In A/B testing, this means that the treatment group and control group should be independent and not influenced by each other. If SUTVA is violated, the results of the experiment may be biased. |
| 2. Ignorability |
Assumes that the treatment assignment is random and not influenced by any other factors that may affect the outcome. In A/B testing, this means that the treatment and control groups should be randomly assigned and not influenced by any external factors. If ignorability is violated, the results of the experiment may be biased. |
| 3. Consistency |
Assumes that the treatment effect is consistent across all units. In A/B testing, this means that the treatment effect should be the same for all users who are exposed to the treatment. If consistency is violated, the results of the experiment may be biased. |
| 4. Compliance |
Assumes that all units comply with the assigned treatment. In A/B testing, this means that all users in the treatment group should receive the treatment and all users in the control group should not receive the treatment. If compliance is violated, the results of the experiment may be biased. |
|
|
|