Thread: Help please: identifying data type

So I've got this project I'm doing in statistics dealing with election results.

Basically I must determine whether winning with over 55% of the votes in presidential elections should be considered a "landslide".

I've got the percentage of votes the winners attained in the past 13 presidential elections.

I see two ways of analyzing this:

Consider the data binomial, where p=1 if a presidential candidate won with 55% of the votes or more; thus k would represent the number of candidates that won with more than 55% of the votes.

Or I consider the data normal, taking a sample mean of the percentage of votes the winner has.

I'm not very sharp when it comes to stastistics, so I'm a bit confused on which route to take.

Any help is much appreciated!

Originally Posted by darkscaryforest

So I've got this project I'm doing in statistics dealing with election results.

Basically I must determine whether winning with over 55% of the votes in presidential elections should be considered a "landslide".

I've got the percentage of votes the winners attained in the past 13 presidential elections.

I see two ways of analyzing this:

Consider the data binomial, where p=1 if a presidential candidate won with 55% of the votes or more; thus k would represent the number of candidates that won with more than 55% of the votes.

Or I consider the data normal, taking a sample mean of the percentage of votes the winner has.

I'm not very sharp when it comes to stastistics, so I'm a bit confused on which route to take.

Any help is much appreciated!

Both will give tests which may indicated how unusual a win with more than 55% of the popular vote is. One will be more powerful than the other (almost certainly the second). But which is preferable depends on what you want to do with the result.

One of the secrets of real statistics is that the test you apply depends on the question you want answered. Frame the question clearly and then we can judge the relative merits of the tests.

Basically I am asking what do you mean by a "landslide"?

RonL

Thanks for your help.

Like you said, I'm just seeking to determine how unusual a win with over 55% of the votes would be.

If such a win of that magnitude seems to be very uncommon, I would consider 55% or more of the votes to be a true "landslide"...since such a powerful win is so uncommon.

But, if the data indicates that elections are won with more than 55% of the votes all the time, I wouldn't consider such a landslide..since it's so common and people are used to such numbers. Perhaps instead, 75% should be considered a landslide?

I suppose the word "landslide" isn't a very appropriate word to use however.

But anyways, thanks again!

Originally Posted by darkscaryforest

Thanks for your help.

Like you said, I'm just seeking to determine how unusual a win with over 55% of the votes would be.

If such a win of that magnitude seems to be very uncommon, I would consider 55% or more of the votes to be a true "landslide"...since such a powerful win is so uncommon.

But, if the data indicates that elections are won with more than 55% of the votes all the time, I wouldn't consider such a landslide..since it's so common and people are used to such numbers. Perhaps instead, 75% should be considered a landslide?

I suppose the word "landslide" isn't a very appropriate word to use however.

But anyways, thanks again!

Can you post the numbers that you do have?

RonL

The percentage of popular vote the winners had in the past 13 elections:

50.73%
47.87%
49.23%
43.01%
53.37%
58.77%
50.75%
50.08%
60.67%
43.42%
61.05%
49.72%
57.37%

Originally Posted by darkscaryforest

The percentage of popular vote the winners had in the past 13 elections:

50.73%
47.87%
49.23%
43.01%
53.37%
58.77%
50.75%
50.08%
60.67%
43.42%
61.05%
49.72%
57.37%

The first observation I would make from this data is that there are four
elections won with more than 55% of the vote, which in 13 elections
on the basis of frequency does not seem so exceptional. However all
four are in the botton 2/3 of the table, which may be significant.

RonL