Repeated time values - Panel Data

[Purnima Kanther]

Hi,

I have election data for a country. I am trying to set a panel but as each place has multiple candidates, I have repeated time values. I can not set my panel on candidate level as they do not always repeat. I want to conduct my analysis at county level. A sample of structure of the data is placed below:

Year	State	County	Constituency	Candidate	Winner
1990	MP	County 1	PC1	1	1
1990	MP	County 1	PC1	2	0
1990	MP	County 1	PC1	3	0
1990	MP	County 1	PC2	4	1
1990	MP	County 1	PC2	5	0
1990	MP	County 2	PC3	6	1
1990	MP	County 2	PC3	7	0
1990	MP	County 2	PC3	8	0
1990	MP	County 2	PC4	9	1
1990	MP	County 2	PC4	10	0
1990	MP	County 2	PC4	11	0
1990	UP	County 1	PC5	12	1
1990	UP	County 1	PC5	13	0
1990	UP	County 1	PC5	14	0
1990	UP	County 1	PC6	15	1
1990	UP	County 1	PC6	16	0
1990	UP	County 1	PC6	16	0
1990	UP	County 2	PC7	17	1
1990	UP	County 2	PC7	18	0
1990	UP	County 2	PC7	19	0
1990	UP	County 2	PC8	20	1
1990	UP	County 2	PC8	21	0
1995	MP	County 1	PC1	1	1
1995	MP	County 1	PC1	2	0
1995	MP	County 1	PC1	22	0
1995	MP	County 1	PC2	4	1
1995	MP	County 1	PC2	23	0
1995	MP	County 2	PC3	24	1
1995	MP	County 2	PC3	25	0
1995	MP	County 2	PC3	26	0
1995	MP	County 2	PC4	9	1
1995	MP	County 2	PC4	10	0
1995	MP	County 2	PC4	11	0
1995	UP	County 1	PC5	30	1
1995	UP	County 1	PC5	31	0
1995	UP	County 1	PC5	14	0
1995	UP	County 1	PC6	15	1
1995	UP	County 1	PC6	32	0
1995	UP	County 1	PC6	32	0
1995	UP	County 2	PC7	17	1
1995	UP	County 2	PC7	33	0
1995	UP	County 2	PC7	19	0
1995	UP	County 2	PC8	34	1
1995	UP	County 2	PC8	21	0

I have tried
xtset year state-county where state-county is a group I created. I do not think year-state-county would be correct either. Can someone suggest a way out to conduct this analysis?

I also need to know if the winner in 1995 was the same as the winner in 1990. Unless I set my panel, I do not think I can find that.

Any help/guidance would be appreciated.

Thank you

[Carlo Lazzaro]

Purnima:
welcome to this forum.
If you do not plan to use teime-series commands such as lags and leads, you can simply:

xtset County

[Purnima Kanther]

Hello,

Thank you for your reply.

I was hoping to use the lags leads to find if the candidate who won the election in 1995 was the same candidate who won the election in 1990. Basically if a new candidate was elected or the incumbent got reelected.

Is there any other way of finding this while using xtset county?

Regards,
Purnima

[Carlo Lazzaro]

Purnima:
none that I'm aware of.

[Purnima Kanther]

Dear Carlo,
Thank you for your reply. I will keep trying to find some other way.

In the meanwhile, if anyone else on this forum has a solution, I am trying to do two things:

(1) Generate a variable reelect to find if the candidate who won the election in 1995 was the same candidate who won the election in 1990 or a new candidate.

(2) Trying to analyse the data and run regressions at County level. However, as I have multiple candidates for each State-County, it gives a repeated time values error. I can set it only at county but if I do that, I do not know how to generate variables like reelect.

Also just for further knowledge, are there any consequences of setting panel only at county without the time variable?

Thank you

[Carlo Lazzaro]

Purnima:
as far as questions (1) and (2) are concerned, you should probably skim through the literature of your research field and see what others did in the past when presented with the same research topic.
As far as your last question is concerned, -xtset panelid- will not produce any consequence, exception made for the unfeasibility of using time-series commands such as lags and leads.