270soft Forum

# Preference Weights

## Recommended Posts

Could someone please tell me what preference weights from the preference_weights.p4e file mean, how they work, and how to make best use of them? Thanks!

• 1
##### Share on other sites

If preferences are on in the scenario.p4e file, like in the Australian scenarios, each riding without a 50% winner redistributes the votes of the 3rd to 16th placed candidates to the 1st and 2nd placed parties. The preference_weights.p4e file is used to determine what proportion of each minor candidate's vote is distributed to each of the major parties. The comments in the file actually describe it pretty well. The preference weights become the w1 to w6 variables in the formula that calculates the proportions.

An example:

LABOR, 40%, 2 stars, Rudd has 18 total attribute points, 5 established rating, 25 distance between platforms

LIBERAL, 40%, 2 stars, Howard has 17 total attribute points, 5 established rating, 75 distance between platforms

GREEN, 20%

Green has 20 relationship with Liberal and 80 relationship plus coalition with Labor.

Preference weights: 300 coalition, 50 party relation, 5 leader attribute, 5 establishment, 25 platform similarity, 5 candidate strength.

Labor's value is: 300*1 + 50*80 + 5*18 + 5*5 + 25*(288-25) + 5*2 = 11000

Liberal's value is: 300*0 + 50*20 + 5*17 + 5*5 + 25*(288-75) + 5*2 = 6445

Squared values: 1.21*10^8, 4.1538025*10^7

Labor's percent gained is: 1.21*10^8/(1.21*10^8+4.1538025*10^7) = 74% of Green's vote

So Liberal gains 26% of Green's vote.

LABOR 40 + 20*0.74 = 54.8

LIBERAL 40 + 20*0.26 = 45.2

How to use them: Reduce a value to zero or low values if you don't want it to affect how voters transfer votes. Increase it to large numbers if you want it to be the overwhelming factor.

As a guideline, in the above example, here's how the different factors contributed to the Labor raw vote transfer proportions:

Coalition 3%

Party relations 36%

Established ~0%

Platform distance 60%

Candidate strength ~0%

What's more, the squaring means that large values take an even larger role in the final total.

So if you want to make one of those factors more important, increase its value.

##### Share on other sites

• 1 year later...

Labor's value is: 300*1 + 50*80 + 5*18 + 5*5 + 25*(288-25) + 5*2 = 11000

Liberal's value is: 300*0 + 50*20 + 5*17 + 5*5 + 25*(288-75) + 5*2 = 6445

Squared values: 1.21*10^8, 4.1538025*10^7

Labor's percent gained is: 1.21*10^8/(1.21*10^8+4.1538025*10^7) = 74% of Green's vote

So Liberal gains 26% of Green's vote.

LABOR 40 + 20*0.74 = 54.8

LIBERAL 40 + 20*0.26 = 45.2

74% seems quite low.

And Labor aren't in coalition with the Greens. This would result in an even lower percentage of Greens votes flowing to Labor. This suggests that the values in the formula need to be tweaked. Particularly since 23% of Nationals votes went to Labor in preferences.

• 1
##### Share on other sites

• 1 year later...

##### Share on other sites

Thanks for spurring this thread again.

I'm disappointed to see the loss of the preference weights from the newest PM4E engine. Instead the has to manually input preference distributions to make things work, which is a tedius problem if you want to have more than a small handful of parties.

• 1

## Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account. ×   Pasted as rich text.   Paste as plain text instead

Only 75 emoji are allowed.