Dear all Stata users,
I am struggling with the interpretation of my diff-in-diff impact. I would be really glad if someone could confirm / criticize my two interpretations.
A )
Please, consider the model (5) of the following table.
The regressor patrilineal indicates a percentage share of patrilineal people living within a certain region: some regions possesses 0 (no one in the region follows patrilineal kinship system), other 1 (the whole region is patrilineal) and some for example 0.8622 (86.22% of the population follow patrilineal kinship system).
The dummy variable After
= 1 if one consider years after implementation of Land Acts of 1999
= 0 if one consider years before 1999
Impact = patrilineal * After
Is the following interpretation correct? Note: my outcome variable is the logarithm of wealth score.
Compared to matrilineal and bilateral societies, the average household’s wealth score of patrilineal societies reduces by 8.96262 units after the passage of Land Acts of 1999. This effects is statistically significant on the 1 per cent significance level.
B)
Again, please consider the model (5). Now, the outcome variable is percentage share of children that a mother lost:
Now, is the following interpretation correct?
Compared to matrilineal and bilateral societies, the average child mortality of patrilineal societies reduces by 0.0119 units = 1.2% per cent ? (as my outcome variable is already a rational number in the form of 0.5 or 0.125)
5/10 = 0.5 (if 5 from 10 children died)
1/8 = 0.125 (if 1 from 8 children died)
Thank you for your time and help.
Kind regards, Veronika
I am struggling with the interpretation of my diff-in-diff impact. I would be really glad if someone could confirm / criticize my two interpretations.
A )
Please, consider the model (5) of the following table.
| Logarithm of wealth score | (1) | (2) | (3) | (4) | (5) |
| Patrilineal | -0.160*** | -0.373*** | -0.345*** | -0.146*** | -0.134*** |
| (0.0112) | (0.0241) | (0.0245) | (0.0230) | (0.0192) | |
| After | -0.0319*** | 0.000325 | 0.0302*** | 0.0302*** | 0.0591*** |
| (0.00866) | (0.00750) | (0.00863) | (0.00863) | (0.00720) | |
| Impact | -0.0197 | -0.0417*** | -0.0475*** | -0.0475*** | -0.0939*** |
| (0.0139) | (0.0125) | (0.0126) | (0.0126) | (0.0107) | |
| Constant | 0.734*** | 0.860*** | 0.867*** | -0.00177 | 0.379** |
| (0.00721) | (0.00942) | (0.00939) | (0.204) | (0.171) | |
| Population | n | n | y | y | y |
| Regional character. | n | n | n | y | y |
| Personal character. | n | n | n | n | y |
| Regional FE | n | y | y | y | y |
| Observations | 23,420 | 23,420 | 23,420 | 23,420 | 23,420 |
| R-squared | 0.030 | 0.214 | 0.216 | 0.216 | 0.446 |
The regressor patrilineal indicates a percentage share of patrilineal people living within a certain region: some regions possesses 0 (no one in the region follows patrilineal kinship system), other 1 (the whole region is patrilineal) and some for example 0.8622 (86.22% of the population follow patrilineal kinship system).
The dummy variable After
= 1 if one consider years after implementation of Land Acts of 1999
= 0 if one consider years before 1999
Impact = patrilineal * After
Is the following interpretation correct? Note: my outcome variable is the logarithm of wealth score.
Compared to matrilineal and bilateral societies, the average household’s wealth score of patrilineal societies reduces by 8.96262 units after the passage of Land Acts of 1999. This effects is statistically significant on the 1 per cent significance level.
B)
Again, please consider the model (5). Now, the outcome variable is percentage share of children that a mother lost:
Code:
share of dead children = number of children died / total number of children born
| Share of dead children | (1) | (2) | (3) | (4) | (5) | |
| Patrilineal | 0.00248 | 0.0456*** | 0.0583*** | -0.0315 | -0.0327 | |
| (0.00686) | (0.0108) | (0.0110) | (0.0247) | (0.0246) | ||
| After | -0.0400*** | -0.0370*** | -0.0237*** | -0.0237*** | -0.0270*** | |
| (0.00438) | (0.00444) | (0.00490) | (0.00490) | (0.00487) | ||
| Impact | -0.00662 | -0.0126 | -0.0151* | -0.0151* | -0.0119 | |
| (0.00766) | (0.00772) | (0.00773) | (0.00773) | (0.00771) | ||
| Constant | 0.129*** | 0.113*** | 0.116*** | 0.310** | 0.193 | |
| (0.00391) | (0.00449) | (0.00451) | (0.150) | (0.150) | ||
| Population | n | n | y | y | y | |
| Regional character. | n | n | n | y | y | |
| Individual character. | n | n | n | n | y | |
| Regional FE | n | y | y | y | y | |
| Observations | 23,420 | 23,420 | 23,420 | 23,420 | 23,420 | |
| R-squared | 0.010 | 0.022 | 0.024 | 0.024 | 0.036 | |
Now, is the following interpretation correct?
Compared to matrilineal and bilateral societies, the average child mortality of patrilineal societies reduces by 0.0119 units = 1.2% per cent ? (as my outcome variable is already a rational number in the form of 0.5 or 0.125)
5/10 = 0.5 (if 5 from 10 children died)
1/8 = 0.125 (if 1 from 8 children died)
Thank you for your time and help.
Kind regards, Veronika
Comment