Coefficients signs

[Olivia Emma]

Hi all


Could someone kindly assist me in determining which of these results align with the given theory? I am seeking guidance on how to identify results that are consistent with the theory in question. To me, the coefficient haz06 is not in accordance with the theory. How can I justify that? Please help

Ed_years the mother education in years
haz06 Length/height-for-age Z-score
waz06 Weight-for-age Z-score
whz06 Weight-for-length/height Z-score
bmiz06 BMI-for-age Z-score

VARIABLES	haz06	whz06	bmiz06	birthweight
ed_years	-0.095*	0.052**	0.426***	13.518***
	(0.049)	(0.031)	(0.075)	(40.163)

Thanks

[Olivia Emma]

Clyde Schechter Please

[Clyde Schechter]

So, I'm confused. Questions:
1. What specifically does your theory predict that you think these results are inconsistent with?
2. What is the outcome variable here? Is it ed_years? What kind of regression was done? Any other variables in it that you didn't mention?
3. You mention a variable waz06, but you do not show any result for it. Was it included in the regression? If so, what was the result for that?
4. What are the units of measurement for birthweight?
5. At what age were the various growth indices measured?
6. What were the inclusion criteria for the study?

Also, please run

corr haz06 waz06 whz06 bmiz06 birthweight

and show the complete output.

[Olivia Emma]

haz06 waz06 whz06 bmiz06 birthweight are the outcome variables and ed_years is the main independent variable and other various Control and Fixed effects are included in the RDD regression.

VARIABLES	haz06	whz06	bmiz06	birthweight
ed_years	-0.095*	0.052**	0.426***	13.518***
	(0.049)	(0.031)	(0.075)	(40.163)

birthweight is grams please measured for childern of age less than 5 years why the sign might of haz06 might be negative please

[Clyde Schechter]

I asked you in #2 for additional information. Thank you for the part that you did supply. But I still need to see the results for variable waz06 (or say that it wasn't included in the regression), and the results of -corr haz06 waz06 whz06 bmiz06 birthweight-

And you still haven't said what the theoretical prediction is that you think the results conflict with.

[Olivia Emma]

. corr ed_years haz06 waz06 whz06 bmiz06 birthweight
(obs=6,500)

| ed_years haz06 waz06 whz06 bmiz06 birthw~t
-------------+------------------------------------------------------
ed_years | 1.0000
haz06 | 0.1544 1.0000
waz06 | 0.1682 0.6683 1.0000
whz06 | 0.0706 -0.0872 0.6680 1.0000
bmiz06 | 0.0461 -0.1903 0.5981 0.9720 1.0000
birthweight | -0.0356 0.1069 0.1848 0.1369 0.1336 1.0000

[Olivia Emma]

results with waz06

	(1)	(2)	(3)	(4)	(5)
VARIABLES	haz06- IV	waz06- IV	whz06- IV	bmiz06- IV	birthweight- IV
ed_years	-0.084*	-0.010	0.073**	0.236***	103.518***
	(0.049)	(0.043)	(0.031)	(0.075)	(40.163)
Observations	8,685	8,812	8,663	8,757	13,676
Mean of Dependent Variable	-1.456	-0.812	0.0404	0.251	3371

I expect the coefficients to me positive as increased maternal education might have a positive impact on the child outcomes ( haz , waz whz )

[Clyde Schechter]

I still don't understand what you are showing me.

Is this one regression, like

regression_command ed_years haz_06 waz06 whz06 bmiz06 birthweight

Or is this five separate regressions:

regression_command ed_years haz06
regression_command ed_years waz06
...
regression_command ed_years birthweight

Or is this five separate regressions:

regression_command haz06 ed_years
regression_command waz06 ed_years
...
regression_command birthweight ed_years

Also, what are the various control and fixed-effects variables you used. (Not just their names: what do they mean?)

[Olivia Emma]

regression_command haz06 ed_years
regression_command waz06 ed_years ...
regression_command birthweight ed_years


control variables
child_birthyear child_gener survey_year region urban_dummy

[Olivia Emma]

and I expect the positive relationship between maternal education and child outcomes (HAZ , WAZ )

[Clyde Schechter]

Well, I think you actually have an interesting, and nuanced set of results. You have positive coefficients for the whz06 and bmi variables, both of which reflect the relationship between body mass and height reflecting nutritional status, but negative coefficients for the haz and waz, which are basically just measures of overall stature.

Since maternal education provides advantages in nutrition and lower risk of infectious disease to the child, it is plausible that these will be reflected in the child's habitus: they will be well nourished, be at lower risk for developmental problems and tend to have higher weight for height. By contrast, I see no obvious reason why maternal education would be positively related to just being taller and heavier--these are much more determined by genetics.

So I don't see any reason to suspect that the modeling or the data are substantially wrong. I think you have probably done things correctly and reached appropriate conclusions. In this case, I think your theoretical expectations were wrong.