Stata Technical Bulletin
27
. list id activity age birth year period
id |
activity |
age |
birth |
year |
period | |
1. |
1 |
nonactiv |
22 |
54 |
76 |
-90 |
2. |
1 |
selfemp |
27 |
54 |
81 |
-90 |
3. |
1 |
selfemp |
39 |
54 |
90 |
-90 |
4. |
1 |
selfemp |
39 |
54 |
91 |
[90-91 |
ε. |
1 |
selfemp |
39 |
54 |
93 |
[91- |
6. |
2 |
nonactiv |
19 |
74 |
90 |
-90 |
7. |
2 |
nonactiv |
19 |
74 |
91 |
[90-91 |
S. |
2 |
nonactiv |
19 |
74 |
93 |
[91- |
9. |
3 |
nonactiv |
19 |
65 |
84 |
-90 |
10. |
3 |
selfemp |
22 |
65 |
87 |
-90 |
11. |
3 |
waged |
28 |
65 |
90 |
-90 |
12. |
3 |
waged |
28 |
65 |
91 |
[90-91 |
13. |
3 |
waged |
28 |
65 |
93 |
[91- |
14. |
4 |
nonactiv |
7 |
70 |
77 |
-90 |
ιε. |
4 |
nonactiv |
20 |
70 |
90 |
-90 |
16. |
4 |
selfemp |
23 |
70 |
91 |
[90-91 |
17. |
4 |
selfemp |
23 |
70 |
93 |
[91- |
18. |
ε |
nonactiv |
4 |
68 |
72 |
-90 |
19. |
ε |
waged |
11 |
68 |
79 |
-90 |
20. |
ε |
nonactiv |
22 |
68 |
90 |
-90 |
21. |
ε |
waged |
26 |
68 |
91 |
[90-91 |
22. |
ε |
waged |
26 |
68 |
93 |
[91- |
23. |
6 |
nonactiv |
12 |
66 |
78 |
-90 |
24. |
6 |
nonactiv |
24 |
66 |
90 |
-90 |
28. |
6 |
waged |
27 |
66 |
91 |
[90-91 |
26. |
6 |
waged |
27 |
66 |
93 |
[91- |
27. |
7 |
nonactiv |
24 |
69 |
90 |
-90 |
28. |
7 |
nonactiv |
24 |
69 |
91 |
[90-91 |
29. |
7 |
nonactiv |
24 |
69 |
93 |
[91- |
30. |
8 |
nonactiv |
16 |
54 |
70 |
-90 |
31. |
8 |
nonactiv |
39 |
54 |
90 |
-90 |
32. |
8 |
nonactiv |
39 |
54 |
91 |
[90-91 |
33. |
8 |
nonactiv |
39 |
54 |
93 |
[91- |
Since every individual is observed as late as 1993 and none is observed in both 1990 and 1991, slicing the data on the
1990-91 interval affects the observations for each individual. For instance, slice adds two observations for individual ‘1’, one
for 1990, and one for 1991. The values of all the variables except year are copied down from the observation where year==93,
the first observation following the 1990-91 interval. As a consequence, the age is incorrect in the added observations. This
problem is easily corrected.
. replace age = year - birth
(13 real changes made)
Now we can examine activity in 1991 (period==l):
. tabulate activity sex if period==l
I Sex
Activityl |
Male |
Female I |
Total |
— | |||
nonactiV I |
0 |
3 I |
3 |
waged I |
1 |
2 I |
3 |
selfemp I |
2 |
0 I |
2 |
— | |||
Total I |
3 |
5 I |
8 |
Apart from descriptive analysis, the slice command is most useful in Cox regression, when we use time-varying regressors
to test the effect of passing through different calendar periods or age groups. For example, we can slice the time into three
periods: from the earliest year through 1980, from 1981 through 1990, and from 1991 until the time at survey (1993):
. use example, clear
. slice year, tvid(id) interval(80,90) saving(new) generate(period)
-80 + 8
[80-90 + 5
[90- --------
13 records added to 20
Note that age has to be corrected again, before running the regression:
. replace age = year - birth
(13 real changes made)
More intriguing information
1. The Works of the Right Honourable Edmund Burke2. A Rare Case Of Fallopian Tube Cancer
3. Survey of Literature on Covered and Uncovered Interest Parities
4. A Theoretical Growth Model for Ireland
5. The name is absent
6. The name is absent
7. Industrial districts, innovation and I-district effect: territory or industrial specialization?
8. Epistemology and conceptual resources for the development of learning technologies
9. Reform of the EU Sugar Regime: Impacts on Sugar Production in Ireland
10. Connectionism, Analogicity and Mental Content