diffindiff 2.5.3__tar.gz → 2.5.5__tar.gz

{diffindiff-2.5.3 → diffindiff-2.5.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: diffindiff
-Version: 2.5.3
+Version: 2.5.5
 Summary: diffindiff: Python library for convenient Difference-in-Differences analyses
 Author: Thomas Wieland
 Author-email: geowieland@googlemail.com
@@ -29,7 +29,7 @@ A case study that utilizes the diffindiff library is available on [arXiv](https:
 
 If you use this software, please cite:
 
-Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.5.3) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
+Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.5.5) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
 
 
 ## Installation
@@ -164,6 +164,7 @@ See the /tests directory for usage examples of most of the included functions.
   - Goldfarb A, Tucker C, Wang Y (2022) Conducting Research in Marketing with Quasi-Experiments. *Journal of Marketing* 86(3): 1-19. [10.1177/00222429221082977](https://doi.org/10.1177/00222429221082977)
   - Isporhing IE, Lipfert M, Pestel N (2021) Does re-opening schools contribute to the spread of SARS-CoV-2? Evidence from staggered summer breaks in Germany. *Journal of Public Economics* 198: 104426. [10.1016/j.jpubeco.2021.104426](https://doi.org/10.1016/j.jpubeco.2021.104426)
   - Li KT, Luo L, Pattabhiramaiah A (2024) Causal Inference with Quasi-Experimental Data. *IMPACT at JMR* November 13, 2024. [AMA](https://www.ama.org/marketing-news/causal-inference-with-quasi-experimental-data/)
+  - Mitze T, Kosfeld R, Rode J, Wälde K (2020) Face masks considerably reduce COVID-19 cases in Germany. *Proceedings of the National Academy of Sciences of the United States of America* 117(51): 32293-32301. [10.1073/pnas.2015954117](https://doi.org/10.1073/pnas.2015954117)
   - Olden A (2018) What do you buy when no one's watching? The effect of self-service checkouts on the composition of sales in retail. Discussion paper FOR 3/18, Norwegian School of Economics, Norway. [http://hdl.handle.net/11250/2490886](http://hdl.handle.net/11250/2490886)
   - Olden A, Moen J (2022) The triple difference estimator. *The Econometrics Journal* 25(3): 531-553. [10.1093/ectj/utac010](https://doi.org/10.1093/ectj/utac010)
   - Strassmann A, Çolak Y, Serra-Burriel M, Nordestgaard BG, Turk A, Afzal S, Puhan MA (2023) Nationwide indoor smoking ban and impact on smoking behaviour and lung function: a two-population natural experiment. *Thorax* 78(2): 144-150. [10.1136/thoraxjnl-2021-218436](https://doi.org/10.1136/thoraxjnl-2021-218436)
@@ -179,9 +180,8 @@ See the /tests directory for usage examples of most of the included functions.
 This software was developed without the use of AI-generated code. The Continue Agent in Microsoft Visual Studio Code using the GPT-5 mini model (by OpenAI) was used solely to assist in drafting and refining docstrings for documentation. The corresponding guidelines and constraints defined by the author are documented in `AGENTS-docstrings.md` in the [public GitHub repository](https://github.com/geowieland/diffindiff_official).
 
 
-## What's new (v2.5.3)
+## What's new (v2.5.5)
 
-- Extensions
-  - Added optional saving of plots in DiffModel plot methods
 - Bugfixes
-  - Fixed typos in messages
+  - didanalysis.did_analysis(): Correct usage of param 'log_outcome' in any case and check whether value of 'log_outcome_add' is appropriate
+  - didanalysis.DiffModel.plot() and didanalysis.DiffModel.plot_counterfactual(): Correct detection of the date format via object metadata

{diffindiff-2.5.3 → diffindiff-2.5.5}/README.md

@@ -21,7 +21,7 @@ A case study that utilizes the diffindiff library is available on [arXiv](https:
 
 If you use this software, please cite:
 
-Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.5.3) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
+Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.5.5) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
 
 
 ## Installation
@@ -156,6 +156,7 @@ See the /tests directory for usage examples of most of the included functions.
   - Goldfarb A, Tucker C, Wang Y (2022) Conducting Research in Marketing with Quasi-Experiments. *Journal of Marketing* 86(3): 1-19. [10.1177/00222429221082977](https://doi.org/10.1177/00222429221082977)
   - Isporhing IE, Lipfert M, Pestel N (2021) Does re-opening schools contribute to the spread of SARS-CoV-2? Evidence from staggered summer breaks in Germany. *Journal of Public Economics* 198: 104426. [10.1016/j.jpubeco.2021.104426](https://doi.org/10.1016/j.jpubeco.2021.104426)
   - Li KT, Luo L, Pattabhiramaiah A (2024) Causal Inference with Quasi-Experimental Data. *IMPACT at JMR* November 13, 2024. [AMA](https://www.ama.org/marketing-news/causal-inference-with-quasi-experimental-data/)
+  - Mitze T, Kosfeld R, Rode J, Wälde K (2020) Face masks considerably reduce COVID-19 cases in Germany. *Proceedings of the National Academy of Sciences of the United States of America* 117(51): 32293-32301. [10.1073/pnas.2015954117](https://doi.org/10.1073/pnas.2015954117)
   - Olden A (2018) What do you buy when no one's watching? The effect of self-service checkouts on the composition of sales in retail. Discussion paper FOR 3/18, Norwegian School of Economics, Norway. [http://hdl.handle.net/11250/2490886](http://hdl.handle.net/11250/2490886)
   - Olden A, Moen J (2022) The triple difference estimator. *The Econometrics Journal* 25(3): 531-553. [10.1093/ectj/utac010](https://doi.org/10.1093/ectj/utac010)
   - Strassmann A, Çolak Y, Serra-Burriel M, Nordestgaard BG, Turk A, Afzal S, Puhan MA (2023) Nationwide indoor smoking ban and impact on smoking behaviour and lung function: a two-population natural experiment. *Thorax* 78(2): 144-150. [10.1136/thoraxjnl-2021-218436](https://doi.org/10.1136/thoraxjnl-2021-218436)
@@ -171,9 +172,8 @@ See the /tests directory for usage examples of most of the included functions.
 This software was developed without the use of AI-generated code. The Continue Agent in Microsoft Visual Studio Code using the GPT-5 mini model (by OpenAI) was used solely to assist in drafting and refining docstrings for documentation. The corresponding guidelines and constraints defined by the author are documented in `AGENTS-docstrings.md` in the [public GitHub repository](https://github.com/geowieland/diffindiff_official).
 
 
-## What's new (v2.5.3)
+## What's new (v2.5.5)
 
-- Extensions
-  - Added optional saving of plots in DiffModel plot methods
 - Bugfixes
-  - Fixed typos in messages
+  - didanalysis.did_analysis(): Correct usage of param 'log_outcome' in any case and check whether value of 'log_outcome_add' is appropriate
+  - didanalysis.DiffModel.plot() and didanalysis.DiffModel.plot_counterfactual(): Correct detection of the date format via object metadata

{diffindiff-2.5.3 → diffindiff-2.5.5}/diffindiff/config.py

@@ -4,15 +4,15 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     1.0.24
-# Last update: 2026-07-25 12:31
+# Version:     1.0.26
+# Last update: 2026-09-11 21:46
 # Copyright (c) 2025-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
 # Basic config:
 
 PACKAGE_NAME = "diffindiff"
-PACKAGE_VERSION = "2.5.3"
+PACKAGE_VERSION = "2.5.5"
 
 VERBOSE = False
 
@@ -26,6 +26,8 @@ FIXED_EFFECTS_THRESHOLD = 3
 
 AUTO_SKIP_CONSTANT_COLUMNS = True
 
+REMOVE_DUPLICATES_FROM_RESULTS_DICT = True
+
 ACCEPT_CONTINUOUS_TREATMENTS = True
 
 # Description texts:

{diffindiff-2.5.3 → diffindiff-2.5.5}/diffindiff/didanalysis.py

@@ -4,8 +4,8 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     2.4.4
-# Last update: 2026-07-23 19:42
+# Version:     2.4.6
+# Last update: 2026-09-11 21:46
 # Copyright (c) 2024-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
@@ -1875,8 +1875,12 @@ class DiffModel:
                 fig, ax = plt.subplots(figsize=(12, 6))
         else:
             fig, ax = plt.subplots(figsize=(plot_size[0], plot_size[1]))       
-    
-        model_data_TG_CG["t"] = pd.to_datetime(model_data_TG_CG["t"])
+
+        date_format = model_config["date_format"]
+        model_data_TG_CG["t"] = pd.to_datetime(
+            model_data_TG_CG["t"],
+            format = date_format
+            )
 
         if not model_config["pre_post"]:
             pre_post_barplot = False
@@ -2022,7 +2026,7 @@ class DiffModel:
             plt.xlabel(x_label)
             plt.ylabel(y_label)
             plt.title(plot_title)
-            ax.xaxis.set_major_formatter(DateFormatter(model_config["date_format"]))
+            ax.xaxis.set_major_formatter(DateFormatter(date_format))
 
         if model_config["pre_post"]:
 
@@ -2222,7 +2226,12 @@ class DiffModel:
             model_data_TG_mean_pred_counterfac.reset_index()
             ],
             axis = 1)
-        model_data_TG_mean[time_col] = pd.to_datetime(model_data_TG_mean[time_col])
+
+        date_format = model_config["date_format"]
+        model_data_TG_mean[time_col] = pd.to_datetime(
+            model_data_TG_mean[time_col],
+            format=date_format
+            )
 
         fig, ax = plt.subplots(figsize=(plot_size[0], plot_size[1]))   
         
@@ -2512,7 +2521,7 @@ def did_analysis(
     unique_units = treatment_diagnostics_results[3]
     unique_time_points = treatment_diagnostics_results[4]
     
-    if no_treatments > 1:        
+    if no_treatments > 1 and unique_units >= config.FIXED_EFFECTS_THRESHOLD:        
         
         intercept = False
         TG_col = []        
@@ -2725,13 +2734,15 @@ def did_analysis(
         print(f"NOTE: Input is {config.MULTIPERIOD_PANELDATA_DESCRIPTION}. Consider including {config.EFFECTS_TYPES['FE']['description']}.")
         
     if log_outcome:
-        
-        if missing_replace_by_zero:
-            data[f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"] = np.log(data[outcome_col]+log_outcome_add)
-        
+
+        if any(data[outcome_col] <= 0):
+            data[f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"] = np.log(data[outcome_col]+log_outcome_add)        
         else:
             data[f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"] = np.log(data[outcome_col])
-        
+
+        if not np.isfinite(data[f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"]).all():
+            print(f"WARNING: Outcome column '{outcome_col}' in input data contains values <= 0, which is not fully compensated by log_outcome_add = {log_outcome_add}. This may lead to NaN Inf values in the log-transformed outcome. Consider using a larger constant for log_outcome_add.")
+
         outcome_col = f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"
 
     if not ITE and not GTE:
@@ -3320,12 +3331,14 @@ def ddd_analysis(
         pre_post = True
         
     if log_outcome:
-        
-        if missing_replace_by_zero:
-            data[f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"] = np.log(data[outcome_col]+log_outcome_add)
-        
+
+        if any(data[outcome_col] <= 0):
+            data[f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"] = np.log(data[outcome_col]+log_outcome_add)        
         else:
             data[f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"] = np.log(data[outcome_col])
+
+        if not np.isfinite(data[f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"]).all():
+            print(f"WARNING: Outcome column '{outcome_col}' in input data contains values <= 0, which is not fully compensated by log_outcome_add = {log_outcome_add}. This may lead to NaN Inf values in the log-transformed outcome. Consider using a larger constant for log_outcome_add.")
         
         outcome_col = f"{config.LOG_PREFIX}{config.DELIMITER}{outcome_col}"
 

{diffindiff-2.5.3 → diffindiff-2.5.5}/diffindiff/didanalysis_helper.py

@@ -4,8 +4,8 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     1.2.3
-# Last update: 2026-07-14 19:44
+# Version:     1.2.4
+# Last update: 2026-08-04 17:35
 # Copyright (c) 2025-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
@@ -25,7 +25,7 @@ def create_fixed_effects(
     type: str = "unit",
     drop_first: bool = False,
     verbose: bool = config.VERBOSE
-    ):
+    ) -> list:
 
     """
     Create dummy variables for fixed effects and attach them to the data.
@@ -118,7 +118,7 @@ def demean_variables(
     max_iter=10,
     tol=1e-8,
     verbose: bool = config.VERBOSE    
-    ):
+    ) -> tuple:
     
     """
     Demean numeric variables by removing unit and/or time means.
@@ -399,7 +399,7 @@ def create_specific_time_trends(
     FE_vars: list,
     type: str = "ITT",
     verbose: bool = config.VERBOSE
-    ):
+    ) -> list:
 
     """
     Create unit-specific time trend variables (interactions with a time counter).
@@ -482,7 +482,7 @@ def create_specific_treatment_effects(
     FE_vars: list,
     type: str = "ITE",
     verbose: bool = config.VERBOSE
-    ):
+    ) -> list:
 
     """
     Create unit-specific treatment effect interaction variables.
@@ -566,7 +566,7 @@ def create_spillover(
     spillover_treatment: list = None,
     spillover_units: list = None,
     verbose: bool = config.VERBOSE
-    ):
+    ) -> list:
 
     """
     Create spillover indicator variables for given treatments and units.
@@ -667,7 +667,7 @@ def create_interactions(
     data: pd.DataFrame,
     interactions: dict = None,
     verbose: bool = config.VERBOSE
-    ):
+    ) -> list:
 
     """
     Create interaction variables by multiplying specified treatment columns.
@@ -787,7 +787,7 @@ def data_diagnostics(
     drop_missing: bool = True,
     missing_replace_by_zero: bool = False,
     verbose: bool = config.VERBOSE
-    ):
+    ) -> dict:
 
     """
     Run diagnostics on model data: missingness, balance and basic stats.
@@ -906,7 +906,7 @@ def treatment_diagnostics(
     pre_post: bool = False,
     confint_alpha = 0.05,
     verbose: bool = config.VERBOSE    
-    ):
+    ) -> list:
     
     """
     Compute diagnostics for each treatment: simultaneity, parallel trends, 
@@ -1057,7 +1057,7 @@ def ols_fit(
     confint_alpha = 0.05,
     cluster_SE_by: str = None,
     verbose: bool = config.VERBOSE
-    ):
+    ) -> list:
     
     """
     Estimate an OLS model using a formula and return coefficients and predictions.
@@ -1130,7 +1130,7 @@ def ml_fit(
     family=sm.families.Gaussian(),
     link=sm.families.links.Identity(),
     verbose: bool = config.VERBOSE
-    ):
+    ) -> list:
     
     """
     Estimate a model by maximum likelihood (GLM) and return coefficients and predictions.
@@ -1217,7 +1217,7 @@ def extract_model_results(
     BG_x_TT_col: list = None,
     covariates: list = None,
     verbose: bool = config.VERBOSE
-    ):
+    ) -> dict:
 
     """
     Compile and format model coefficient results into a structured dict.
@@ -1311,6 +1311,9 @@ def extract_model_results(
                 "CI_lower": float(coef_conf_intervals.loc[treatment, 0]),
                 "CI_upper": float(coef_conf_intervals.loc[treatment, 1]),
                 }
+            
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            ATE = remove_duplicates_from_dict(ATE)
         
         model_results = {config.EFFECTS_TYPES["ATE"]["model_results_key"]: ATE}
     
@@ -1329,6 +1332,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[TG_, 1]),
                 }
                         
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            beta_1 = remove_duplicates_from_dict(beta_1)
+            
         model_results[config.EFFECTS_TYPES["beta_1"]["model_results_key"]] = beta_1
 
     if (len(TT_col) > 0) and (any(col in coefficients for col in TT_col)):
@@ -1346,6 +1352,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[TT_, 1]),
                 }
         
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            delta_0 = remove_duplicates_from_dict(delta_0)
+        
         model_results[config.EFFECTS_TYPES["delta_0"]["model_results_key"]] = delta_0
 
     if "Intercept" in coefficients:
@@ -1377,6 +1386,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[AATE_, 1]),
                 }
             
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            AATE = remove_duplicates_from_dict(AATE)
+            
         model_results[config.EFFECTS_TYPES["AATE"]["model_results_key"]] = AATE
     
     if (len(ATT_col) > 0) and (any(col in coefficients for col in ATT_col)):
@@ -1394,6 +1406,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[ATT_, 1]),
                 }
                        
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            ATT = remove_duplicates_from_dict(ATT)
+            
         model_results[config.EFFECTS_TYPES["ATT"]["model_results_key"]] = ATT    
     
     if (len(spillover_vars) > 0) and (any(col in coefficients for col in spillover_vars)):
@@ -1411,6 +1426,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[spillover_var, 1]),
                 } 
                        
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            spillover_coef = remove_duplicates_from_dict(spillover_coef)
+            
         model_results[config.EFFECTS_TYPES["spillover"]["model_results_key"]] = spillover_coef
     
     fixed_effects = [
@@ -1435,6 +1453,9 @@ def extract_model_results(
                 "Coefficient_type": config.EFFECTS_TYPES["FE"]["types"][0]["description"]
                 }
             
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            FE_unit_coef = remove_duplicates_from_dict(FE_unit_coef)
+            
         fixed_effects[0] = {config.EFFECTS_TYPES["FE"]["types"][0]["model_results_key"]: FE_unit_coef}
     
     if (len(FE_time_vars) > 0) and (any(col in coefficients for col in FE_time_vars)):
@@ -1452,7 +1473,10 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[time_dummy, 1]),
                 "Coefficient_type": config.EFFECTS_TYPES["FE"]["types"][1]["description"]
                 }
-            
+        
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            FE_time_coef = remove_duplicates_from_dict(FE_time_coef)
+                    
         fixed_effects[1] = {config.EFFECTS_TYPES["FE"]["types"][1]["model_results_key"]: FE_time_coef}
         
     if (len(FE_group_vars) > 0) and (any(col in coefficients for col in FE_group_vars)):
@@ -1470,7 +1494,10 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[group_dummy, 1]),
                 "Coefficient_type": config.EFFECTS_TYPES["FE"]["types"][2]["description"]
                 }
-            
+        
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            FE_group_coef = remove_duplicates_from_dict(FE_group_coef)
+
         fixed_effects[2] = {config.EFFECTS_TYPES["FE"]["types"][2]["model_results_key"]: FE_group_coef}
 
     model_results[config.EFFECTS_TYPES["FE"]["model_results_key"]] = fixed_effects
@@ -1491,6 +1518,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[ITT_var, 1]),
                 }      
         
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            ITT_coef = remove_duplicates_from_dict(ITT_coef)
+        
         model_results["individual_time_trends"] = ITT_coef
 
     if (len(ITE_vars) > 0) and (any(col in coefficients for col in ITE_vars)):
@@ -1516,6 +1546,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[ITE_var, 1]),
                 }            
             
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            ITE_coef = remove_duplicates_from_dict(ITE_coef)
+            
         model_results["individual_treatment_effects"] = ITE_coef
 
     if (len(GTT_vars) > 0) and (any(col in coefficients for col in GTT_vars)):
@@ -1534,6 +1567,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[GTT_var, 1]),
                 }      
         
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            GTT_coef = remove_duplicates_from_dict(GTT_coef)
+
         model_results["group_time_trends"] = GTT_coef
 
     if (len(GTE_vars) > 0) and (any(col in coefficients for col in GTE_vars)):
@@ -1558,6 +1594,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[GTE_var, 1]),
                 }      
         
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            GTE_coef = remove_duplicates_from_dict(GTE_coef)
+
         model_results["group_treatment_effects"] = GTE_coef
 
     if (len(covariates) > 0) and (any(col in coefficients for col in covariates)):
@@ -1577,6 +1616,9 @@ def extract_model_results(
                     "CI_lower": float(coef_conf_intervals.loc[covariate, 0]),
                     "CI_upper": float(coef_conf_intervals.loc[covariate, 1]),
                     }
+            
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            covariates_effects = remove_duplicates_from_dict(covariates_effects)
                 
         model_results["covariates_effects"] = covariates_effects    
 
@@ -1595,6 +1637,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[TG_x_BG_x_TT_, 1]),
                 }
         
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            TDATE = remove_duplicates_from_dict(TDATE)
+
         model_results = {config.EFFECTS_TYPES_DDD["TDATE"]["model_results_key"]: TDATE}
     
     if (len(BG_col) > 0) and (any(col in coefficients for col in BG_col)):
@@ -1612,6 +1657,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[BG_, 1]),
                 }            
 
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            BG = remove_duplicates_from_dict(BG)
+
         model_results[config.EFFECTS_TYPES_DDD["beta_2"]["model_results_key"]] = BG
         
     if (len(TG_x_BG_col) > 0) and (any(col in coefficients for col in TG_x_BG_col)):
@@ -1627,7 +1675,10 @@ def extract_model_results(
                 "p": float(coef_p[TG_x_BG_]),
                 "CI_lower": float(coef_conf_intervals.loc[TG_x_BG_, 0]),
                 "CI_upper": float(coef_conf_intervals.loc[TG_x_BG_, 1]),
-                }            
+                }
+            
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            TG_x_BG = remove_duplicates_from_dict(TG_x_BG)
 
         model_results[config.EFFECTS_TYPES_DDD["beta_4"]["model_results_key"]] = TG_x_BG
 
@@ -1646,6 +1697,9 @@ def extract_model_results(
                 "CI_upper": float(coef_conf_intervals.loc[BG_x_TT_, 1]),
                 }            
 
+        if config.REMOVE_DUPLICATES_FROM_RESULTS_DICT:
+            BG_x_TT = remove_duplicates_from_dict(BG_x_TT)
+
         model_results[config.EFFECTS_TYPES_DDD["beta_6"]["model_results_key"]] = BG_x_TT
     
     if verbose:
@@ -1658,7 +1712,7 @@ def fit_metrics(
     outcome_col,
     model_predictions,
     indep_vars_no: int = None
-    ):
+    ) -> list:
 
     """
     Compute fit metrics for a model given data and predictions.
@@ -1704,7 +1758,7 @@ def fit_metrics(
 
     return fit_metrics_result
 
-def create_timestamp(function):
+def create_timestamp(function) -> dict:
 
     """
     Create a standard timestamp dictionary for logging or metadata.
@@ -1743,4 +1797,35 @@ def create_timestamp(function):
         "datetime": now.strftime("%Y-%m-%d %H-%M-%S")
     }
 
-    return timestamp_dict
+    return timestamp_dict
+
+def remove_duplicates_from_dict(
+    any_dict: dict
+    ) -> dict:
+    
+    """
+    Remove duplicate values from a dictionary.
+
+    Parameters
+    ----------
+    any_dict : dict
+        Dictionary from which to remove duplicates.
+
+    Returns
+    -------
+    dict
+        Dictionary with unique values.
+    """
+    
+    any_dict_unique = {}
+    seen = set()
+
+    for key, value in any_dict.items():
+
+        identifier = tuple(sorted(value.items()))
+
+        if identifier not in seen:
+            seen.add(identifier)
+            any_dict_unique[key] = value
+
+    return any_dict_unique

{diffindiff-2.5.3 → diffindiff-2.5.5}/diffindiff/diddata.py

@@ -4,8 +4,8 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     2.3.2
-# Last update: 2026-07-16 18:52
+# Version:     2.3.3
+# Last update: 2026-08-04 17:51
 # Copyright (c) 2024-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
@@ -1986,10 +1986,11 @@ class DiffData:
                 }
             )
 
+        columns = {}
         units_not_included = []
+        valid_units = []
 
         for unit in units:
-            
             did_modeldata_unit = did_modeldata.loc[
                 (did_modeldata[config.UNIT_COL].astype(str) == str(unit))
                 & (did_modeldata[outcome_col].notna()),
@@ -1998,9 +1999,19 @@ class DiffData:
 
             if len(did_modeldata_unit) != len(outcome_matrix):
                 units_not_included.append(unit)
-                units.remove(unit)
             else:
-                outcome_matrix[tools.clean_column_name(unit)] = did_modeldata_unit.reset_index(drop=True)
+                valid_units.append(unit)
+                columns[tools.clean_column_name(unit)] = did_modeldata_unit.reset_index(drop=True)
+
+        outcome_matrix = pd.concat(
+            [
+                outcome_matrix, 
+                pd.DataFrame(columns)
+            ],
+            axis=1
+            )
+
+        units = valid_units
 
         if len(units_not_included) > 0:
             print(f"WARNING: {len(units_not_included)} analysis units were not included due to NaN data.")

{diffindiff-2.5.3 → diffindiff-2.5.5}/diffindiff/tests/tests_diffindiff.py

@@ -4,16 +4,18 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     2.1.0
-# Last update: 2026-07-10 11:35
+# Version:     2.2.0
+# Last update: 2026-08-04 18:42
 # Copyright (c) 2025-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
+
 """
 Please note that this tests script is a collection of examples for the use of most functions of the diffindiff package. 
 Although the data is real (i.e., non-simulated), no results relevant to the real world can be derived from it.
 """
 
+from pathlib import Path
 import pandas as pd
 from diffindiff.didanalysis import did_analysis
 from diffindiff.diddata import create_groups, create_treatment, merge_data, create_data
@@ -22,7 +24,11 @@ from diffindiff.diddata import create_groups, create_treatment, merge_data, crea
 # Example 1: Effect of a curfew in German counties in the first
 # wave of the COVID-19 pandemic (DiD pre-post analysis)
 
-curfew_DE=pd.read_csv("data/curfew_DE.csv", sep=";", decimal=",")
+curfew_DE = pd.read_csv(
+    Path(__file__).parent / "data" / "curfew_DE.csv",
+    sep=";",
+    decimal=","
+)
 # Dataset with daily and cumulative SARS-CoV-2 infections of German counties
 # Data source: Wieland (2020) https://doi.org/10.18335/region.v7i2.324
 
@@ -193,7 +199,12 @@ curfew_model_prepost_AT.plot(
 # Plot DiD pre vs. post results
 # with user-determined style
 
-counties_DE=pd.read_csv("data/counties_DE.csv", sep=";", decimal=",", encoding='latin1')
+counties_DE = pd.read_csv(
+    Path(__file__).parent / "data" / "counties_DE.csv",
+    sep=";",
+    decimal=",",
+    encoding='latin1'
+)
 # Dataset with German county data
 
 curfew_data_prepost_withcov = curfew_data_prepost.add_covariates(
@@ -405,7 +416,9 @@ curfew_model_extended.plot(
 # Example 3: Nighttime curfew and other NPI in Hesse
 # (Staggered adoption)
 
-Corona_Hesse=pd.read_excel("data/Corona_Hesse.xlsx")
+Corona_Hesse = pd.read_excel(
+    Path(__file__).parent / "data" / "Corona_Hesse.xlsx",
+)
 # Test data effective reproduction number and Corona NPI Hesse
 # Data source: Wieland (2025) https://doi.org/10.1007/s10389-024-02218-x
 
@@ -530,4 +543,75 @@ Hesse_model6=did_analysis(
 # Model with two interventions and one interaction of the two treatments
 
 Hesse_model6.summary()
-# Model summary
+# Model summary
+
+
+# Example 4: Mandatory face masks in Jena (Germany) during the first Corona wave
+# data: German counties during the first Corona wave
+
+facemasks_data=create_data(
+    outcome_data=curfew_DE,
+    unit_id_col="county",
+    time_col="infection_date",
+    outcome_col="infections_cum_per100000",
+    treatment_group= 
+        curfew_DE.loc[curfew_DE["REG_NAME"] == "Jena"]["county"],
+    control_group= 
+        curfew_DE.loc[curfew_DE["REG_NAME"] != "Jena"]["county"],
+    treatment_name="Mandatory face masks",
+    study_period=["2020-03-01", "2020-04-26"],
+    treatment_period=["2020-04-06", "2020-04-26"],
+    freq="D"
+    )
+# Creating DiD dataset by defining groups and treatment time at once
+# Treatment mandatory face masks in Jena beginning April 6, 2020 (first German city with mandatory face masks)
+
+facemasks_data.summary()
+# Summary of created treatment data
+
+facemasks_data_analysis = facemasks_data.analysis(
+    FE_unit=True,
+    FE_time=True,
+    intercept=False,
+    verbose=True
+    )
+# DiD analysis as two-way fixed effects model
+
+facemasks_data_analysis.summary()
+# Summary of model results
+
+
+facemasks_data_analysis_demean = facemasks_data.analysis(
+    FE_unit=True,
+    FE_time=True,
+    intercept=False,
+    demean=True,
+    verbose=True
+    )
+# DiD analysis as model with demeaened variables instead of fixed effects
+
+facemasks_data_analysis_demean.summary()
+# Summary of model results
+
+
+facemasks_data_synth = facemasks_data.add_synthetic(process_unit="Jena")
+# Add synthetic control for Jena
+
+facemasks_data_synth.summary()
+# Summary of synthetic DiD data
+
+print(facemasks_data_synth.get_synthetic_control_weightings())
+# Weights
+
+print(facemasks_data_synth.data[0].head)
+# Treatment unit with counterfactual (synthetic) control unit
+
+facemasks_data_synth_analysis = facemasks_data_synth.analysis(
+    FE_time=True,
+    log_outcome=True,
+    verbose=True
+    )
+# Synthetic DiD analysis 
+
+facemasks_data_synth_analysis.summary()
+# Summary of model results

{diffindiff-2.5.3 → diffindiff-2.5.5}/diffindiff.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: diffindiff
-Version: 2.5.3
+Version: 2.5.5
 Summary: diffindiff: Python library for convenient Difference-in-Differences analyses
 Author: Thomas Wieland
 Author-email: geowieland@googlemail.com
@@ -29,7 +29,7 @@ A case study that utilizes the diffindiff library is available on [arXiv](https:
 
 If you use this software, please cite:
 
-Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.5.3) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
+Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.5.5) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
 
 
 ## Installation
@@ -164,6 +164,7 @@ See the /tests directory for usage examples of most of the included functions.
   - Goldfarb A, Tucker C, Wang Y (2022) Conducting Research in Marketing with Quasi-Experiments. *Journal of Marketing* 86(3): 1-19. [10.1177/00222429221082977](https://doi.org/10.1177/00222429221082977)
   - Isporhing IE, Lipfert M, Pestel N (2021) Does re-opening schools contribute to the spread of SARS-CoV-2? Evidence from staggered summer breaks in Germany. *Journal of Public Economics* 198: 104426. [10.1016/j.jpubeco.2021.104426](https://doi.org/10.1016/j.jpubeco.2021.104426)
   - Li KT, Luo L, Pattabhiramaiah A (2024) Causal Inference with Quasi-Experimental Data. *IMPACT at JMR* November 13, 2024. [AMA](https://www.ama.org/marketing-news/causal-inference-with-quasi-experimental-data/)
+  - Mitze T, Kosfeld R, Rode J, Wälde K (2020) Face masks considerably reduce COVID-19 cases in Germany. *Proceedings of the National Academy of Sciences of the United States of America* 117(51): 32293-32301. [10.1073/pnas.2015954117](https://doi.org/10.1073/pnas.2015954117)
   - Olden A (2018) What do you buy when no one's watching? The effect of self-service checkouts on the composition of sales in retail. Discussion paper FOR 3/18, Norwegian School of Economics, Norway. [http://hdl.handle.net/11250/2490886](http://hdl.handle.net/11250/2490886)
   - Olden A, Moen J (2022) The triple difference estimator. *The Econometrics Journal* 25(3): 531-553. [10.1093/ectj/utac010](https://doi.org/10.1093/ectj/utac010)
   - Strassmann A, Çolak Y, Serra-Burriel M, Nordestgaard BG, Turk A, Afzal S, Puhan MA (2023) Nationwide indoor smoking ban and impact on smoking behaviour and lung function: a two-population natural experiment. *Thorax* 78(2): 144-150. [10.1136/thoraxjnl-2021-218436](https://doi.org/10.1136/thoraxjnl-2021-218436)
@@ -179,9 +180,8 @@ See the /tests directory for usage examples of most of the included functions.
 This software was developed without the use of AI-generated code. The Continue Agent in Microsoft Visual Studio Code using the GPT-5 mini model (by OpenAI) was used solely to assist in drafting and refining docstrings for documentation. The corresponding guidelines and constraints defined by the author are documented in `AGENTS-docstrings.md` in the [public GitHub repository](https://github.com/geowieland/diffindiff_official).
 
 
-## What's new (v2.5.3)
+## What's new (v2.5.5)
 
-- Extensions
-  - Added optional saving of plots in DiffModel plot methods
 - Bugfixes
-  - Fixed typos in messages
+  - didanalysis.did_analysis(): Correct usage of param 'log_outcome' in any case and check whether value of 'log_outcome_add' is appropriate
+  - didanalysis.DiffModel.plot() and didanalysis.DiffModel.plot_counterfactual(): Correct detection of the date format via object metadata

{diffindiff-2.5.3 → diffindiff-2.5.5}/diffindiff.egg-info/requires.txt

@@ -1,4 +1,3 @@
-geopandas
 pandas
 numpy
 statsmodels>=0.14.5
@@ -6,9 +5,6 @@ scipy>=1.17
 scikit-learn
 xgboost
 lightgbm
-shapely
-requests<3.0
-contextily
 openpyxl
 matplotlib
 patsy

{diffindiff-2.5.3 → diffindiff-2.5.5}/setup.py

@@ -7,7 +7,7 @@ def read_README():
     
 setup(
     name='diffindiff',
-    version='2.5.3',
+    version='2.5.5',
     description='diffindiff: Python library for convenient Difference-in-Differences analyses',
     packages=find_packages(include=["diffindiff", "diffindiff.tests"]),
     include_package_data=True,
@@ -20,7 +20,6 @@ setup(
         'diffindiff': ['tests/data/*'],
     },
     install_requires=[
-        'geopandas',
         'pandas',
         'numpy',
         'statsmodels>=0.14.5',
@@ -28,9 +27,6 @@ setup(
         'scikit-learn',
         'xgboost',
         'lightgbm',
-        'shapely',
-        'requests<3.0',
-        'contextily',
         'openpyxl',
         'matplotlib',
         'patsy',