Stata Technical Bulletin
11
. tab married, summ(mhiθ) | |||
I |
Summary of Baseline Mental Health | ||
Mean |
Std. Dev. |
Freq. | |
0 I |
65.238349 |
23.932533 |
— 751 |
i I |
71.337259 |
21.6785 |
1104 |
____________ ^-_ Total I |
68.868105 |
22.809219 |
— 1855 |
. tab married, summ(mhi3mo)
I Summary of Mental Health at 3 months
married∣ |
Mean |
Std. Dev. |
Freq. | |
------------+- |
— | |||
0 I |
68.699844 |
22.002264 |
1376 | |
1 I |
73.239149 |
20.31232 |
1968 | |
------------+- |
— | |||
Total I |
71.371301 |
21.139313 |
3344 | |
. tab married, |
. summ(mhilyr) | |||
I |
Summary of Mental Health at |
1 year | ||
— | ||||
0 I |
69.756461 |
21.821531 |
761 | |
i I |
74.669736 |
18.927619 |
1086 | |
———————————-⅛- |
— | |||
Total I |
72.645371 |
20.309153 |
1847 | |
. tab married, |
, summ(mhi2yr) | |||
I |
Summary of Mental Health |
at 2 |
years | |
— | ||||
0 I |
70.728295 |
21.383028 |
741 | |
i I |
74.994872 |
19.236051 |
1040 | |
———————————-⅛- |
— | |||
Total I |
73.219727 |
20.26075 |
1781 | |
. tab married, |
, summ(mhi4yr) | |||
I |
Summary of Mental Health |
at 4 |
years | |
— | ||||
0 I |
72.62061 |
21.52095 |
579 | |
i I |
76.28414 |
18.028531 |
847 | |
———————————-⅛- |
— | |||
Total I |
74.796634 |
19.597544 |
1426 |
The first measurement (mhiθ) is available on a random half of the sample only, but mhi3mo is available on just about everybody.
Between 3 months and one year, approximately half of the sample was discarded by the original investigators. There is additional
attrition during the follow-up period.
It appears that married persons have “better” mental health (higher values), but that the gap may be narrowing over
time. One purpose of longitudinal studies is to distinguish cohort effects—relationships that appear to be present but are really
manifestations of group associations—from causal effects or trends. For example, it could be that marriage produces good mental
health—a real effect—or it could be that people with good mental health tend to get married—a manifestation. If the latter is
the case, scores would regress toward the mean over time.
A multivariate examination is now warranted. I will begin with a multivariate examination of the cross-sectional effects:
. corr married imale Iagecont Inonwht
(obs=3691)
I married imale iagecont inonwht
--------+------------------------------------
married I 1.0000
imaleI 0.2456 1.0000
iagecontI 0.0663 0.0872 1.0000
inonwht∣ -0.0749 -0.0814 -0.0904 1.0000
Fortunately, there does not appear to be much intercorrelation between the demographic variables so we do not have to be overly
concerned with which variables we include in the model. Even if we omitted a relevant variable, it would have little effect on
the estimated coefficients of the others.
More intriguing information
1. EXECUTIVE SUMMARIES2. Survey of Literature on Covered and Uncovered Interest Parities
3. 09-01 "Resources, Rules and International Political Economy: The Politics of Development in the WTO"
4. Program Semantics and Classical Logic
5. The Challenge of Urban Regeneration in Deprived European Neighbourhoods - a Partnership Approach
6. Sex differences in the structure and stability of children’s playground social networks and their overlap with friendship relations
7. The name is absent
8. Dementia Care Mapping and Patient-Centred Care in Australian residential homes: An economic evaluation of the CARE Study, CHERE Working Paper 2008/4
9. A Hybrid Neural Network and Virtual Reality System for Spatial Language Processing
10. A Rare Presentation of Crohn's Disease