Procedure

CONTENTS	ADDRESS OPTIMIZATION
FAQ Statement History Glossary Platform Promotion Initiatives LTE/Articles LP Links Liberty links	You may devise a better procedure for the tools available to you, but it should be no less sophisticated than the following one used by San Diego Libertarian Party Chair, Edward Teyssier to maximize the return on investment. Plan Envelope addressing and Follow Up I. Working with ACCESS: Insert the Registrary of Voters (ROV) data into MS ACCESS, Filemaker Pro or other data base management system. One file will be the registrants, the other file will be the precinct/district mapping information. Using the registrant file in ACCESS, remove all the non-LP registrants using the query function (find "Libertarians", find omitted, delete all). Now we're dealing with just 11,000 names instead of 1. 3 million. Using the precinct/district file in ACCESS, link the partisan districts (AD, SD, CD) to the names in the reduced registrant file. Save this as an ACCESS file, "LPMASTER". From the LPMASTER file in ACCESS create (Save As) an MS EXCEL file. The same name can be used, but this file will have an MS EXCEL extension, ie: .xls instead of .axs. II. WORKING WITH EXCEL FILES: a. Add "score" data to registrant name files. i. Membership Add a column to be used for indicating SDLP membership status, this is column "SDLP". Convert the SDLP membership data into a separate EXCEL file. Open the SDLP membership data in another EXCEL window and sort by street name, street number, last name. Now do a side by side comparison, and when a registrant is also a member, put a "1" in the membership column in LPMASTER. ii. Stuff code. The file now has partisan district information for each registrant. Add a column to be used for stuffing envelopes. This is the "stuff" code. It is created from the district and membership columns, and will look like the following example: "A76-S38-C52-L1" In the foregoing example, the registrant is in the 76th Assembly District, the 38th State Senate District, and the 52nd Congressional District, and that they are a SDLP party member (ie:L=1). iii. How many per household? Add a column to be used for indicating how many LP registrants live at the same address, this is column, "MANY". Sort by street name, street number, last name. If more than one LP living at that same address, put a number in the "MANY" column representing the number living at that address. Sort on address rather than last names as some people are living together either unmarried or married, but have kept their last names. The file now also has whether each registrant is a party member as well as how many additional LP registrants live at their address. iv. Score based on voting patterns. Create a new column called, "votescore" to be used to add a weight based upon the registrants voting pattern. Voters that have voted in all six of the last major elections are more likely to sign our petitions. Those registrants will get the highest scores. Even if they're not "perfect" voters, we should consider sending them petitions with a greater preference towards those most likely to respond. For example, registrants who have voted in the last two elections are more likely to still be alive and living at the last address they voted at than are people who haven't voted in the last two elections. Also, as we learned from last time, younger voters who have just recently registered Libertarian and will be voting for the first time, are more likely to take the time to return signed petitions. v. Score based on Telephone number. Create another new column, "TELEPHONE", and determine (sort) whether a telephone number exists for this registrant. If so, put a "1" in this column (fill down). Otherwise, a zero. Or use a calculation [if (column T = "", "1", "0")] in the TELEPHONE column fields that sets itself to "1" if a value exists in the telephone number field, else a "0". The data base from Joe Dehn has a "confidence factor" of 1 through 5 for the telephone number, with 1 being the best. I suggest we use that factor to modify the wieght applied to the telephone adder as follows: IF (Tele_Conf>0) then Tele_Adder=(0. 8-(Tele_Conf/10)) Else Tele_Adder=0 If the factor is zero, then there isn't a number and the adder is zero. If there is any factor then subtract 0. 1 for every point the factor is greater than zero. vi. Score based on how many per household. No reason to send multiple envelopes to same address! When more than one LP registrant lives at the same address, create a new 'combined' record for the combined names. For example, Mary Smith and John Smith should be combined into "Mary & John Smith". Who ever has the highest score should be named first and combined record will have that highest score. Keep the 'combined' record into CLPMASTER. The records that were used to create the combined record are now redundant and shall be deleted or hidden so that CLPMASTER just has the essential records. (Thought: Determine some way to turn-on/turn-off, ie: "hide" the redundant data so that we don't have delete and then un-delete that data later.) c. Weighting formula: File now has a 'score' for each record (registrant) indicating our best estimate of how likely sending an envelope to that address will get us a return. Score based on voting history: Voted in 6th to last election: add 0. 1 points Voted in 5th to last election: add 0. 3 points Voted in 4th to last election: add 0. 5 points Voted in 3rd to last election: add 0. 9 points Voted in 2nd to last election: add 1. 5 points Voted in last election: add 2. 5 points ---------- ie: total possible based on voting history: 5. 8 points adder for LP membership: Add 5. 0 points adder for newly registred in last 6 mos. Add 4. 4 points. adder for tel. no. available: Add 0. 7 points Multiplier based on number of registrants at same address: Multiply subtotal of all the above by following: If 1 person, multiply by: 1. 0 If 2 people, multiply by: 1. 5 If 3 people, multiply by: 2. 0 If 4 people, multiply by: 2. 5 If 5 people, multiply by: 3. 0, etc. , ie: f(x) = (x+1)/2 d. Sorting and selecting registrants: Sort this weighted, scored and combined CLPMASTER first by Assembly Districts (AD), and then score. Plan on giving each AD candidate an equal number of envelopes. Assuming that we'll be mailing 3,500 envelopes and that there are seven (7) AD candidates, then we'll needed to select the top 500 scoring registrants per AD, ie: 3,500/7 = 500. (Obviously, if we only have 6 AD candidates then we'll use 3,500/6=583 per AD candidate, and if we only have five, 3,500/5=700, etc. ) Save the selected records into a new file called "Mailing". Also, be sure to check that all the candidate's have at least the same, minimum number (whether it's 500 or 583 or 700 or whatever) as the AD candidates. The AD's are the smallest district, so giving 500 to each of them SHOULD give more than that to the SD and CD candidates. But check just to be sure. If not, go back to CLPMASTER and re-sort using a lower cut-off score for some AD's and a higher one for other AD's to make sure everyone gets a fair number of envelopes. Also, make sure that if any one has an over abundance of envelopes, like more than 1200, that those envelopes couldn't be re-distributed more evenly. Save this file as "MAILING". e. When done, create a final "FMAILING" file. This is the file that will go to the mailing house, with instructions on how to print the envelopes and to include both the SCORE and STUFF codes. The extra information that is in the "MAILING" file, such as date of birth, state of birth, etc. is extraneous and might confuse or distract the mailing house personnel. The "FMAILING" file should only have the information the mailing house needs to address and encode the envelopes, ie: name, address, STUFF and SCORE information. f. Create a separate district files for each candidate. Using the CLPMASTER file, give each candidate a floppy disk file that shows all the registered LP voters in his district for them to follow up. For each record, show whether they were sent an envelope, what their score was, and their telephone number, if available, etc. g. Create a Return Mailing, "RMAILING" file. This is a 'check-off' file. This is the file to be used to create both the check-off sheets and check-off file so we can keep a record of everyone who has returned envelopes with the signed petitions inside. Use the print-out to help keep track by hand. Then update the file on the computer file for dissemination to the candidates for follow-up. i. Create four new columns to add a check off square to "check" if they returned each of the three petitions they were given. Instead of a "check" mark, the actual district number should be entered. This is to be sure the person signed the right petition. So it acts as a check on how accurately we stuffed the envelopes. Also, we need to record the number of signtatures they got on each petition. ii. Add a new "DATE REC'D" column to use to keep track of the date the envelope was returned. To be used by person opening envelopes. iii. Add a new "DATE ENTERED" column to indicate whether or not the information collected by checking off by hand has been entered into the computer. To be used by person entering the data into the computer. iv. Create a new column for comments like, "included membership dues", or "included donation", etc. or for some other response to our mailing, other than just signing and returning petitions. As the signed petitions are returned this file will be critical in helping candidates follow-up on their signature collection efforts. Also, this file will be kept and analyzed to study how well our "scoring" system worked. We'll do a statistical analysis on each weighting factor so we know better next time how to weigh each factor in the score. DONE!

CONTENTS

ADDRESS OPTIMIZATION

You may devise a better procedure for the tools available to you, but it should be no less sophisticated than the following one used by San Diego Libertarian Party Chair, Edward Teyssier to maximize the return on investment.

Plan Envelope addressing and Follow Up

I. Working with ACCESS: Insert the Registrary of Voters (ROV) data into MS ACCESS, Filemaker Pro or other data base management system. One file will be the registrants, the other file will be the precinct/district mapping information. Using the registrant file in ACCESS, remove all the non-LP registrants using the query function (find "Libertarians", find omitted, delete all). Now we're dealing with just 11,000 names instead of 1. 3 million. Using the precinct/district file in ACCESS, link the partisan districts (AD, SD, CD) to the names in the reduced registrant file. Save this as an ACCESS file, "LPMASTER".

From the LPMASTER file in ACCESS create (Save As) an MS EXCEL file. The same name can be used, but this file will have an MS EXCEL extension, ie: .xls instead of .axs.

II. WORKING WITH EXCEL FILES:

a. Add "score" data to registrant name files.

i. Membership

Add a column to be used for indicating SDLP membership status, this is column "SDLP".

Convert the SDLP membership data into a separate EXCEL file.

Open the SDLP membership data in another EXCEL window and sort by street name, street number, last name.

Now do a side by side comparison, and when a registrant is also a member, put a "1" in the membership column in LPMASTER.

ii. Stuff code.

The file now has partisan district information for each registrant.

Add a column to be used for stuffing envelopes. This is the "stuff" code. It is created from the district and membership columns, and will look like the following example:

"A76-S38-C52-L1"

In the foregoing example, the registrant is in the 76th Assembly District, the 38th State Senate District, and the 52nd Congressional District, and that they are a SDLP party member (ie:L=1).

iii. How many per household?

Add a column to be used for indicating how many LP registrants live at the same address, this is column, "MANY". Sort by street name, street number, last name.

If more than one LP living at that same address, put a number in the "MANY" column representing the number living at that address.

Sort on address rather than last names as some people are living together either unmarried or married, but have kept their last names.

The file now also has whether each registrant is a party member as well as how many additional LP registrants live at their address.

iv. Score based on voting patterns.

Create a new column called, "votescore" to be used to add a weight based upon the registrants voting pattern. Voters that have voted in all six of the last major elections are more likely to sign our petitions. Those registrants will get the highest scores.

Even if they're not "perfect" voters, we should consider sending them petitions with a greater preference towards those most likely to respond. For example, registrants who have voted in the last two elections are more likely to still be alive and living at the last address they voted at than are people who haven't voted in the last two elections. Also, as we learned from last time, younger voters who have just recently registered Libertarian and will be voting for the first time, are more likely to take the time to return signed petitions.

v. Score based on Telephone number.

Create another new column, "TELEPHONE", and determine (sort) whether a telephone number exists for this registrant. If so, put a "1" in this column (fill down). Otherwise, a zero. Or use a calculation [if (column T = "", "1", "0")] in the TELEPHONE column fields that sets itself to "1" if a value exists in the telephone number field, else a "0".

The data base from Joe Dehn has a "confidence factor" of 1 through 5 for the telephone number, with 1 being the best. I suggest we use that factor to modify the wieght applied to the telephone adder as follows:

IF (Tele_Conf>0) then Tele_Adder=(0. 8-(Tele_Conf/10)) Else Tele_Adder=0

If the factor is zero, then there isn't a number and the adder is zero. If there is any factor then subtract 0. 1 for every point the factor is greater than zero.

vi. Score based on how many per household.

No reason to send multiple envelopes to same address!

When more than one LP registrant lives at the same address, create a new 'combined' record for the combined names. For example, Mary Smith and John Smith should be combined into "Mary & John Smith". Who ever has the highest score should be named first and combined record will have that highest score. Keep the 'combined' record into CLPMASTER. The records that were used to create the combined record are now redundant and shall be deleted or hidden so that CLPMASTER just has the essential records.

(Thought: Determine some way to turn-on/turn-off, ie: "hide" the redundant data so that we don't have delete and then un-delete that data later.)

c. Weighting formula:

File now has a 'score' for each record (registrant) indicating our best estimate of how likely sending an envelope to that address will get us a return. Score based on voting history:

Voted in 6th to last election: add 0. 1 points

Voted in 5th to last election: add 0. 3 points

Voted in 4th to last election: add 0. 5 points

Voted in 3rd to last election: add 0. 9 points

Voted in 2nd to last election: add 1. 5 points

Voted in last election: add 2. 5 points

----------

ie: total possible based on voting history: 5. 8 points

adder for LP membership: Add 5. 0 points

adder for newly registred in last 6 mos. Add 4. 4 points.

adder for tel. no. available: Add 0. 7 points

Multiplier based on number of registrants at same address:

Multiply subtotal of all the above by following:

If 1 person, multiply by: 1. 0

If 2 people, multiply by: 1. 5

If 3 people, multiply by: 2. 0

If 4 people, multiply by: 2. 5

If 5 people, multiply by: 3. 0, etc. , ie: f(x) = (x+1)/2

d. Sorting and selecting registrants:

Sort this weighted, scored and combined CLPMASTER first by Assembly Districts (AD), and then score. Plan on giving each AD candidate an equal number of envelopes. Assuming that we'll be mailing 3,500 envelopes and that there are seven (7) AD candidates, then we'll needed to select the top 500 scoring registrants per AD, ie: 3,500/7 = 500.

(Obviously, if we only have 6 AD candidates then we'll use 3,500/6=583 per AD candidate, and if we only have five, 3,500/5=700, etc. )

Save the selected records into a new file called "Mailing".

Also, be sure to check that all the candidate's have at least the same, minimum number (whether it's 500 or 583 or 700 or whatever) as the AD candidates. The AD's are the smallest district, so giving 500 to each of them SHOULD give more than that to the SD and CD candidates. But check just to be sure. If not, go back to CLPMASTER and re-sort using a lower cut-off score for some AD's and a higher one for other AD's to make sure everyone gets a fair number of envelopes. Also, make sure that if any one has an over abundance of envelopes, like more than 1200, that those envelopes couldn't be re-distributed more evenly.

Save this file as "MAILING".

e. When done, create a final "FMAILING" file.

This is the file that will go to the mailing house, with instructions on how to print the envelopes and to include both the SCORE and STUFF codes. The extra information that is in the "MAILING" file, such as date of birth, state of birth, etc. is extraneous and might confuse or distract the mailing house personnel.

The "FMAILING" file should only have the information the mailing house needs to address and encode the envelopes, ie: name, address, STUFF and SCORE information.

f. Create a separate district files for each candidate.

Using the CLPMASTER file, give each candidate a floppy disk file that shows all the registered LP voters in his district for them to follow up. For each record, show whether they were sent an envelope, what their score was, and their telephone number, if available, etc.

g. Create a Return Mailing, "RMAILING" file.

This is a 'check-off' file. This is the file to be used to create both the check-off sheets and check-off file so we can keep a record of everyone who has returned envelopes with the signed petitions inside. Use the print-out to help keep track by hand. Then update the file on the computer file for dissemination to the candidates for follow-up. i. Create four new columns to add a check off square to "check" if they returned each of the three petitions they were given. Instead of a "check" mark, the actual district number should be entered. This is to be sure the person signed the right petition. So it acts as a check on how accurately we stuffed the envelopes.

Also, we need to record the number of signtatures they got on each petition.

ii. Add a new "DATE REC'D" column to use to keep track of the date the envelope was returned. To be used by person opening envelopes.

iii. Add a new "DATE ENTERED" column to indicate whether or not the information collected by checking off by hand has been entered into the computer. To be used by person entering the data into the computer.

iv. Create a new column for comments like, "included membership dues", or "included donation", etc. or for some other response to our mailing, other than just signing and returning petitions.

As the signed petitions are returned this file will be critical in helping candidates follow-up on their signature collection efforts. Also, this file will be kept and analyzed to study how well our "scoring" system worked. We'll do a statistical analysis on each weighting factor so we know better next time how to weigh each factor in the score.

DONE!