I have a dataset with variables for year, city, indicator, and indicator_options. I have aggregated them together to get the count of all those for indicator_options within the indicator, city and year. I am trying to suppress data that is between 1-5 and then also considering secondary/complementary suppression so that if there is only one missing value, the next lowest value is suppressed. I have been able to do the first round of suppression but am getting stuck with the secondary suppression.   I asked ChatGPT, which gave me a pretty good solution:     DATA trends; SET trends; BY options indicator city year; /* Step 1: Initialize variables for tracking */ RETAIN missing_count next_min_value; /* Initialize for the first record in each group */ IF first.year THEN DO; missing_count = 0; next_min_value = .; END; /* Step 2: Count the missing values in the group */ IF count_supp = . THEN missing_count + 1; /* Step 3: Find the next lowest non-missing value (next smallest value) */ IF count_supp NE . THEN DO; IF next_min_value = . OR count_supp < next_min_value THEN next_min_value = count_supp; END; /* Step 4: Modify the next smallest non-missing value if exactly one missing value exists in the group */ IF last.year AND missing_count = 1 THEN DO; IF count_supp = next_min_value THEN count_supp = .; /* Set the next smallest non-missing value to missing */ END; /* Step 5: Output the modified rows */ count_supp_new = count_supp; /* Create the new column for modified values */ OUTPUT; /* Drop temporary variables */ DROP missing_count next_min_value; RUN;   However, the problem arises with Step 4 which seems to just be looking at the last row, not the next lowest value within the group. I can't figure out how to modify the code so that it considers the whole group.   Here is the output with the code: year city indicator options count count_supp count_supp_new 2017 Paris job title Consultant 8 8 8 2019 Paris job title Consultant 6 6 6 2021 Paris job title Consultant 4 . . 2023 Paris job title Consultant 8 8 .   Here is what it should look like with the next lowest value marked as missing/suppressed. year city indicator options count count_supp count_supp_new 2017 Paris job title Consultant 8 8 8 2019 Paris job title Consultant 6 6 . 2021 Paris job title Consultant 4 . . 2023 Paris job title Consultant 8 8 8 ... View more