mail quantities determination
Locked
 
|
I have a question regarding determining the mail quantities for a direct
mail. This is a non-stat questions, so pls bear. Say for example we are planning to mail 50,000 mail pieces. We would select the records by the logistic regression scores, and maybe go down till the fifth decile to reach the quantity. But we also want to test the model overall in all the deciles (1 - 10 deciles) by mailing some quantities in each of the remaining deciles 6-10. My supervisor wants to determine the optimal quantities to mail in the rest of the deciles so that we can accurately validate the model performance in all the deciles. In the past we have randomly selected, say 2000 mail pieces based on the 50K total. But is there any scientific method to determine the quantities? Can someone share your thoughts pls? Thanks. Renji Abraham ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
|
As always, your response rate will be a key factor of sample size and
how wide your margin of error needs to be. I have found that sample size calculators on the net to be helpful. I have a feeling that for direct mail, 2,000 pieces will not be enough. Go to: http://www.ezsurvey.com/samplesize.html RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of <R. Abraham> Sent: Monday, December 08, 2008 8:55 AM To: [hidden email] Subject: mail quantities determination I have a question regarding determining the mail quantities for a direct mail. This is a non-stat questions, so pls bear. Say for example we are planning to mail 50,000 mail pieces. We would select the records by the logistic regression scores, and maybe go down till the fifth decile to reach the quantity. But we also want to test the model overall in all the deciles (1 - 10 deciles) by mailing some quantities in each of the remaining deciles 6-10. My supervisor wants to determine the optimal quantities to mail in the rest of the deciles so that we can accurately validate the model performance in all the deciles. In the past we have randomly selected, say 2000 mail pieces based on the 50K total. But is there any scientific method to determine the quantities? Can someone share your thoughts pls? Thanks. Renji Abraham ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
|
Rodrigo,
The number I provided in my previous email were not real. The actual numbers for a real mailing per month for a single product is something like this. Total Mail pieces - 1 Mil Mailed per Decile Decile 1 - 200K Decile 2 - 200K Decile 3 - 200K Decile 4 - 200K Decile 5 - 200K ----------------------- Total - 1 Mil ----------------------- Since we also want to find the performance of the bottom deciles, we include a sample of 25K in each of the bottom decile (Decile 6 - Decile 10), based on the scores (quantity is not determined on a scientific basis). I tried the sample size calculator that you had suggested, but the sample size it showed was very small (not what I was thinking of). Or am I missing something? Thanks. Renji "Guerrero, Rodrigo" <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 11:11 AM Please respond to "Guerrero, Rodrigo" <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination As always, your response rate will be a key factor of sample size and how wide your margin of error needs to be. I have found that sample size calculators on the net to be helpful. I have a feeling that for direct mail, 2,000 pieces will not be enough. Go to: http://www.ezsurvey.com/samplesize.html RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of <R. Abraham> Sent: Monday, December 08, 2008 8:55 AM To: [hidden email] Subject: mail quantities determination I have a question regarding determining the mail quantities for a direct mail. This is a non-stat questions, so pls bear. Say for example we are planning to mail 50,000 mail pieces. We would select the records by the logistic regression scores, and maybe go down till the fifth decile to reach the quantity. But we also want to test the model overall in all the deciles (1 - 10 deciles) by mailing some quantities in each of the remaining deciles 6-10. My supervisor wants to determine the optimal quantities to mail in the rest of the deciles so that we can accurately validate the model performance in all the deciles. In the past we have randomly selected, say 2000 mail pieces based on the 50K total. But is there any scientific method to determine the quantities? Can someone share your thoughts pls? Thanks. Renji Abraham ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
The sample size calculator tells you the size of the ACHIEVED sample *if
*you believe that non-response is at random. What kind of return rates have you had in the past? <R. Abraham> wrote: > Rodrigo, > The number I provided in my previous email were not real. The actual > numbers for a real mailing per month for a single product is something > like this. > > Total Mail pieces - 1 Mil > > Mailed per Decile > Decile 1 - 200K > Decile 2 - 200K > Decile 3 - 200K > Decile 4 - 200K > Decile 5 - 200K > ----------------------- > Total - 1 Mil > ----------------------- > > Since we also want to find the performance of the bottom deciles, we > include a sample of 25K in each of the bottom decile (Decile 6 - Decile > 10), based on the scores (quantity is not determined on a scientific > basis). I tried the sample size calculator that you had suggested, but the > sample size it showed was very small (not what I was thinking of). Or am I > missing something? > > Thanks. > Renji > > > > > > > "Guerrero, Rodrigo" <[hidden email]> > Sent by: "SPSSX(r) Discussion" <[hidden email]> > 12/08/2008 11:11 AM > Please respond to > "Guerrero, Rodrigo" <[hidden email]> > > > To > [hidden email] > cc > > Subject > Re: mail quantities determination > > > > > > > As always, your response rate will be a key factor of sample size and > how wide your margin of error needs to be. I have found that sample > size calculators on the net to be helpful. I have a feeling that for > direct mail, 2,000 pieces will not be enough. > > Go to: http://www.ezsurvey.com/samplesize.html > > RG > > Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The > Scooter Store | 830.627.4317 > > -----Original Message----- > From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of > <R. Abraham> > Sent: Monday, December 08, 2008 8:55 AM > To: [hidden email] > Subject: mail quantities determination > > I have a question regarding determining the mail quantities for a direct > mail. This is a non-stat questions, so pls bear. > > Say for example we are planning to mail 50,000 mail pieces. We would > select the records by the logistic regression scores, and maybe go down > till the fifth decile to reach the quantity. But we also want to test > the > model overall in all the deciles (1 - 10 deciles) by mailing some > quantities in each of the remaining deciles 6-10. My supervisor wants to > determine the optimal quantities to mail in the rest of the deciles so > that we can accurately validate the model performance in all the > deciles. > In the past we have randomly selected, say 2000 mail pieces based on the > 50K total. But is there any scientific method to determine the > quantities? > Can someone share your thoughts pls? Thanks. > > Renji Abraham > > ===================== > To manage your subscription to SPSSX-L, send a message to > [hidden email] (not to SPSSX-L), with no body text except the > command. To leave the list, send the command > SIGNOFF SPSSX-L > For a list of commands to manage subscriptions, send the command > INFO REFCARD > > > The information transmitted is intended only for the addressee(s) and may > contain confidential or privileged material, or both. Any review, > receipt, dissemination or other use of this information by non-addressees > is prohibited. If you received this in error or are a non-addressee, > please contact the sender and delete the transmitted information. > > ===================== > To manage your subscription to SPSSX-L, send a message to > [hidden email] (not to SPSSX-L), with no body text except the > command. To leave the list, send the command > SIGNOFF SPSSX-L > For a list of commands to manage subscriptions, send the command > INFO REFCARD > > ===================== > To manage your subscription to SPSSX-L, send a message to > [hidden email] (not to SPSSX-L), with no body text except the > command. To leave the list, send the command > SIGNOFF SPSSX-L > For a list of commands to manage subscriptions, send the command > INFO REFCARD > > > ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD
Art Kendall
Social Research Consultants |
Re: mail quantities determination
|
|
In reply to this post by <R. Abraham>
If you have a large response rate, say 1%+, then you will not need a
large sample to prove there is a difference in the deciles. If your response rate is low, say under .3%, then you will need something in the 200k range. Try using a population size that is the total size of the mail file, not your mailing. I get a mailing sample size of 500k for a .01% error range, 95% confidence, 7 Million population size, and a .15% response rate. RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 From: [hidden email] [mailto:[hidden email]] Sent: Monday, December 08, 2008 10:52 AM To: Guerrero, Rodrigo Cc: [hidden email] Subject: Re: mail quantities determination Rodrigo, The number I provided in my previous email were not real. The actual numbers for a real mailing per month for a single product is something like this. Total Mail pieces - 1 Mil Mailed per Decile Decile 1 - 200K Decile 2 - 200K Decile 3 - 200K Decile 4 - 200K Decile 5 - 200K ----------------------- Total - 1 Mil ----------------------- Since we also want to find the performance of the bottom deciles, we include a sample of 25K in each of the bottom decile (Decile 6 - Decile 10), based on the scores (quantity is not determined on a scientific basis). I tried the sample size calculator that you had suggested, but the sample size it showed was very small (not what I was thinking of). Or am I missing something? Thanks. Renji "Guerrero, Rodrigo" <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 11:11 AM Please respond to "Guerrero, Rodrigo" <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination As always, your response rate will be a key factor of sample size and how wide your margin of error needs to be. I have found that sample size calculators on the net to be helpful. I have a feeling that for direct mail, 2,000 pieces will not be enough. Go to: http://www.ezsurvey.com/samplesize.html RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of <R. Abraham> Sent: Monday, December 08, 2008 8:55 AM To: [hidden email] Subject: mail quantities determination I have a question regarding determining the mail quantities for a direct mail. This is a non-stat questions, so pls bear. Say for example we are planning to mail 50,000 mail pieces. We would select the records by the logistic regression scores, and maybe go down till the fifth decile to reach the quantity. But we also want to test the model overall in all the deciles (1 - 10 deciles) by mailing some quantities in each of the remaining deciles 6-10. My supervisor wants to determine the optimal quantities to mail in the rest of the deciles so that we can accurately validate the model performance in all the deciles. In the past we have randomly selected, say 2000 mail pieces based on the 50K total. But is there any scientific method to determine the quantities? Can someone share your thoughts pls? Thanks. Renji Abraham ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ====================To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
|
Thanks for your replies.
Actually the mail quantities is already determined using past response rates, fatigue factors, etc. But that mail quantity will be met within the first 5 deciles. But what I need is to determine the additional quantities (just a token sample) for the remaining 5 deciles in order to accurately validate the model performance for all deciles. Is there any method to determine the additional quantity for just the remaining deciles in order to compare the response rates for all 10 deciles. I hope I am explaining my problem well. Thanks. Renji "Guerrero, Rodrigo" <[hidden email]> 12/08/2008 12:26 PM To <[hidden email]> cc <[hidden email]> Subject RE: mail quantities determination If you have a large response rate, say 1%+, then you will not need a large sample to prove there is a difference in the deciles. If your response rate is low, say under .3%, then you will need something in the 200k range. Try using a population size that is the total size of the mail file, not your mailing. I get a mailing sample size of 500k for a .01% error range, 95% confidence, 7 Million population size, and a .15% response rate. RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 From: [hidden email] [mailto:[hidden email]] Sent: Monday, December 08, 2008 10:52 AM To: Guerrero, Rodrigo Cc: [hidden email] Subject: Re: mail quantities determination Rodrigo, The number I provided in my previous email were not real. The actual numbers for a real mailing per month for a single product is something like this. Total Mail pieces - 1 Mil Mailed per Decile Decile 1 - 200K Decile 2 - 200K Decile 3 - 200K Decile 4 - 200K Decile 5 - 200K ----------------------- Total - 1 Mil ----------------------- Since we also want to find the performance of the bottom deciles, we include a sample of 25K in each of the bottom decile (Decile 6 - Decile 10), based on the scores (quantity is not determined on a scientific basis). I tried the sample size calculator that you had suggested, but the sample size it showed was very small (not what I was thinking of). Or am I missing something? Thanks. Renji "Guerrero, Rodrigo" <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 11:11 AM Please respond to "Guerrero, Rodrigo" <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination As always, your response rate will be a key factor of sample size and how wide your margin of error needs to be. I have found that sample size calculators on the net to be helpful. I have a feeling that for direct mail, 2,000 pieces will not be enough. Go to: http://www.ezsurvey.com/samplesize.html RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of <R. Abraham> Sent: Monday, December 08, 2008 8:55 AM To: [hidden email] Subject: mail quantities determination I have a question regarding determining the mail quantities for a direct mail. This is a non-stat questions, so pls bear. Say for example we are planning to mail 50,000 mail pieces. We would select the records by the logistic regression scores, and maybe go down till the fifth decile to reach the quantity. But we also want to test the model overall in all the deciles (1 - 10 deciles) by mailing some quantities in each of the remaining deciles 6-10. My supervisor wants to determine the optimal quantities to mail in the rest of the deciles so that we can accurately validate the model performance in all the deciles. In the past we have randomly selected, say 2000 mail pieces based on the 50K total. But is there any scientific method to determine the quantities? Can someone share your thoughts pls? Thanks. Renji Abraham ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
|
Renji,
I've been watching the back-and-forth on your question all day and I am wondering if this isn't just simple problem of determining how many people in each decile are needed to give a specified confidence interval about the expected value for that decile. If so, then I'd think that any basic survey sampling book would have that formula (and it's undoubtedly on the web in many places). Am I completely misunderstanding what you are wanting to do? Gene Maguin ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
|
Gene,
Sorry for the confusion with my question. Let me try to explain it once more. We have a model which is used for scoring and selecting records for mailing campaign. From the response rates obtained from past mailing campaigns and several other factors, we normally calculate the quantity to be mailed from the population. For example, Population - 1 mil Mail Quantity - 500K After scoring the population with the model, the records are ranked from 1 -10 based on the scores (records with highest scores as 1, and records with lowest scores as 10). To meet our required mail quantity of 500K, we would have to select records from Decile 1-5. But since we also want to find out how the model performs (in terms of response rates) in the bottom deciles (i.e. deciles 6-10), we want to mail small samples that fall in the deciles 6-10. So we select 10K records from each of the bottom deciles (6-10). The 10K quantity from each decile is not calculated using any scientific method. What I would like to know is if there is any scientific method to determine the additional quantities for just the bottom deciles. Hope it's more clearer now. Thanks again. Renji Gene Maguin <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 04:22 PM Please respond to Gene Maguin <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination Renji, I've been watching the back-and-forth on your question all day and I am wondering if this isn't just simple problem of determining how many people in each decile are needed to give a specified confidence interval about the expected value for that decile. If so, then I'd think that any basic survey sampling book would have that formula (and it's undoubtedly on the web in many places). Am I completely misunderstanding what you are wanting to do? Gene Maguin ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
|
Renji,
Yes, that's pretty much what I figured. Here is how I'm thinking of this problem right now. Suppose your expected (i.e., predicted) response rate is 2%. Then the variance is .98*.02 = .0196 and the sampling variance, assuming a sample of 10,000, is .0196/10000 = 1.96E-06 and the standard error is .00138. If you take a 1SD confidence interval your response rate would be .01862 to .02138, or 1862 to 2138 persons. Gene Maguin >>Let me try to explain it once more. We have a model which is used for scoring and selecting records for mailing campaign. From the response rates obtained from past mailing campaigns and several other factors, we normally calculate the quantity to be mailed from the population. For example, Population - 1 mil Mail Quantity - 500K After scoring the population with the model, the records are ranked from 1 -10 based on the scores (records with highest scores as 1, and records with lowest scores as 10). To meet our required mail quantity of 500K, we would have to select records from Decile 1-5. But since we also want to find out how the model performs (in terms of response rates) in the bottom deciles (i.e. deciles 6-10), we want to mail small samples that fall in the deciles 6-10. So we select 10K records from each of the bottom deciles (6-10). The 10K quantity from each decile is not calculated using any scientific method. What I would like to know is if there is any scientific method to determine the additional quantities for just the bottom deciles. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
|
In reply to this post by Guerrero, Rodrigo
Rodrigo,
Thanks for your reply. The overall mail response rate typically is between 0.5% and 1% depending on the product type, but the response rates go down further towards the lower deciles. Also a typical mail quantity is around a million pieces. However in the lower deciles, I just need to mail a sufficiently small quantity just to validate the overall model performance. The quantity you've suggested is far greater than what I can afford. Renji Abraham "Guerrero, Rodrigo" <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 12:26 PM Please respond to "Guerrero, Rodrigo" <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination If you have a large response rate, say 1%+, then you will not need a large sample to prove there is a difference in the deciles. If your response rate is low, say under .3%, then you will need something in the 200k range. Try using a population size that is the total size of the mail file, not your mailing. I get a mailing sample size of 500k for a .01% error range, 95% confidence, 7 Million population size, and a .15% response rate. RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 From: [hidden email] [mailto:[hidden email]] Sent: Monday, December 08, 2008 10:52 AM To: Guerrero, Rodrigo Cc: [hidden email] Subject: Re: mail quantities determination Rodrigo, The number I provided in my previous email were not real. The actual numbers for a real mailing per month for a single product is something like this. Total Mail pieces - 1 Mil Mailed per Decile Decile 1 - 200K Decile 2 - 200K Decile 3 - 200K Decile 4 - 200K Decile 5 - 200K ----------------------- Total - 1 Mil ----------------------- Since we also want to find the performance of the bottom deciles, we include a sample of 25K in each of the bottom decile (Decile 6 - Decile 10), based on the scores (quantity is not determined on a scientific basis). I tried the sample size calculator that you had suggested, but the sample size it showed was very small (not what I was thinking of). Or am I missing something? Thanks. Renji "Guerrero, Rodrigo" <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 11:11 AM Please respond to "Guerrero, Rodrigo" <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination As always, your response rate will be a key factor of sample size and how wide your margin of error needs to be. I have found that sample size calculators on the net to be helpful. I have a feeling that for direct mail, 2,000 pieces will not be enough. Go to: http://www.ezsurvey.com/samplesize.html RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of <R. Abraham> Sent: Monday, December 08, 2008 8:55 AM To: [hidden email] Subject: mail quantities determination I have a question regarding determining the mail quantities for a direct mail. This is a non-stat questions, so pls bear. Say for example we are planning to mail 50,000 mail pieces. We would select the records by the logistic regression scores, and maybe go down till the fifth decile to reach the quantity. But we also want to test the model overall in all the deciles (1 - 10 deciles) by mailing some quantities in each of the remaining deciles 6-10. My supervisor wants to determine the optimal quantities to mail in the rest of the deciles so that we can accurately validate the model performance in all the deciles. In the past we have randomly selected, say 2000 mail pieces based on the 50K total. But is there any scientific method to determine the quantities? Can someone share your thoughts pls? Thanks. Renji Abraham ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ====================To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
|
OK,
Sample size is really dependent on what response rate you expect to get. Let's say that Decile 4 has a response rate of .4 and that you would expect Decile 5 to have a 25% higher response rate, or .5%. To prove statistically significant result, Decile 4 response rate cannot be higher than .45% or lower than .35% This is a 12.5% error rate. At a 95% confidence, you will need a minimum sample size of 61,220. Try this calculator to help you. I think it will work better than the other I pointed you to. http://www.moore.rrd.com/wwwRMS/WhatWeHave/MinSample.asp RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of <R. Abraham> Sent: Friday, December 12, 2008 4:07 PM To: [hidden email] Subject: Re: mail quantities determination Rodrigo, Thanks for your reply. The overall mail response rate typically is between 0.5% and 1% depending on the product type, but the response rates go down further towards the lower deciles. Also a typical mail quantity is around a million pieces. However in the lower deciles, I just need to mail a sufficiently small quantity just to validate the overall model performance. The quantity you've suggested is far greater than what I can afford. Renji Abraham "Guerrero, Rodrigo" <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 12:26 PM Please respond to "Guerrero, Rodrigo" <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination If you have a large response rate, say 1%+, then you will not need a large sample to prove there is a difference in the deciles. If your response rate is low, say under .3%, then you will need something in the 200k range. Try using a population size that is the total size of the mail file, not your mailing. I get a mailing sample size of 500k for a .01% error range, 95% confidence, 7 Million population size, and a .15% response rate. RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 From: [hidden email] [mailto:[hidden email]] Sent: Monday, December 08, 2008 10:52 AM To: Guerrero, Rodrigo Cc: [hidden email] Subject: Re: mail quantities determination Rodrigo, The number I provided in my previous email were not real. The actual numbers for a real mailing per month for a single product is something like this. Total Mail pieces - 1 Mil Mailed per Decile Decile 1 - 200K Decile 2 - 200K Decile 3 - 200K Decile 4 - 200K Decile 5 - 200K ----------------------- Total - 1 Mil ----------------------- Since we also want to find the performance of the bottom deciles, we include a sample of 25K in each of the bottom decile (Decile 6 - Decile 10), based on the scores (quantity is not determined on a scientific basis). I tried the sample size calculator that you had suggested, but the sample size it showed was very small (not what I was thinking of). Or am I missing something? Thanks. Renji "Guerrero, Rodrigo" <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 11:11 AM Please respond to "Guerrero, Rodrigo" <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination As always, your response rate will be a key factor of sample size and how wide your margin of error needs to be. I have found that sample size calculators on the net to be helpful. I have a feeling that for direct mail, 2,000 pieces will not be enough. Go to: http://www.ezsurvey.com/samplesize.html RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of <R. Abraham> Sent: Monday, December 08, 2008 8:55 AM To: [hidden email] Subject: mail quantities determination I have a question regarding determining the mail quantities for a direct mail. This is a non-stat questions, so pls bear. Say for example we are planning to mail 50,000 mail pieces. We would select the records by the logistic regression scores, and maybe go down till the fifth decile to reach the quantity. But we also want to test the model overall in all the deciles (1 - 10 deciles) by mailing some quantities in each of the remaining deciles 6-10. My supervisor wants to determine the optimal quantities to mail in the rest of the deciles so that we can accurately validate the model performance in all the deciles. In the past we have randomly selected, say 2000 mail pieces based on the 50K total. But is there any scientific method to determine the quantities? Can someone share your thoughts pls? Thanks. Renji Abraham ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ====================To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: mail quantities determination
|
|
Thanks Rodrigo!! It really helps.
Also many thanks to Gene Maguin, and Art for your valuable suggestions. Renji Abraham "Guerrero, Rodrigo" <[hidden email]> 12/12/2008 06:22 PM To <[hidden email]> cc <[hidden email]> Subject RE: Re: mail quantities determination OK, Sample size is really dependent on what response rate you expect to get. Let's say that Decile 4 has a response rate of .4 and that you would expect Decile 5 to have a 25% higher response rate, or .5%. To prove statistically significant result, Decile 4 response rate cannot be higher than .45% or lower than .35% This is a 12.5% error rate. At a 95% confidence, you will need a minimum sample size of 61,220. Try this calculator to help you. I think it will work better than the other I pointed you to. http://www.moore.rrd.com/wwwRMS/WhatWeHave/MinSample.asp RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of <R. Abraham> Sent: Friday, December 12, 2008 4:07 PM To: [hidden email] Subject: Re: mail quantities determination Rodrigo, Thanks for your reply. The overall mail response rate typically is between 0.5% and 1% depending on the product type, but the response rates go down further towards the lower deciles. Also a typical mail quantity is around a million pieces. However in the lower deciles, I just need to mail a sufficiently small quantity just to validate the overall model performance. The quantity you've suggested is far greater than what I can afford. Renji Abraham "Guerrero, Rodrigo" <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 12:26 PM Please respond to "Guerrero, Rodrigo" <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination If you have a large response rate, say 1%+, then you will not need a large sample to prove there is a difference in the deciles. If your response rate is low, say under .3%, then you will need something in the 200k range. Try using a population size that is the total size of the mail file, not your mailing. I get a mailing sample size of 500k for a .01% error range, 95% confidence, 7 Million population size, and a .15% response rate. RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 From: [hidden email] [mailto:[hidden email]] Sent: Monday, December 08, 2008 10:52 AM To: Guerrero, Rodrigo Cc: [hidden email] Subject: Re: mail quantities determination Rodrigo, The number I provided in my previous email were not real. The actual numbers for a real mailing per month for a single product is something like this. Total Mail pieces - 1 Mil Mailed per Decile Decile 1 - 200K Decile 2 - 200K Decile 3 - 200K Decile 4 - 200K Decile 5 - 200K ----------------------- Total - 1 Mil ----------------------- Since we also want to find the performance of the bottom deciles, we include a sample of 25K in each of the bottom decile (Decile 6 - Decile 10), based on the scores (quantity is not determined on a scientific basis). I tried the sample size calculator that you had suggested, but the sample size it showed was very small (not what I was thinking of). Or am I missing something? Thanks. Renji "Guerrero, Rodrigo" <[hidden email]> Sent by: "SPSSX(r) Discussion" <[hidden email]> 12/08/2008 11:11 AM Please respond to "Guerrero, Rodrigo" <[hidden email]> To [hidden email] cc Subject Re: mail quantities determination As always, your response rate will be a key factor of sample size and how wide your margin of error needs to be. I have found that sample size calculators on the net to be helpful. I have a feeling that for direct mail, 2,000 pieces will not be enough. Go to: http://www.ezsurvey.com/samplesize.html RG Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of <R. Abraham> Sent: Monday, December 08, 2008 8:55 AM To: [hidden email] Subject: mail quantities determination I have a question regarding determining the mail quantities for a direct mail. This is a non-stat questions, so pls bear. Say for example we are planning to mail 50,000 mail pieces. We would select the records by the logistic regression scores, and maybe go down till the fifth decile to reach the quantity. But we also want to test the model overall in all the deciles (1 - 10 deciles) by mailing some quantities in each of the remaining deciles 6-10. My supervisor wants to determine the optimal quantities to mail in the rest of the deciles so that we can accurately validate the model performance in all the deciles. In the past we have randomly selected, say 2000 mail pieces based on the 50K total. But is there any scientific method to determine the quantities? Can someone share your thoughts pls? Thanks. Renji Abraham ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ====================To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |