Positive theories of decision making are included in policy analysis to frame questions concerning welfare. Currently, well-being research lays out a number of proposals that would encompass non-standard models of choice, Bernheim, (2008) and these works consider an implicit or explicit definition of welfare. Some approaches assume that well-being flows from the achievement of defined objectives and these objectives guide choices. Other approaches consider that well-being is directly measurable [Berheim, Rangel (2007)]. Though the notion of well-being is intuitively appealing, the two existing approaches — revealed well-being and measured well-being — encounter serious conceptual difficulties. In the case of revealed well-being, Berheim et al. (2007)’s standard normative analysis evaluates the decision maker of well-being according to her true objectives, that is what her choices reveal; also analyzed in Sen [1973] and in Koszegi and Rabin [2008].
The central problem is that most of the approaches that study well-being are based on conventional rationalization and in order to capture revealed aspects of well-being; researchers should open the door to unconventional rationalizations. In this case, the literature examines two distinct strategies for rationalizing non-standard choice patterns: one strategy, broadens the preference domain while maintaining the assumption that choice always maximizes a single coherent objective function, the second strategy relaxes the latter assumption, either by adopting a model that accounts for divergences between preference and behavior, or by supposing that the individual pursues multiple conflicting objectives. For example, if we consider well-being choice as a consumption behavior; the research of Beshears, Choi, Laibson, and Madrian (2006) identify factors that increase the likelihood between revealed preferences and normative preferences: they considered “passive choice, complexity, limited personal experience, third-party marketing, and intertemporal choice. Their work discusses six approaches that contribute to the identification of normative preferences: structural estimation, active decisions, asymptotic choice, aggregated revealed preferences, reported preferences and informed preferences. Each of this approaches relies on a consumer behavior approach to infer some property of normative preferences without equating revealed and normative preference and it illustrates these issues with evidence from savings and investment outcomes”.
Other proposals measure well-being directly, as urged by Kahneman, D., Wakker, P.P., & Sarin. (1997), Kahneman (1999), Frey and Stutzer (2002), Kahneman and Sugden (2005), Layard (2005a, 2005b), and Koszegi and Rabin (2008). In this approach, well-being analysis might build upon a sizable body of work in psychology concerning the measurement of happiness and life satisfaction. Happiness research has already achieved a toe-hold in economics; see, e.g. Frey and Stutzer (2000,2004), Kimball and Willis (2006). However, much of the literature considers the concepts of”well-being,” and “self-reported happiness” as if they were equivalent. For example, in Bernheim (2008) a choice-based approach to welfare analysis can simplify the identification problem by equating well-being with a choice well-being. Likewise, one could simply equate well-being with self-reported happiness; Berheim et al. (2007) stated that an alternative interpretation of standard welfare economics holds that well-being is defined in terms of choice rather than underlying objectives. This perspective has a long intellectual tradition; see Little (1949).
~
Continuing with Berheim (2008), “choice” is more due to two types of justifications: instrumental and non-instrumental. Instrumental justifications provide reasons to expect that a choice public policy will improve people’s well-being. One version holds that while choice might be an imperfect proxy for well-being, no better proxy is available. While instrumental justifications portray choice as deriving its normative significance from its connection to well-being, non-instrumental justifications maintain that choices are normatively compelling simply because they are choices; hence it is possible to define well-being in terms of choice without implicitly invoking other objectives.
If we focus on the decision-making process, cognitive science focuses on one of the oldest subject areas of scientific reflection, human thinking itself. Empirical studies of happiness and life satisfaction have spread into a wide range of disciplines including computational social science and applied mathematics. In practice, economists rely on revealed preference indirectly, evaluating policy options by how they affect economic indicators. Benjamin-Heffetz-Kimball-ReesJones (2012) state that economists increasingly use survey-based measures of subjective well-being (SWB) as empirical proxies for utility. In many applications, SWB data are used to test or estimate preference models, or conduct welfare evaluations in situations where it is difficult to test and estimate credibly with choice-base revealed preference methods. The work of Benjamin-Heffetz-Kimball-Szembrot (2014) proposed an experimental methodology for estimating well-being’s relative marginal utilities which are the base to estimate a well-being index. They developed a theory in which utility depends on “fundamental aspects” of well-being measurable with surveys.
The main discussion to determine a well-being index from the perspective of Regional Science is the argument that pre-existing characteristic of the location plays an important role in determining people’s well-being. Aslam-Corrado (2011) showed that the different geographical locations where individuals live are considered drivers of well-being and these spatial drivers differ within and between places. Not only are individual-level effects significant, but also these spatial factors influence individuals. There are several spatial factors that we might expect have an impact on people’s well-being choices. To narrow these aspects, this research will consider high-crime and lower crime environments. The following section will review some fundamental work that analyzes choices and risk preferences in violent scenarios.
There are few studies in the literature that test choices in violent scenarios. Most of the analyses have focused on the economic or social consequences. For example, Hoeffler-Reynal (2003) examined the economic and human cost of civil war during 1960-1999. Using a global data set they found that a civil war of five years reduces the average annual growth rate by more than two percent. Their survey of the human costs of conflict showed that even long after the war stops the mortality rate increase mainly due to the destruction of public health infrastructure and population displacements. Zussman-Zussman (2006) evaluated the effect of contra-terrorism policies by conducting an indirect test to evaluate the effectiveness of Israel’s assassinations policy. Their approach was built on the fact that terrorism has had a significant adverse macroeconomic effect on Israel’s economy. Abadie-Gardeazabal (2003) investigated the economic effects of conflict in the Basque Country as a case study. Their main conclusions are that after the outbreak of terrorism in the late 1960’s, the per capita GDP in the Basque Country declined about 10 percentage points relative to a synthetic control region without terrorism. Besley-Mueller (2012) exploited data on the pattern of violence across regions and over time by estimating the impact of the peace process in Northern Ireland on house prices.
The main discussion to determine a well-being index from the perspective of Regional Science is the argument that pre-existing characteristic of the location plays an important role in determining people’s well-being. Aslam-Corrado (2011) showed that the different geographical locations where individuals live are considered drivers of well-being and these spatial drivers differ within and between places. Not only are individual-level effects significant, but also these spatial factors influence individuals. There are several spatial factors that we might expect have an impact on people’s well-being choices. To narrow these aspects, this research will consider high-crime and lower crime environments. The following section will review some fundamental work that analyzes choices and risk preferences in violent scenarios.
There are few studies in the literature that test choices in violent scenarios. Most of the analyses have focused on the economic or social consequences. For example, Hoeffler-Reynal (2003) examined the economic and human cost of civil war during 1960-1999. Using a global data set they found that a civil war of five years reduces the average annual growth rate by more than two percent. Their survey of the human costs of conflict showed that even long after the war stops the mortality rate increase mainly due to the destruction of public health infrastructure and population displacements. Zussman-Zussman (2006) evaluated the effect of contra-terrorism policies by conducting an indirect test to evaluate the effectiveness of Israel’s assassinations policy. Their approach was built on the fact that terrorism has had a significant adverse macroeconomic effect on Israel’s economy. Abadie-Gardeazabal (2003) investigated the economic effects of conflict in the Basque Country as a case study. Their main conclusions are that after the outbreak of terrorism in the late 1960’s, the per capita GDP in the Basque Country declined about 10 percentage points relative to a synthetic control region without terrorism. Besley-Mueller (2012) exploited data on the pattern of violence across regions and over time by estimating the impact of the peace process in Northern Ireland on house prices.
On the other hand, Akresh-Walque (2008) examined the impact of Rwanda’s 1994 genocide on children schooling. The authors combined two cross-sectional households’ surveys collected before and after the genocide. The authors used an identification strategy on which pre-war data was used to control for an age group’s baseline schooling to exploits variation across provinces in the intensity of killings and on which children’s cohorts were school-aged when exposed to the war. Their findings showed a strong negative impact of the genocide on schooling, with exposed children completing one-half year less education. Beber-Blattman (2010) investigated child soldiering, they followed a field study method interviewing former members of Uganda’s Lord’s Resistance Army. The theories they used can be captured by a principal-agent model that incorporates punishments, indoctrination, and age-varying productivity. They found that children are more easily indoctrinated and disoriented than adults, but are less effective guerrillas; hence the optimal targets of coercion were young adolescents. Rockmore (2011), studied the effects in general welfare due to violence in Northern Uganda. This work is the first estimate of the economic cost of risk violence separate from the actual experience of violence and finds that there is a significant mechanism by which conflict influences development.
Furthermore, the major contribution to the literature of choice and risk perception has been made by Slovic et al (1984,1980). Their work indicated that subjective judgments, made by experts are a major component in any risk assessment. If such judgments are faulty, efforts at public and environmental protection are likely to be misdirected. In their paper about perceiving fear, they made an analysis of biases exhibited by lay people and experts when they make judgments about risk. Their work analyzed similarities and differences between lay and expert evaluations in the context of a specific set of activities and technologies. On the other hand, Lerner et al. (2003) studied the effects of fear and anger on perceived risk of terrorism by setting an experimental design on which they predicted opposite effects for anger and fear on risk judgments and policy preferences. This research aims to identify well-being preferences in geographical areas affected by Drug War violence in Mexico; the following section provides a background of this war.
The Mexican government has addressed the Drug War with all of its military force. Beithel (2013) stated that ”Mexico’s drug trafficking-related violence has been punctuated by more than 1,300 beheadings, public hanging of corpses, killing of innocent bystanders, narcobloqueosfootnote{Block Trafficking by gangs members in strategic points of a city to avoid police authorities or military to accede to a violent event. Usually by kidnapping public transportation or private cars which are taken randomly and turn them on fire}, narcomantasfootnote{A message left by a drug cartel on a cloth banner, usually containing threats or explanations of criminal activity}, car bombs, torture, and assassinations of journalists and government officials involved in crime. In March 2012, the head of the U.S. Northern Command, General Charles Jacoby, testified to the Senate Armed Services Committee that Mexico at that time had succeeded in capturing or killing 22 out of 37 of the Mexican government’s most wanted drug traffickers. General Jacoby noted that their removal had not had “any positive effect” in reducing the violence, which continued to climb in 2011. Beithel (2013) stated that with the end of President Calderon’s term in 2012, several observers maintained that between 47,000 to 65,000 organized crime-related killings had occurred during his tenure, roughly 10,000 murders a year. The Trans- Border Institute (TBI) at the University of San Diego reported that 120,000 to 125,000 people were killed (all homicides) during Calderón’s administration. Addressing the question as to whether violence had leveled off or declined in 2012, TBI estimated that total homicides in Mexico fell in 2012 by 8.5%
The aim of this paper is to implement an experimental survey that will be the base to develop an index which will be used to analyze well-being preferences in risk environments. This experimental survey is based on a modified version borrow from Benjamin et. al (2014). We might expect several drivers through which drug war violence might affect people’s well-being preferences. De Choudhury et al (2014) found that emotional responses in social media might indicate desensitization to violence experienced in communities embroiled in an armed conflict. This diminishing sensitivity might suggest that people living in high-risk environments might act in a less risk averse way as they are not as concerned about risks that seem small relative to their general setting (Quiggin, 2003). However, the Mexican population explicitly exposed to drug war violence might perceive the environment as riskier and might behave risk averse when they make a pairwise comparison to other choices. The proposed methodology on this research is based on the preference-based theory based on an application of conjoint analysis (Green and Rao (1971)). In particular, we will use a revealed preference approach based on personal and policy hypothetical scenarios- to elicit such set of preferences. The adopted method could be a potential mechanism that public institutions could include in the process of policy making. In addition, there are two methodological points to take care when making this index: first, while the idea is to relate choice behavior to SWB measures, these measures are based on reports on general levels of SWB, whereas the survey questions elicit preferences and predictions comparing the SWB consequences of specific choices. Second, the data collected in research are based on choices framed in hypothetical scenarios. This is a limitation because the two might not be the same. However, using hypothetical scenarios it will allow addressing a much wider variety of relevant real-world choice situations [Benjamin et al. (2012, 2014)]
The temporality of this study will allow both to capture a ranking of preferences and to test the hypothesis that state that residential characteristics (e.g. levels of violence ) affect behavioral well-being responses to the conflict. However, the geographical variation in violence might be endogenous to some unobservable factors that are correlated with the behavioral responses, then both violence exposure and choices would be biased. Our study aims to contribute to the literature by identifying a set of well-being preferences of a population in violent and non-violent locations exposed to a plausibly exogenous change in the violence environment. The following section will discuss the methodology to develop this well-being index.
This experimental survey will be an approximation to construct a well-being index in Mexico. There are some differences between the method used in Benjamin et al (2014) and the present in this paper. First, in traditional SWB measures a utility would be modeled as a function of market goods as well as non-market goods such as leisure, social relationships and compromised with the community, crime awareness, violence exposure, which is included in the present survey. In addition, there is a combination of both types of goods since people also made decisions about private and public goods. Second, the survey in Benjamin et al. (2014) uses an important wide list of fundamental aspects of well-being. Since there are aspects of well-being that resemble each other, the present survey worked behind the idea of overlap-detection by using a variation on the original SP survey by reducing (1) the number of hypothetical scenarios from 30 to 15, and (2) by considering that the combinatorial and fundamental aspects of well-being must appear balanced on each scenario, bundles of 6 and 8 aspects allow respondents to pairwise comparisons. The present survey considers aspects that can be combined by incorporating hypothetical personal and policy aspects in a single scenario. This makes easy to compare if the relative marginal utility of the joint aspects has an effect on preferences. Furthermore, if we assume that a utility $u(w)$ depends on a vector $w$ of aspects of well-being, any vector proportional to the vector of marginal utilities $D_{w}$$u(w)$ can then be used as relative weights for combining the components of $w$ into an individual-level index that tracks small changes in well-being. For large changes in the aspects, the index can be used to track these changes.
The version of the survey on this research used hypothetical scenarios, 3 personal choice scenarios, 3 policy vote scenarios and 9 combined hypothetical scenarios. This experimental framework resembles the analysis reported in Rao (1971) Tversky and Griffin (2000). The current phenomena of violence and corruption related to drug violence in Mexico make the scenarios relevant to the Mexican population. In addition, the survey is directed to respondents residents in violent and nonviolent cities which put respondents in a situation to valued well-being aspects from different spatial characteristics. The prototype online survey will estimate a vector proportional to the vector of marginal utilities $D_{w}$$u_{w}$. In each scenario, it is possible to elicit respondent’s preferences between two options that differ only on how they compare a small set of aspects. This framework anchored on preference-based theory will help to fill the gap in understanding people’s judgments about well-being in risky environments: People will reveal their preference-based on hypothetical choices according to with the following hypotheses:
Residents in violent places might rank well-being’s emotional & affective aspects higher than people who reside in nonviolent places.
Residents in nonviolent places might rank well-being’s risk perception and personal safety aspects higher than people who reside in violent places.
The Benjamin et al. (2014)’s experimental survey requires that the list of aspects as an argument of utility must be exhaustive and non-overlapping. The list of candidates of fundamental aspects is based on (1) common well-being questions asked in large-scale surveys and (2) a mainstream of well-being studies. The present research made a modification to the former list by using 120 private aspects and 30 public aspects socioeconomic reality in Mexico. From a structural-functional view the items of the survey might be grouped in families of questions which can cover up to 12 general themes: 1 Happiness and life satisfaction, 2 Negative emotions, 3 Social network support, 4 Social engagement, 5 Health, 6 Freedom and personal autonomy, 7 Personal and progress inter-generational, 9 Corruption and state of law, 10 Security, 11 Drug War and 12 Government. ( See Appendix A.1)
We modified the experimental survey considering 15 hypothetical choice scenarios, one per screen and they are preceding by demographic questions and screen of instructions. The scenarios consider possible fundamental and combinatorial aspects of well-being. We have five different lists of well-being aspects, to simplify we name an “x-list” for those aspects that include 115-you private aspects, a named “X-list” contains 5-only you private aspects, a “y list” contains 115-other’s private aspects, a “Z-list contain 30-public aspects and a” z list” contains 115-people’s private aspects. Each scenario has either 6 or 8 aspects and they were selected randomly by the computer. The algorithm optimizes the selection of aspects to avoid repetition in each scenario ( see Appendix ). An example of a personal and policy scenario is reproduced below. Likewise in Benjamin et al (2014, p13), each screen has three components. First, the textit{preamble} frames the scenario as a choice between two options that are neutrally labeled “Option 1” and “Option 2” which describes the weight in the impact between the two options. The second component is the textit{aspect table} that describe the consequences of the two options. by marking them with an “X”. Finally, the third and final component is the textit{ choice question} which elicits the respondent’s stated preference between the two options
The Figure A.1 in the appendix shows an example for personal choice and policy vote. In the estimation procedures, the dependent variable is the response to the choice question, and the independent variables are the relative ratings of the aspects (“X’s”). Because the survey randomly assigns the weight of each aspect between the options, is possible to identify the relative marginal utility. The computer assigns the rating aspects “X’s” balance between the two options.
textit{Preamble}: In general, there are 15 versions. The first version introduces textit{personal-choice} scenarios. Here the opening clause “Imagine that you are making a personal decision between two options” will focus on private-good aspects on which a personal choice seems to be the relevant setting for eliciting these aspects that are weighted randomly by the computer and respondents analyze the impact on their well-being. This preamble uses a list of 115 ” private goods”-relating to an individual’s own well-being and use the words treatment textit{you} and textit{you only }. The difference between these two words are as follows: you (e.g your health) could in principle pertain to everyone, however, textit{you only} pertains to the respondent and could not pertain to everyone.
The second preamble version introduces policy-vote scenarios: the opening clause “Imagine that you and everyone else in Mexico are voting on a public policy.” Policy-vote scenarios will be related to public policy in Mexico (drug war, inequality, insecurity, corruption, and freedom of speech) and have two purposes. This preamble uses 30 public aspects and 120 well-being own to others/people aspects. This preamble uses the word treatment textit{people/nation/society} and when it refers to a country it refers to Mexico. Since this preamble considers public goods, it has been proposed 30 public aspects that cannot typically be affected by one individual’s personal choice. If a national SWB survey is to be used for evaluating policy, it may be useful to elicit the relative weights in a setup where the aspects pertain to everyone.
Personal scenarios will draw aspects randomly from a set of 120 private aspects (which consists of 115-you, 5 -only you-aspects). The policy scenarios proposed will draw aspects randomly from a set of 115 everyone- and 30 public-aspects; these 145 aspects are effectively public goods because they affect everyone in the same way. Following the strategy of Benjamin et. al. (2014,2012), and reducing the number of private and public aspects of each scenario this experiment will randomly draw aspects making groups of {6,8} aspects. The rating of each aspect is randomly assigned from a seven rating scale. The choice response scale is identical across all scenarios and it is designed to elicit the intensity of preference on a six-point scale.
The choice response scale is identical across all scenarios and it is designed to capture the preferences on a six-point scale (“Much Prefer Option 1″, “Somewhat Prefer Option 1″, “Slightly Prefer Option 1″, “Slightly Prefer Option 2″) and it is omitted an “indifferent” choice option. The the rest of combinatorial scenarios use the same frame to elicit preferences.
The recruitment in traditional behavioral experiments is typically conducted by using Amazon Turk. This paper focuses on the Mexican Population on which there is a limited existence of Amazon Turk users. One promising method for recruitment is the use of on-line advertisements(ads). On-line ads compare with traditional recruitment strategies are currently selected methods in terms of financial cost [Williams, Proetto, Casiano,&Frankling,2012], and a number of labor hours required [Battistella, Kalyan,&Prior,2010]. Facebook ads as a recruitment method have a low per-participant cost allowing to target ads to users profile most likely to be interested in a specific area. Some research has used Facebook advertisements as one of a variety of recruitment strategies with successful results see [Arcia(2014),Ahmed, Jacob, Allen & Benowitz (2011), Batisella et al., (2010), Jones & Magee (2011), Richiardi, Pivetta, & Merletti (2012); Ryan & Xenos (2011), Williams et. al (2012)]
~
Inclusion in this study was limited to the Mexican population aged 18 to 60 years old that have access to the Internet and have a Facebook account. Facebook was used for several reasons: First, the Digital Market Outlook showed that in 2015 the number of Facebook users in Mexico was 41.26 million. In 2018, the number of Facebook users is expected to reach 54.16 million. Second, social media have emerged as prominent information sharing ecosystems in the context of a variety of crisis, the response in social media and how they might indicate desensitization or sensibility to violence experienced have been and increased interest in HCI (Human Computing Interaction) research [see Choudhury,Monroy-Hernandez, Mark(2014)]. For these reasons, Facebook was selected to implement the recruitment. The sample collected auxiliary demographic variables -gender, marital status, education level, age, household size, income level, state and municipality-. The recruitment was made by using Facebook’s self-service application to create 4 ads, each accompanied by a different image but with identical copy: Do you live in Mexico? Participate in this survey and win dollars in this study. Clicking on the ad led users to the study survey welcome page on which it is explained that if they decided to participate they will have 1-20 chance of winning a $10 dollars in Amazon gift cards. The page explained that all information provided is anonymous and if they consent to participate they log in by using their Facebook account. Not recording of personal information was made. Ad images include a map of Mexico, a Cornell University Logo and a text “Survey Social Networks and Subjective Well-Being” (See supplementary Figure 1 attached in Annex 1). Ad images were adhering to Facebook guidelines. Ads were target such that they appear on desktop devices and it appeared on the Facebook desktop news feed and on desktop right column randomly.
The designated market area was violent and nonviolent Mexican cities selected based on the Mexico Peace Index 2015 calculated by the Institute for Economics & Peace (this index is based on 7 key aspects: homicides, violent crime, weapons crime, incarceration, police funding, organized crime and justice efficiency). The ads were shown on Facebook pages of Mexicans users aged 18 to 60 years old that stated the resident in Mexico. The optimization and delivery method selected was the cost-per-click (CPC) payment method. This option allows a buyer to pay only when users click on the ads whereas in the cost-per-thousand-impressions model cost is based on the number of times (in thousands) that the ad appears on user’s screen. The ads in CPC model entering in an auction where the buyer can choose a bid on what link clicks are worth for the buyer; in this case the model selected was an automatic bid where Facebook set automatically the bid to get most clicks at the lowest price. The delivery method selected was a standard approach which it means that the ads were shown throughout the day. Cornell University, Institutional Review Board for Human Participants approved the procedure (Protocol ID#: 1511005970). The study was coded using NubiS which is a complete data collection tool that has been developed by the team behind the Understanding America Study (UAS) panel at the Center for Economic and Social Research at the University of Southern California and it was hosted by the same research center.
Prototype Model: Personal and Policy Benchmark specifications
Each observation $s$ captures the information from the scenario $s$ faced by the respondents, corresponding to the prospected survey screens like in the example, $StatedPreference_{s}$ will encode the response to the choice question, while $AspectRatings$ will encode the differences between the two options. The same model will be used to analyze the results for personal, policy vote and the rest of combinatorial scenarios. This experimental design uses six points on the choice scale assigning six numerical values $(-3,-2,-1, 1,2,3)$, and the seven columns in the aspect table that have the values $(-3,-2,-1,0,1,2,3)$. The econometric specification was made by coding the verbal scale of the independent and dependent variable as exogenous imposed numerical scales. To obtain the values for the choice scale we used the standard normal cdf to calculate the expected value of latent preference intensity conditional on observed intensity category; linearly scale this conditional expectations to be on (-1,+1) interval, symmetrize them around zero by taking the average of the absolute value of each corresponding conditional expectations.
The Facebook ads ran from January 11 to June 10, 2016. According to data provided by the Facebook self-service ad management application, the ads were shown a total of 2,455,893 times to 848,598 Facebook users over the 33-weeks campaign for a mean of 2.89 impressions per unique user. The ads received 7,828 clicks by 6,490 unique users for a total click-through rate of 0.32% at a mean cost of $0.08 per click. The success of the ads is measured by clicks and click-through rate which varied substantially depending on the ad image. The total cost of the Facebook ads was $578.44. The participants mean age is 25-29 years. Only participants who reached the end of the survey were considered for analysis and the demographic characteristic is summarized in Table 1. All participants reported their State and County of residence. The sample considered is (N= $259$) internet completed survey respondents which are demographic diverse -albeit no representative- sample of the Mexican adult population.
The estimation of personal choice scenarios show that personal growth and happiness measures- “Your ability to dream and pursued your dreams”, ‘Your ability to be yourself and express yourself”, “Your having a role to play in society”, “Your success at accomplish your goals”,” You are having more options and possibilities in your life and the freedom to choose among them”, “Your ability to fully experience the entire range of healthy human emotions”, ” Your feeling that you have enough time and money for the things that are most important for you”, ” -are the highest ranking aspects more than the affective and evaluative SWB ones- ” how much time you feel happy”[70] or “how happy you feel”[83] and positive affect measures “How often you smile of laugh”[38]. The main results are reported in Table 2 in appendix
~
textit{Risk, Safety emotions and health}: consist of seven own introspection measures and three aspect classes. Two own measures related with Drug War Violence-” Your neighborhood is safe (no thieves, extortion, smuggling, kidnapping or murders)”[9], and “Your home is free of addictions”[11] are following by ‘Your health”[19] and ‘Your physical safety”[27] -which aspects lies on the top of the table with measures 0.231-0.220 and 0.188-0.165.”Your mental health and emotional stability” lies below this rank [87].
textit{Negative emotions}: Benjamin et al (2014), Deaton et al (2011) suggest capturing measures negative emotions. The survey captured ten measures which have the following rank; sleep[18], stress[20], anger[31], feeling full of energy [41], anxious[52], frustration[75] and sad [97].
~
textit{Social Media and Eudaimonic SWB} By using social media (Facebook, Twitter) we observed people reactions towards different events of corruption and violence in Mexico which suggest using measures to capture people involved in the community. The aspects “Your sense that are making a difference, actively contributing to the well-being of others”[25] and “Your sense of community, belonging and connection with other people”[67] lied on first and middle on Table.
textit{Happiness and life satisfaction}: Table 3 on Appendix show preferences of the respondents who live in Non-violent locations. This group ranked evaluative and affective SBW aspects such as “Your sense that you know what to do when you face choices in your life”, ” The happiness of your friends”, ” Your ability to pursued your dreams” and “Your health” with coefficients in the range 0.763-0.503. Residents of violent places [N=164] in Table 4 rank as follows: “You are getting the thins you want out of life”, ” Your sense that everything happens for a reason”, “Your knowledge, skills, and access to information,” and “Your ability to fully experience entire range of healthy human emotions” with coefficients in the range 0.416-0.267.
Preferences related with textit{Risk,Safety emotions and health} showed that residents of non-violent locations have the following order of preference- “Your health”[4], “Your physical safe and security”[6],
“Your home is free of addictions”[9], “Your neighborhood is safe (i.e no thieves, extortion, smuggling, kidnapping or murders)[17], “In your neighborhood there is not presence of drug trafficking organizations”[25],”you feel safe in your neighborhood”[70], and “your neighborhood is free of drug war violence”[83].
~
With respect to the residents of violent location, Table 4 show the rank in the following order: “Your neighborhood is safe (i.e no thieves, extortion, smuggling, kidnapping or murders)[19], Your health”[55], “Your home is free of addictions”[64], “Your physical safety and security”[92],”you feel safe in your neighborhood”[94],”your neighborhood is free of drug war violence”[95], “In your neighborhood there is not presence of drug trafficking organizations”[118].It is important to notice that the rank of negative emotions show slight differences between non-violent and violent locations with the following rough order of preference, listed as [non-violent]-[violent]: sleep[5]-sleep[48], stress[18]-stress[12], anger[23]-anger[40], worry[31]-worry[114], energy[37]-energy[81], depressed[73]-depressed[101], sad[74]-sad[87], fear[100]-fear[93], pain[101]-pain[104].
In summary, this stated-preference-based approach yields high marginal utilities estimates on measures related to – the quality of life, well-being of family- and with aspects related with risk and safety emotions such as “physical safe”, “security”,”home free of addictions”, “neighborhood safe”, or “neighborhood free of drug violence”. These measures have not previously been asked in large-scale surveys. The results -albeit not representative of the Mexican population- allow distinguishing preferences of residents in violent and non-violent locations.
Policy scenarios are reported in Table 2 of appendix and use hypothetical specifications related with public aspects. In such scenarios, respondents vote on policy trading off 115 everyone aspects (personal aspects that pertain to everyone) and 30 public aspects (public goods that pertain to socioeconomic and political issues in Mexico). There are three policy scenarios that each respondent faced. The comparison between the two panels is possible due to the use of same numerical scales as in the personal scenario case. The survey collected fewer data in these policy scenarios, errors standard are larger and the correlation between the 115-textit{you} and 115-textit{everyone} coefficient pairs is low. However, the results may reflect respondents greater uncertainty towards other. If the set of public a policy scenarios were high correlated it might be interpreted as a confusion to distinguished between scenarios that affect the respondents and aspects that affect the society on general. We observed that respondents state preferences with high intensity in policy scenarios. The high rank textit{30 public aspects} include: ” Freedom of Conscience and belief in Mexico”[1], “The amount of order and stability in Society”[16],” Equal Opportunity in Mexico”[18],”Low rate of unemployment in Mexico”[28],”The average income of people in Mexico”[35], “The low the rate of criminality in your town ( i.e no thieves, extortion, smuggling, kidnapping or murders)”[37]. “The well-being of people in Mexico”[39], “The low rate of addictions at home in Mexico”[41].” The efficiency of the local government of your town to provide basic public services”[45], “The general confidence in local institutions in Mexico”[57], “Low rate of drug trafficking in your town”[60].
Policy scenarios-textit{everyone} aspects related with textit{Risk,Safety emotions and health} showed a ranking order as follows: “People’s neighborhood is free of drug trafficking”[5] and ” In people’s neighborhood there is not presence of drug trafficking organizations”[14],” People feel safe in their neighborhood”[11], “People’s neighborhood is free of drug violence ( i.e murders by drug gangs, shootings, narcobloqueos, public hanging of corpses, beheadings)”[34], “People physical safety”[65]. Negative emotions are ranked on the following order of preference, frustration[2], anger[10], stress[58], pain[61], fear[77], sad[107].
The corresponding textit{30 public aspects} stated preferences for violent and non-violent locations are reported in Tables 3 and 4 on appendix, these include “Freedom of Conscience and belief in Mexico [1;20]”, “People’s neighborhood is free of drug trafficking [2;13]”, “People’s neighborhood is free of drug violence ( i.e murders by drug gangs, shootings, narcobloqueos, public hanging of corpses, beheadings) [42;56]”, “Low rate of unemployment in Mexico [56;25]”.
Policy scenarios-textit{everyone} aspects related with textit{Risk,Safety emotions and health} showed a ranking order as follows:” People feel safe in their neighborhood”[17,90], “People mental health and emotional stability [9,127], “Peoples’ health [138,7]” By comparing the overall results of personal and policy scenarios we can reject our hypotheses. There is a set of preferences on policy aspects related with security and drug violence. We suggest to interpreting this set of aspects carefully, since, one concern is that the geographical variation in violence might be endogenous to some unobservable factors that are correlated with the behavioral responses, in addition to reference group size of analysis ( violent and non-violent locations) might influence an affect the identifiability. However, the results provide new findings on Human Computer Interaction research, in his case affective responses indicate has not yet reach desensitization to violence experienced in communities embroiled in an armed conflict.