9-2

Starting with the variables based on the donor's ZIP code, MALEVET (% Male veterans) would be quantitative. This variable will most likely not affect the operational performance; being a male veteran will not affect the operational performance, and the variable does not correlate with the goal. VIETVETS (% Vietnam Veterans) is also quantitative, and just like MALEVET, the variables do not correlate. WWIIVETS ( % WWII Veterans) is quantitative and does not correlate with operational performance. LOCALGOV (% Employed by the local government), STATEGOV (% Employed by the state government), and FEDGOV (% Employed by the federal government) are all quantitative, and just like the other variables based on the donor's zip code, they do not correlate with the operational performance.  Variables that are specific to the individual donors include HOMEOWNER, which is categorical, it identifies if the participants own a home or not. This variable will most likely not affect the operational performance. HIT (Number of times donor has responded to mail order offers other than PVA's) is quantitative and can be used to determine if the donor will participate in the PVA's fundraiser.  CARDPROM (Number of card promotions received lifetime) is quantitative. This variable can be of use in connection with how much the donor has donated to determine if the donor will give to the organization. MAXADATE (Date of most recent promotion received in YYMM Year  Month format) is a categorical variable. MAXADATE can identify when the last mail-in request was sent to the donor, giving an idea of the recent donor activity. NUMPROM (Number of

Strategy Achieving PVA's goal of decreasing their mailing expense by $20 million, analyzing certain variables will give us a better understanding of which donors to go after. Analyzing the variables that tell which donors regularly donate, how much they donate, and when last they donated, can pinpoint the data we wish to receive. The method I will use is descriptive statistics, analyzing the GIFTAMNT as the main variable in determining which donors to go after. Histograms and scatterplots for each variable will determine the meaningful correlations in donor choice.  Based off the information from the summary statistics, we see the average amount of money gifted is about $15, with a standard deviation of $12, meaning there is a huge jump of donations from the donators. Gifting anywhere from $0 to over $200. This histogram shows the data in visual, there are several outliers or abnormalities in the data. Statistical Method Statistical methods help turn data into information, which turns into knowledge and, from there, into logical business decisions. ( Sharpe et al., 2019 ) The two types of statistical methods needed to justify the organization's results are descriptive and inferential. These two methods go hand in hand in understanding and effectively using data.  The descriptive statistics from GIFTAMT shows the average amount of money gifted is about $15, with a standard deviation of $12, meaning there is a huge jump of donations from the

Summary of Results The results of the multiple regression analysis with p-values less than 0.05, show there are no random fluctuations between the estimate and p-vale. When the p-value and estimate are lower than 0, we can say there is a very unlikely that the variables are due to random fluctuations. The results from the R-square show there is a 54% chance that collectively these variables have a proportion of variance relative to GIFTAMNT. The R-square results are a little over half, which shows these variables contribute to a little over half of the donations. (APPENDIX AC) Improvement Solution After calculating the multiple regressions and analyzing the scatterplots between each variable and GIFTAMNT. PVA should target the donors based off these variables, HOMEOWNERS, LASTGIFT, MINRAMNT, CLUSTER2, AVGGIFT, and MAAXRAMNT. These variables all show positive significance to GIFTAMNT. If PVA changes their directions on sending mailing gifts to these donors they can reduce the amount of money they spend while increasing their donations, contributing to them saving $4 million dollars.

graph, and a visualization of the descriptive summary helps us understand and identify any patterns in the data. In this case, the mean or average amount of money donors have given and the standard deviation. The histogram is positively skewed, with outliers. The assumption of this model is normal, and the average amount of money donated is roughly $15. The histogram is unimodal and symmetrical. ( Sharpe et al., 2019 ) Appendix B - Scatterplot of HOMEOWNER and GIFTAMNT

understand the relationship between the variables, I recommend Multiple Regression to analyze and provide a solution to the problem of identifying whether looking at MALEVET is a valid variable to use when examining which variable will help acquire more donations. Appendix E - Scatterplot of WWIIVETS variable in comparison to GIFTAMNT Nonlinear The scatterplot WWIIVETS (% WWII Veterans) shows a nonlinear scatterplot with a linear assumption of the donors who are World War 2 veterans. Examining the scatterplot, I see an average of 10-50% of World War 2 veterans who donate about $15 each to the organization. To understand the relationship between the variables, I recommend Multiple Regression to analyze

Appendix H - Scatterplot of STATEGOV variable in comparison to GIFTAMNT Nonlinear The scatterplot STATEGOV (% Employed by the State Government) shows a nonlinear scatterplot with a linear assumption of the donors employed by the state government. To understand the relationship between the variables, I recommend Multiple Regression to analyze and provide a solution to the problem of identifying whether looking at STATEGOV is a valid variable to use when examining which variable will help acquire more donations.

solution to the problem of identifying whether looking at CARDPROM is a valid variable to use when examining which variable will help acquire more donations. Appendix K - Scatterplot of MAXADATE2 variable in comparison to GIFTAMNT Nonlinear The scatterplot MAXADATE2 (Date of most recent promotion received in YYMM Year Month Format) shows a nonlinear scatterplot and when they've received their most recent marketing from PVA. After analyzing the scatterplot, there are a significant number of donors who received the promotion in September 2015. To understand the relationship between the variables, I

recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at MAXADATE2 is a valid variable to use when examining which variable will help acquire more donations. Appendix L - Scatterplot of NUMPROM variable in comparison to GIFTAMNT Nonlinear The scatterplot NUMPROM (Number of promotions received lifetime) shows a nonlinear scatterplot with a linear assumption of the donors and how many received a promotion in a lifetime. After analyzing the scatterplot, there are a significant number of donors who received the promotion in September 2015. To understand the relationship between the variables, I recommend Multiple Regression to recognize and provide a solution to the problem of

identifying whether looking at NUMPROM is a valid variable to use when examining which variable will help acquire more donations. Appendix M - Scatterplot of CARDPM12 variable in comparison to GIFTAMNT Nonlinear The scatterplot CARDPM12 (Number of card promotions received in last 12 months) shows a nonlinear scatterplot of how many received a card promotion in the last 12 months. After analyzing the scatterplot, there are a significant number of donors who received the promotion in in June or on. Month six. To understand the relationship between the variables, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at CARDPM12 is a valid variable to use when examining which variable will help acquire more donations.

Appendix N - Scatterplot of NUMPRM12 variable in comparison to GIFTAMNT Nonlinear The scatterplot NUMPRM12 (Number of promotions received in last 12 months) shows a nonlinear scatterplot of how many received a promotion in the last 12 months. After analyzing the scatterplot, there are a significant number of donors who’ve gifted from $1-25. To understand the relationship between the variables, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at NUMPRM12 is a valid variable to use when examining which variable will help acquire more donations.

Appendix O - Scatterplot of NGIFTALL variable in comparison to GIFTAMNT Nonlinear The scatterplot NGIFTALL (Number of gifts given lifetime to date) shows a nonlinear scatterplot of how many donors have given gifts in their lifetime to date. After analyzing the scatterplot, there are a significant number of donors who’ve gifted from $1-25. To understand the relationship between the variables, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at NGIFTALL is a valid variable to use when examining which variable will help acquire more donations.

Appendix P - Scatterplot of CARDGIFT variable in comparison to GIFTAMNT Nonlinear The scatterplot CARDGIFT (Number of gifts to card promotions given lifetime to date) shows a nonlinear scatterplot of how many donors have given gifts in their lifetime to date. There are a between 1-10 group of donors who’ve donated a significant amount. To understand the relationship between the variables, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at CARDGIFT is a valid variable to use when examining which variable will help acquire more donations.

Appendix Q - Scatterplot of MINRAMNT variable in comparison to GIFTAMNT Positive Linear The scatterplot MINRAMNT (Amount of smallest gift to date) shows a positive linear scatterplot of how many donors have given the smallest amount to date. The smallest amount given is between $0-10. To understand the relationship between the variables, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at MINRAMNT is a valid variable to use when examining which variable will help acquire more donations.

Appendix R - Scatterplot of MINRDATE2 variable in comparison to GIFTAMNT Nonlinear The scatterplot MINRDATE2 (Date associated with the smallest gift to date—YYMM format) shows a nonlinear scatterplot of how many donors have given the smallest amount to date. The year with the most activity is 2010-2015. To understand the relationship between the variables even more in-depth, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at MINRDATE2 is a valid variable to use when examining which variable will help acquire more donations.

Appendix S - Scatterplot of MAXRAMNT variable in comparison to GIFTAMNT Positive Linear The scatterplot MAXRAMNT (Amount of largest gift to date in $) shows a positive linear scatterplot of how many donors have given the largest amount to date. The data shows more values near 0 going up to about $50. To understand the relationship between the variables even more in-depth, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at MAXRAMNT is a valid variable to use when examining which variable will help acquire more donations.

Appendix T - Scatterplot of MAXRDATE variable in comparison to GIFTAMNT Nonlinear The scatterplot MAXRDATE (Date associated with the largest gift to date—YYMM format) shows a nonlinear scatterplot of how many donors have given the largest amount to date. The data shows donors donated or gifted significantly more in 2015. To understand the relationship between the variables even more in-depth, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at MAXRDATE is a valid variable to use when examining which variable will help acquire more donations.

Appendix U - Scatterplot of LASTGIFT variable in comparison to GIFTAMNT Positive Linear The scatterplot LASTGIFT (Amount of most recent gift in $) shows a positive linear scatterplot of how much has been given by donors recently. The data conveys a significant amount of donors donate about $1-20 to the organization. To understand the relationship between the variables even more in-depth, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at LASTGIFT is a valid variable to use when examining which variable will help acquire more donations.

Appendix V - Scatterplot of AVGGIFT variable in comparison to GIFTAMNT Positive Linear The scatterplot AVGGIFT (Average amount of gifts to date in $) shows a positive linear scatterplot of how much has been given by donors recently. The data shows a lot of donors donate about $1-25 to the organization. To understand the relationship between the variables even more in-depth, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at AVGGIFT is a valid variable to use when examining which variable will help acquire more donations.

Appendix W - Scatterplot of CONTROLN variable in comparison to GIFTAMNT Nonlinear The scatterplot CONTROLN (Control number—unique record identifier) shows a nonlinear scatterplot of how much has been given by donors recently. The data graphs out the identifier on the scatterplot there is no real significance of the data, just a visual of the identifier for each donor. To understand the relationship between the variables even more in-depth, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at CONTROLN is a valid variable to use when examining which variable will help acquire more donations.

Appendix X - Scatterplot of HPHONE_D variable in comparison to GIFTAMNT Nonlinear The scatterplot HPHONE_D (Indicator variable for presence of a published home phone number: 1=Yes; 0=No) shows a nonlinear scatterplot of what donors have home phone numbers. The data shows there are more donors with home phone numbers than those without. To understand the relationship between the variables even more in-depth, I recommend Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at

HPHONE_D is a valid variable to use when examining which variable will help acquire more donations. Appendix Y -Scatterplot of CLUSTER2 variable in comparison to GIFTAMNT Nonlinear The scatterplot CLUSTER2 (Marketing Cluster Code—nominal field) shows a nonlinear scatterplot of marketing cluster codes. The data shows the marketing code associated with each donor. To understand the relationship between the variables even more in-depth, I recommend

Multiple Regression to recognize and provide a solution to the problem of identifying whether looking at CLUSTER2is a valid variable to use when examining which variable will help acquire more donations. Appendix Z - Scatterplot of CHILDREN variable in comparison to GIFTAMNT Nonlinear The scatterplot CHILDREN (Number of children living at home) show a nonlinear scatterplot of how many children live at home. The data shows significantly more donors have no children living at home. To understand the relationship between the variables even more in-depth, I recommend Multiple Regression to recognize and provide a solution to the problem of

identifying whether looking at CHILDREN a valid variable to use when examining which variable will help acquire more donations. APPENDIX AA – Multiple Regression between GIFTAMNT and every variable Multiple linear regression results: Parameter estimates: Parameter Estimate Std. Err. Alternative DF T-Stat P-value Intercept -139.7955 220.299 ≠ 03616 -0.63457163 0.5257 HOMEOWNER -0.78773019 0.30152264 ≠ 03616 -2.6125076 0.009 HIT -0.012254142 0.013146086 ≠ 03616 -0.93215138 0.3513 MALEVET 0.012543926 0.014349803 ≠ 03616 0.87415319 0.3821 VIETVETS -0.0051712551 0.011092739 ≠ 03616 -0.46618378 0.6411 STATEGOV 0.051863556 0.026706414 ≠ 03616 1.9419888 0.0522 LOCALGOV 0.0059163673 0.032229237 ≠ 03616 0.18357144 0.8544 WWIIVETS -0.0080025371 0.010018498 ≠ 03616 -0.79877616 0.4245 NUMPRM12 -0.092328116 0.061614074 ≠ 03616 -1.4984907 0.1341 CARDPM12 0.13278976 0.17743309 ≠ 03616 0.74839341 0.4543 NUMPROM 0.0040825446 0.03180203 ≠ 03616 0.12837371 0.8979 MAXADATE2 0.89998567 1.3859631 ≠ 03616 0.64935759 0.5161 FEDGOV 0.0012052412 0.036960721 ≠ 036160.032608705 0.974 CARDPROM 0.047829167 0.079530885 ≠ 03616 0.6013911 0.5476 LASTGIFT 0.59109941 0.021652584 ≠ 03616 27.299255<0.0001 MAXRDATE2 0.0098996672 0.0072744438 ≠ 03616 1.360883 0.1736 MAXRAMNT 0.0067681028 0.0081785452 ≠ 03616 0.82754362 0.408 MINRAMNT -0.19045816 0.039091704 ≠ 03616 -4.8720865<0.0001 MINRDATE2 0.0014126278 0.0061037357 ≠ 03616 0.23143659 0.817 CARDGIFT -0.12724046 0.076191589 ≠ 03616 -1.6700066 0.095 NGIFTALL -0.059310231 0.041263434 ≠ 03616 -1.4373557 0.1507 CLUSTER2 -0.017751322 0.0075444445 ≠ 03616 -2.3528999 0.0187 HPHONE_D 0.17543637 0.28458589 ≠ 03616 0.61646195 0.5376 CONTROLN -0.00000252495510.0000026204352 ≠ 03616 -0.96356326 0.3353 AVGGIFT 0.38267851 0.045026243 ≠ 03616 8.4990103<0.0001 AGE 0.0064900956 0.0095301905 ≠ 03616 0.68100377 0.4959 CHILDREN 0.040242306 0.24349601 ≠ 03616 0.16526885 0.8687

Analysis of variance table for multiple regression model: Source DF SS MS F-stat P-value Model 26287983.3911076.284167.24213<0.0001 Error 3616239484.1866.229031 Total 3642527467.57 Summary of fit: Root MSE: 8.1381221 R-squared: 0.546 R-squared (adjusted): 0.5427 Residuals stored in new column: Residuals Predicted values stored in new column: Pred. Values APPENDIX AB – Residual vs. Predicted scatterplot of GIFTAMNT and every other variable Negative Skewed

APPENDIX AC - Multiple Regression of HOMEOWNER, MINRAMNT, LASTGIFT, CLUSTER 2, and AVGGIFT Multiple linear regression results: Dependent Variable: GIFTAMNT Independent Variable(s): HOMEOWNER, MINRAMNT, LASTGIFT, CLUSTER2, AVGGIFT GIFTAMNT = 3.9652946 + -0.61398813 HOMEOWNER + -0.15162642 MINRAMNT + 0.60608717 LASTGIFT + -0.017142051 CLUSTER2 + 0.3768565 AVGGIFT Parameter estimates: Parameter Estimate Std. Err. Alternative DF T-Stat P-value Intercept 3.9652946 0.41579255 ≠ 03637 9.536714<0.0001 HOMEOWNER -0.61398813 0.29198558 ≠ 03637-2.1028029 0.0356 MINRAMNT -0.15162642 0.033891857 ≠ 03637-4.4738304<0.0001 LASTGIFT 0.60608717 0.020442098 ≠ 03637 29.648971<0.0001 CLUSTER2 -0.0171420510.0073948975 ≠ 03637-2.3180918 0.0205 AVGGIFT 0.3768565 0.040047463 ≠ 03637 9.4102464<0.0001 Analysis of variance table for multiple regression model: Source DF SS MS F-stat P-value Model 5284269.4956853.897850.24364<0.0001 Error 3637243198.0866.867771 Total 3642527467.57 Summary of fit: Root MSE: 8.1772716 R-squared: 0.5389 R-squared (adjusted): 0.5383 Residuals stored in new column: Residuals Predicted values stored in new column: Pred. Values