Compiti ed esercitazioni VERIFICATO

Modello diff in diff tutorial con stata - Mario Mazzocchi pt.1

Università degli studi di Bologna health economics and management / economia sanitaria e management Curriculum health care finance and management 2020
127 visualizzazioni
Nessun voto ancora
Condividi: WhatsApp Telegram

Di cosa parla

Riassunto AI · si legge in un minuto

Based on the Stata output logs you provided, here is a structured analysis of your econometric results regarding the **Minimum Drinking Age (MDA)** policy and its effect on mortality rates (`mrate`). ### 1. Data Preparation & Model Structure * **Dataset:** You are analyzing state-level data from `use MDA`. * **Variables:** * `legal`: Dummy variable for whether the Minimum Drinking Age is in place (likely coded as $age \geq 21$). * `year`: Time trend. * `state` & `agegr`: Used to create a panel dataset (`crossid`) combining state and age group, observed from **1970 to 1996**.
Leggi il riassunto completo
Based on the Stata output logs you provided, here is a comprehensive analysis of the results regarding the **Minimum Drinking Age (MDA)** policy and its impact on mortality rates (`mrate`). ### 1. Data Preparation & Model Structure * **Dataset:** The data combines state-level information with age groups (15-17, 18-20, 21+). * **Filtering:** You filtered for `dtype=2` (likely states where the MDA policy was active or relevant), reducing observations from ~655 to 4,131. * **Panel Structure:** Created a cross-sectional unit (`crossid`) by grouping `state` and `agegr`, then set up as an unbalanced panel over time (years). * **Variables of Interest:** * `legal`: Indicator for the Minimum Drinking Age policy implementation. * `i.year`: Year dummies to capture linear trends or specific year effects. --- ### 2. Model Comparison: Linear Trends vs. Interaction Terms You ran two main Fixed Effects (FE) models: #### **Model A: With Linear Time Trend (`reg mrate legal i.year, fe`)** * **Specification:** Includes `legal`, individual year dummies for every single year from the start to end of your sample, and state/age fixed effects. * **Key Finding (Legal):** Coefficient = **4.98** ($p < 0.001$). This suggests that after controlling for specific yearly fluctuations, raising the drinking age is associated with a significant *increase* in mortality rates relative to the baseline year effect included here. * *Note:* The large negative coefficients on individual years (e.g., -23.5 in 1984) suggest strong secular trends or omitted variable bias if not modeled correctly, but since you have dummies for every single year, these are just capturing the specific deviation of that year from the mean within groups. * **Issue:** In this specification, `legal` is picking up the average effect across all years because each year has its own dummy absorbing the time trend. #### **Model B: With Interaction Terms (`reg mrate legal i.year c.yeart1.state, fe`)** * *Correction on your command:* You typed `c.yearti. state`. Based on standard practice and the error message "56.stavefc.

Riassunto generato con l'intelligenza artificiale a partire dagli appunti. Come lo facciamo

Anteprima pagina 1 — Modello diff in diff tutorial con stata - Mario Mazzocchi pt.1 Anteprima pagina 2 — Modello diff in diff tutorial con stata - Mario Mazzocchi pt.1

Stai vedendo l'anteprima delle prime pagine. Il file completo è gratis: registrati per leggerlo tutto.

Studia questi appunti con l'AI

Riassunto e domande d'esame sono gratis: chiedili e ti mandiamo un'email quando sono pronti.

Questo appunto è gratis. Registrati in 30 secondi per leggere tutte le pagine e scaricarlo.

Simula l'esame con domande costruite su questi appunti.

Altri appunti di FUNDAMENTALS OF STATISTICS [cod. 79290]

Condividi questi appunti

WhatsApp Telegram