Research using ecological momentary assessment often asks participants to respond to repeated prompts (often 40-120 times) and answer the same questions repeatedly over a short time period. This method introduces high participant burden. Without large incentive structures (i.e. high compensation), participants routinely do not complete a substantial portion of these potential prompts, resulting in large proportions of missing data. A widespread practice in ecological momentary assessment research is to discard data from persons who have missed 20-30% or more of these prompts. In applying these arbitrary compliance thresholds, researchers are severely biasing their model estimates. Researchers are often aware of the pitfalls of applying listwise deletion in other contexts. However, compliance thresholds are far worse than listwise deletion. In this widespread practice, researchers discard complete data, and consequently bias the ability to study their desired phenomena of interest. The current presentation will show the results of a simulation study, along with an empirical illustration of data using ecological momentary assessment to study anxiety and depression in daily life. The results demonstrate that true parameters can be reliably estimated in the presence of very large proportions of missing values (70-90% missing) by utilizing techniques designed to handle missing data (i.e. full information maximum likelihood and multiple imputation)..