Abstract

We present two methods-oriented studies on sexuality, one dealing with the discussion of sexuality in the context of a relationship, the second with the societal protection of sex workers. Both studies used consumer respondents to evaluate systematically varied combinations of messages about the topic, the combinations created by experimental design, following the method of Mind Genomics. Study 1 on discussions of sexual intimacy presents Mind Genomics to understand the way people process information, their criteria for decision-making, and the nature of possibly easy-to-understand mind-sets, i.e., different criteria of importance assigned to the same pieces of information. Study 2 on the protection and recourse given to legal workers shows how to assess the interaction between person and situation as drivers of judgments and drivers of engagement. Both studies point to the emerging science of Mind Genomics as an easy, rapid, and cost-effective ways to create archival databases, to introduce new ways of thinking, and to democratize research world-wide, respectively.

Introduction

During the past three decades the focus of researchers has steadily increased on issues involving intimacy, specifically sexual intimacy between consenting partners (love, romance), as well as sexual intimacy as a business (sex workers.) Sexuality in its many manifestations has always attracted research because of its centrality in daily life, but as society has evolved, issues of sexuality have become intertwined with emotions, with public health (e.g., sexually transmitted disease), and finally with issues of the law (e.g., prostitution and the issues revolving around sex workers.)

The topics of love, sexuality, sexual exploitations, and societal reactions each have spawned enormous literatures. Table 1 shows the number of ‘hits’ for Google® and for Google Scholar®, for each of these topics, at the time of this writing, December 2019,

Table 1. Number of citations dealing with sex and its ramifications.

No set of studies can hope to be comprehensive, given the long history of the study of sexuality, the many manifestations in daily life, and the many cultures as well as stages of individual development that must be considered. Rather, we introduce here a new approach to the study of sexuality, the science of Mind Genomics, designed to take small snapshots of a topic, focus in depth on a specific, limited topic, and work with small, affordable samples of respondents.

The worldview of Mind Genomics involves a small, limited topic, investigating the patterns of decision making within that topic. Rather than emerging out of the history of the hypothetico-deductive method, isolating a variable and studying that variable in an experiment, Mind Genomics proceeds in the reverse direction. One might think of the Mind Genomics researcher as a cartographer faced with a new land. The cartographer measures the relevant variables of a topographical area, deduces the nature of the structure below, and maps the land. The cartographer creates maps, not theories. In the case of Mind Genomics, the ‘land’ is the world of sexuality. The cartography of this paper deals with the reactions to issues of sexual intimacy (one set of experiments), and reactions to issues of sex workers (another set of experiments.)

Exploring two topics of sex using Mind Genomics to generate insights and hypotheses

The topic of sexual behavior spans a wide range of topics, from the physical to the emotional to the legal, and to the societal. It is impossible to cover even a very small fraction of the topics with a set of experiments or surveys. The strategy of this paper is to demonstrate how the emerging science of Mind Genomics can generate an affordable, powerful database at the start of a research initiative, using simple ideas, simple thinking, consumer research, and powerful analyses, meaningful even with samples that are traditionally considered ‘small’.

The emerging science of Mind Genomics (Moskowitz & Gofman, 2007) [1], traces its intellectual heritage to the systematized thinking using experimental design to structure the test stimuli, as well as to sociology and consumer research for transforming the ideas into questions to be answered, and finally to the Socratic method to create the system as an inductive knowledge-development technique, easily applied in practice

Experimental design

Experimental design allows a researcher to understand the effects of a variable, either tested along in ‘splendid isolation’ or tested as part of a mixture (Box, Hunter & Hunter, 1978) [2]. Mind Genomics deals with the ordinary situation, wherein a person is presented with a combination of ideas, as the typical situation of daily life. The person responds to the combination, making a decision. But just what specific component of the combination or set of components ‘drive’ that decision? Experimental design sets up efficient combinations of independent variables, messages or elements in the language of Mind Genomics. It is the response to these systematically created mixtures, which, through regression reveals, quite directly the contribution of each Message or Element to the response. The response, in turn, is what the respondent answers.

Sociology and consumer research

These social science disciplines rely upon the responses of people to questions about behavior, or upon the measurement of the behavior of people in situations, i.e., upon attitude versus upon behavior, respectively. Where possible a meaningful behavioral measure may be better than an attitude, although the term ‘meaningful’ is important as a qualifier.

Over almost a century there has been a subtle current of belief that implicit measures are better than explicit ones, e.g., that EEG (brain waves) or GSR (activation) or pupil behavior (dilation, pupil motion) somehow are better than simple attitudinal ratings because the former are more objective, more biological (Boring, 1929) [3].

The foregoing use of ‘meaningful’ is not what is meant here. Rather, the term ‘meaningful’ is used in the sense that the measure to be meaningful must be a direct correlate of the mind of the person, whether person in society or an ordinary citizen faced with a choice. Mind Genomics uses the responses to combinations of messages, i.e., combinations of elements as the meaningful measure, since a great deal of behavior in everyday life is responses to mixtures. Mind Genomics goes the additional step by creating combinations of these messages, presenting them to respondents, measuring the reactions, and then estimating the contribution of each message.

The Socratic Method

The approach is grounded empiricism, not in the hypothetico-deductive method. There is no hypothesis to be tested. Rather, there is a topic to be studied. The topic of interest is presented to the researcher, who must create four questions which ‘tell a story’ about the topic. The questions are not necessarily final, but rather represent the way the topic is thought about, either those who are grounded in the topic, or even novices with no idea at all, so-called ‘newbies’. The four questions each motivate four answers, or a total of 16 answers, as shown in the next sections. The researcher then combines these answers into small vignettes, obtains responses to the vignettes, and shows how the different answers shed light on the topic.

The best way to show the Mind Genomics method is through a case history, dealing with a topic relevant to an individual, or even beyond the individual to a group, and to society. This paper focuses on two aspects of sexual behavior, the first dealing with discussions of sexual intimacy and disease protection between consenting partners, the second dealing with protection of the ‘sex’ worker. These are but two of the perhaps hundreds of topics in the rainbow of topics in sexuality. We show how a one-day experiment can produce data for each topic, making it feasible to explore hundreds of topics about sexuality in the time frame of a year, with affordable, rapid, insightful and archival data.

Study 1 – Discussinag disease prevention between two consenting & emotionally-involved partners

A great deal has been written about sexual relations between consenting partners, from issues to measurements (e.g., Fisher et. al., 2013; Montesi, et. al., 2013; Stephenson, et. al., 2010). [4, 5, 6] The topics range from the emotions felt by the participants to the behavior of adolescents versus older individuals, and on to the issues caused by the ravages of sexually transmitted disease (Harvey et al., 2016; Katz et al., 2000; Peplau et. al., 2007; Widman et. al., 2006). [7, 8, 9, 10] Our focus in this experiment is the couple’s discussion of issues around the prevention of sexually transmitted diseases using methods under their control. The study was motivated by author Ortiz’s plan to sponsor a campaign to reduce sexually transmitted disease.

Method

The Mind Genomics study begins with the creation of the four questions and the four answers to each question. These appear in Table 2 and were created by author Ortiz as part of a campaign against sexually transmitted diseases. The important thing to realize from Table 2 is that the study does not exhaust the topic. Indeed, Mind Genomics studies are not designed as single, exhaustive treatments of a subject, treatments which generate a large volume of disparate information. Rather, Table 2 shows a preliminary attempt to understand four aspects of the topic. The reality is that there may be 40 or 400 aspects of the topic. When one attempts to cover a topic thoroughly, the entire endeavor may collapse because, in common folk wisdom ‘the perfect is often the enemy of the good.’ The Mind Genomics strategy is to create a set of such small studies, accrete the results, and identify emergent patterns ‘from the bottom up.’ Mind Genomics represents the inductive way to learn, i.e., by discovering patterns, rather than by confirming or disconfirming ingoing hypotheses.

Table 2. Sexual Intimacy: Four questions and four answers to each question.

The researcher combines these elements (the answers A1-D4) into small, easy to read combinations, so-called vignettes. The actual experimental design is created as ‘kernel,’ in which the 16 elements are statistically independent of each other, allowing for subsequent analysis by OLS (ordinary least-squares) regression. The kernel, or basic experimental design is permuted so that the design structure remains the same, but the individual combinations changes in a permutation pattern (Gofman & Moskowitz, 2010.) [11] Table 3 shows the experimental design for one respondent (independent variables in the subsequent analysis), and then the ratings, binary transformation (Bin) and Consideration or response Time (CT) (the dependent variables in the subsequent analysis.)

Table 3. Example of the data from one respondent, prepared for statistical analysis.

The structure of the vignettes follows these conventions:

The experimental design, metaphorically a booklet of recipes of the same ingredients to create different dishes. The experimental design specifies the composition of vignettes comprising two elements, three elements, and four elements, respectively.

Each respondent is required to evaluate 24 vignettes, all different from each other. Across the 24 vignettes, each element appears five times and is absent 19 times. A vignette comprises at most one element or answer from any question (Table 3.) This strategy of testing both complete vignettes (one answer from each question) and incomplete vignettes (no answer from either one or two questions) ensures that the analysis of the data by OLS (ordinary least-squares) regression generates coefficients having absolute value, where ratios of coefficients are meaningful.

Each respondent evaluates 24 unique, different vignettes. The underlying experimental design ensures that the 24 vignettes for each respondent differ from the 24 vignettes for any other respondent. The benefit to this permutation scheme is that the Mind Genomics experiment covers a great deal of the so-called ‘design space’. The benefit to the researcher is one need not know ‘what works’ ahead of the study. In contrast, in other research methods using experimental design of messages (so-called conjoint analysis; Green & Srinivasan, 1990), [12] the researcher selects one set of combinations, and tests that set with many people in order to suppress the variation by averaging. Whether averaging out the variation in the typical approach or averaging out the variation by looking at a great deal of the design space ultimately proves to be better is still a matter of dispute.

Table 3 show the 24 vignettes as rows. The 16 elements are shown as A1-D4, corresponding to the four questions and the four answers in each question featured in Table 1.

The column labelled Rat is the 9-point rating assigned to the vignette by the respondent. Table 3 shows the respondent ratings of all 24 vignettes.

The column labelled Top3 is the ‘binary’ transformation of the 9-point ratings, with ratings of 1–6 transformed to 0, and a very small random number added to the transformed number For ratings 7–9 the rating is transformed to 100, and again a very small random number is added to the transformed number.

The addition of the random number is done so that the regression analysis will not ‘crash’ when the analysis creates individual-level models to generate mind-set segments. When a respondent rates all 24 vignettes between either 1–6 or between 7–9, respectively, then transformed ratings will all become either 0 or 100, respectively for the Top 3 measure, and the regression model using the Top3 measure as the dependent variable will ‘crash.’ Adding a small random number prevents that crash, ensuring that the statistical analysis proceeds without incident.

Finally, the column labelled CT is Consideration Time, or Response Time, defined as the number of seconds elapsing between the presentation of the vignette on the screen and the rating assigned by the respondent. The vignettes are short, so that any Consideration Time longer than 9 seconds is assumed to reflect the respondent’s multi-tasking and is brought to the value 9.0. The use of the term Consideration Time makes the number more meaningful to the reader, because the magnitude of the CT can be associated with the time it takes the respondent to consider the element.

The analysis of Mind Genomics data proceeds in a straightforward manner, enabled by the experimental design for the creation of the different vignettes. The experimental design created for a single individual ensures that the 16 elements or answers for that individual appear independently of each other among the 24 vignettes. Putting together a set of such experimental designs, each different from the others simply by a permutation scheme, maintain the statistical independence of the 16 elements.

An easy-to-interpret analysis (OLS Regression) relates the presence/absence of the 16 elements to the binary rating. OLS regression uses the 16 elements as independent variables, and the binary transformation, Top3, as the dependent variable. The regression incorporates the relevant cases, namely the 24 rows from each respondent who belongs to the subgroup. Thus, when it comes to the model or equation for ‘males,’ only the data from the male respondents are used. Each male respondent contributes 24 cases or observations.

The regression model estimates the parameters of this simple equation: Top3 = k₀ + k₁(A1) + k2(A2) … + k₁₆(D4). Top3 is defined as ‘comfortable talking about the topic.

The parameters for the total panel and key subgroups appear in Table 4. The table shows total panel, gender, age groups, relationship status, and the response from all respondents, but broken out into the results from Vignettes 1–12 (Half1) and then from Vignettes 13–24 (Half2). This final comparison shows us whether the respondents ‘change their criteria’ as the study proceeds

Table 4. Parameters (additive constant, coefficients) for equations relating the presence/absence of the 16 elements for binary transformed rating ‘comfortable talking about the topic (prevention of sexually transmitted disease.)’. The table is sorted by the coefficients for the total panel.

The additive constant is a measure of basic comfort talking about the topic, but with no elements in the vignette. The basic comfort for the total panel is 61, meaning that in the absence of any elements, 61% of the responses will be 7–9. That is, about 3 in 5 times the response will be ‘comfortable.’ The only group showing less comfort is the younger respondents (additive constant = 43), whereas their complementary age group, the older respondents, age 25 and older, is more comfortable (additive constant = 73).

There are some elements which ‘stand out’ from the others, topics about which the respondents feel very comfortable discussing. The elements below list the strong performing elements. Although there are strong performing elements, as shown by the coefficient, an underlying theme or story does not appear.

Total – None

Males

Knowing partner’s prior sexual history
Getting tested regularly
Both partners using birth controls
Talking about safe sex during the first conversation about intimacy

Females

Partner says: “Let’s use protection”
Discussing STD precautions planning through texting

Age 25 or older – None

Age 24 or younger

Talking about safe sex before engaging in sexual activity
Both partners using birth controls

Single – None

In a relationship

Both partners using birth controls
Getting tested regularly
Knowing partner’s prior sexual history
Using condoms during sex

First half of the individual’s vignettes (vignette 01- vignette 12)

Both partners using birth controls
Partner says: “Safe sex is the best move”

Second half (vignette 13 – vignettes 24)

Discussing STD precautions planning through texting

An increasing focus of Mind Genomics is upon Consideration Time (CT). In experimental psychology the term Consideration Time may be replaced by either Reaction Time or Response Time. CT is defined as the number of seconds (to the nearest tenth of second) between the presentation of the test stimulus, the vignette, and the rating assigned by the respondent. The term Consideration Time’ is used to underscore that the response is not only the time to perceive and react, but to read and consider.

The computation of response time is straightforward. The Mind Genomics algorithm relates the response time to the presence/absence of the elements, using the same form of equation as done for the Top3 value (comfort, in Table 3). The only difference is that the equation for consideration time has no additive constant. That is, the ingoing assumption is that without any elements in the vignette, the consideration time should be 0.

Table 5 shows the six elements with long consideration times in at least one group of responses or in either the first half or the second half of the Mind Genomics experiment, respectively. In turn, Table 6 shows the Consideration Times for the full set of elements across the different subgroups.

Table 5. The six elements showing long (estimated) consideration times of 1.5 seconds or longer.

To give a perspective, the typical consideration time of a full vignette for less serious topics may be 1–2 seconds. People make up their mind quickly for topics considered to be of minor import, perhaps System 1 in the language of Nobel Laureate Daniel Kahneman in his book Thinking Fast, Thinking Slow (Kahneman, 2011) [13] In contrast, topics of sexual discussion may involve System 2, the slower, more deliberate thinking which is the hallmark of a serious topic.

Table 6. The full set of consideration times for the total panel and key subgroups.

Three emergent mind-sets

One of the ongoing tenets of Mind Genomics is that within any topic where human judgment plays a role, there are usually at least two different groups of people, having different criteria about the same topic. That is, for those topics involving judgment, people disagree. The disagreement may be minor, or major, depending upon the people, the topic, and the information presented.

Researchers have uncovered these differences as a matter of course when studying the criteria for human judgment. The differences themselves exist, but Mind Genomics goes one step further beyond noting the differences. Mind Genomics attempts to uncover, classify and then understand the nature of these differences, creating a set of mind-sets embodying the different criteria for judgment. Mind Genomics can go one step further, creating a tool, the PVI (personal viewpoint identifier), to predict the way new people will respond to the information, i.e., an assignment tool. The analogy is to color science and colorimetry. Mind Genomics creates the ‘color science’ for a topic, and then crafts the tool to identify these mind-sets in the population at large. In the interest of length, the PVI for these data are not presented in this paper.

Mind Genomics follows these steps to identify the emergent mind-sets, with all the information needed present in the data from the basics study:

1. Create the data matrix, with the rows corresponding to the respondents, and the columns corresponding to the elements. For the data presented here, the data matrix comprises 16 columns, one for each element. (The additive constant is not used). The data matrix comprises 50 rows, one row for each respondent.
2. Define the distance between rows (respondents) by a single number. The choice of the number can range from the simple Euclidean distance to a distance between patterns, defined as (1-Pearson correlation between two rows). Mind Genomics uses the latter (1 – Pearson Correlation, or 1-R).
3. The distance metric (1-R) ranges from a low of 0 when two rows are perfectly correlated, to a high of 2 when two rows are perfectly but inversely correlated.
4. The program, k-means clustering (Dubes & Jain, 1980), [14] creates complementary and exhaustive groups, called clusters or segments.
5. Mind Genomics creates two clusters and assigns each respondent to one of the two clusters.
6. Mind Genomics then creates three clusters, and assigns every respondent to one of the three clusters
7. The data from respondents in each cluster are analyzed separately, first for the model for comfort (Top3) and then for the model for Consideration Time.
8. The strongest performing elements for each set of clusters are used to determine whether there is a coherent story (interpretability), and whether the number of clusters is as few as necessary (parsimony). It is important to have as few clusters (mind-sets) as possible, provided that the clusters are interpretable, i.e., make sense.

Table 7 suggests three mind-sets, based upon the clustering using the coefficients for comfortable. Recall that the ingoing coefficients come from the data wherein the response (1–9 scale) was converted to 0 (ratings 1–6) or 100 (ratings 7–9.).

Table 7. Coefficients for ‘Comfortable with talking about the topic of preventing sexually transmitted disease,’ as well as Consideration Time, for three emergent mind-sets.

The three mind-sets can be really divided into one group which feels comfortable with actual conversation as shown by quotation marks (Mind-Set 1), and the remaining two groups, which are less responsive to the elements. We might be satisfied with two mind-sets, not three, one responsive to conversation (Mind-Set 1), and others. On the other hand, the differences between Mind-Set 2 (Discuss safe sex as a prelude to intimacy) and Mind-Set 3 (Safe sex as the responsibility of both partners) points to some key differences between these two groups. That difference between Mind-Sets 2 and 3 is underscored by the differences between the mind-sets in terms of Consideration Time. Mind-Set 2 (discuss safe sex) spends a lot longer than Mind-Set 3 (focuses on responsibility) when reading and rating the vignettes.

Study 2 – Recourse & Protection for the sex worker

The recent literature is replete with discussions of sex trafficking, and other offenses (Van der Meulen, et. al., 2018; Kempadoo & Doezema, 2018) [15, 16] Those stories talk about the system which creates and benefits from the sex worker, and not generally about the sex worker in terms of emotions and personal development (Bekteshi et. al., 2012; McClain & Garrity, 2011.) [17, 18].

This second study was inspired by the interests of marketing students in a graduate course in Bogota, Colombia. The students under the instruction of a8uthor Herrera, investigated the nature and magnitude of the interaction between the WHO (who is the sex worker), the DANGER (what is the danger facing a sex worker in Colombia), as they drive the response of ‘protection of’ and ‘legal recourse available to’ the sex worker. Over the past decades there has been a recognition that prostitution and allied activities constitute a profession with the workers deserving he benefits and protection due to any person who works in a job. The study approach was the same, in terms of creating the four questions, developing four answers to each question (Table 8), and then presenting the vignettes to the respondents. The 24 students themselves offered to be respondents, and so we present this second study as a methodological advancement within the emerging science of Mind Genomics.

Table 8. Sex worker – Four questions and four answers to each question.

Creating scenarios to uncover interactions among answers

The first study presented in the previous sections treated all 16 answers as independent variables, which in fact they are. In this second study, we created the study specifically to comprise a WHO (the sex worker), the danger that the person would face (DANGER), and then two different types of protection (ongoing physical safety, legal recourse, respectively.) Thus, the first two answers are really ‘set-ups’ to frame the information, that information given by protection and recourse. The objective was to identify how different ‘set-ups,’ i.e., combinations of WHO and DANGER, drive the response to protections and to recourse, respectively. The analysis below explicates the approach to study interactions, using two sets of vignettes. The first set comprises a single sex worker exposed to four different dangers. The second set comprises four sex workers, each facing the same danger.

Set 1 – sex worker constant, danger varies

Select one person to study. It does not matter which one, since we are interested in the method. For the sake of simplicity, we study one specific sex worker; an older woman who has gone bankrupt. We create five different strata, varying by the danger to which the individual (older woman) can be exposed. Each stratum thus can be defined as having one type of worker (the older woman), and one type of danger. Each individual danger and ‘no danger’ jointly define the stratum. For each stratum we run a simple model using the eight elements as predictors, the four elements describing physical protection, and the four elements describing legal recourse. Our model has no additive constant, because the rating is ‘agree/disagree.’ The additive constant makes no intuitive sense. We create this model for the rating question, again converted to binary (Top2, for agree), and then for consideration time. Table 9 presents the coefficients for agree (coefficients of 60 or higher shown in shaded cells, bold type.) Table10 presents the coefficients for consideration time (5 seconds and higher shown in shaded cell, bold type.) Both tables also show the average coefficient across all eight elements.

Table 9. Interactions between Sex Worker, Danger as stratifying variables, and legal recourse and protection as variables to be considered when disagreeing or agreeing.

The coefficients are high because two of the variables are not considered in the model. Thus, the binary transformed rating, ‘agree’ (4–5), must be allocated across eight elements, not 16 elements, even though the vignettes still comprised 2–4 elements.

What is remarkable about the table is the dramatic interaction among the ingoing facts of the case, specifically WHO the sex worker is, and the DANGER the sex worker faces, and the specific protections and recourses selected.

1. On average across the eight elements (four protection, four recourse), the level of agreement is similar close across all four Dangers for the single person (older woman)
2. Yet, the specific interactions are dramatic. For example, when the Danger is shame and disgraceful feelings inside’ the sex worker, the strongest Recourse is: Special “shaming” notices for those who hurt sex workers. In contrast, when the Danger is getting beaten up and robbed, the strongest performing else is the legal Recourse: Steep fines for those who cheat sex workers.

When we move to Consideration Time (Table 10), we see that with an older woman who has gone bankrupt, we emerge with dramatically different Consideration Times. The longest Consideration Time comes from the combination of the older woman with ‘not getting paid’ and with ‘shunned as undesirable person’, both an average of 4.6 seconds.

Table 10. Interactions between Sex Worker, Danger as stratifying variables, and legal recourse and protection as variables driving ‘Consideration Time’ when rating disagree vs agree.

There is also a noticeable interaction between the person (older woman who has gone bankrupt), the nature of the danger from the outside (not getting paid / shunned as undesirable), versus from the inside (‘’shame and disgraceful feelings.”). The outside actions / dangers generate longer Consideration Times.

The Consideration Times do not generate as clear a pattern as do the Agreement coefficients. So-called ‘objective measures’ in research may be attractive because of a belief that they are ‘tapping something real,’ but the interpretation of what they are tapping may be harder, and undoubtedly problematic.

Set 2 – danger constant, person varies

Select one danger to study. It does not matter which danger is held constant for purposes of explicating the approach. For simplicity, we focus on an emotional danger from the person’s self-image, ‘shame disgraceful feeling inside.’ As before, we create five different strata anew, varying by the sex worker. Thus, each of five strata has one danger (shame disgraceful feeling inside) and one of four sex workers, as well as the case of ‘no sex worker’.

For each of the five strata we run a simple model using the eight elements as predictors, as we did before, the four for physical protection, and the four for legal protection, respectively Our model has no additive constant. Table 11 presents the coefficients for agree (coefficients of 60 or higher shown in shaded cells, bold type.) Table 12 presents the coefficients for consideration time (5 seconds and higher shown in shaded cell, bold type.) Both tables also show the average coefficient across all eight elements.

1. On average, for a given danger, the average coefficients vary, from a high achieved by vignettes featuring the young woman who is just starting out (average coefficient = 35), to a low achieved by vignettes featuring an older woman who has gone bankrupt (average = 21).
2. When the danger is ‘shame and disgraceful feelings inside’), most of the strong performing elements are plausible, i.e., legal recourse, rather than protection. The shame and disgraceful feelings do not present danger.

Table 11. Interactions between Danger and Worker as stratifying variables, and legal recourse and protection as variables to be considered when disagreeing or agreeing.

Finally, Table 12 shows the how Consideration Time for each of the protection and recourse elements vary with the single fixed danger (shame and disgrace inside), the four different types of sex workers, and the Consideration Time. All Consideration Times are high (4.2- 4.8) except for the older woman who has gone bankrupt (2.0). For the younger sex workers, the focus is protection. For the older sex worker, the focus is legal recourse.

Table 12. Interactions between Danger and Worker as stratifying variables, and legal recourse and protection as variables affecting Consideration Time when assigning a rating of disagree agree for legal recourse and protection.

Discussion – Mind Genomics as a tool to map and to understand relationships

As suggested by the introduction, the field of sexuality, and especially the sexual behavior of intimate couples and the issues involved with sex workers have created in their wake an enormous literature. This paper does not address that literature, and especially does not attempt to answer questions raised by previous studies. Such an effort requires an encyclopedia of papers, not a single short research note. Rather, the objective here is to introduce a way to understand a topic from the inside-out, from the mind of the person, from a combination of psychological ‘thinking’ and consumer research methods.

The tradition of today’s science can be summarized by the term ‘hypothetico-deductive.’ The term means that we create a hypothesis about the nature of behavior, and then perform the requisite experiments either to falsify the hypothesis, or to not-falsify it. Not falsifying a hypothesis does not mean that the hypothesis is correct, but rather that for the time-being the hypothesis may be accepted. The focus of today’s research thus becomes increasingly narrow. The rigors of scientific research demand an almost superhuman concentration to focus the research on the specific problem. Little is left to the exploration of new ideas.

When it comes to the study of human behavior, the many aspects, the nuances, and the impossible-to-remove interactions among the variables make the hypothetico-deductive system interesting, but not particularly productive. One has pieces of information, some convincing than others. Yet, one is missing a narrative, not necessary spun from narratives and stories, but rather emerging from easy-to-do studies. The sheer difficulty of doing inexpensive, comprehensive, focused experiments with people force the researcher either to rely on questionnaires (self-reports), or to weave a story from interviews, or a limited number of experiments.

The approach presented here, Mind-Genomics, demonstrates the opportunity to create a new archival literature on people, personal relations, focusing either on specifics, on limited topics, or on a set of topics which bring into focus a bigger picture. What we see in these two studies is the relative ease of doing computer-aided experiment with messaging in order to identify how the person thinks about a topic. The experiments are short, iterative, yet generate information emerging from the structure of the experiment. The test stimuli are cognitively rich. The richness means that beyond the emergent patterns (what other studies discover) lies the responses to individually, meaningful, relevant, and possible important stimuli. The responses to the individual stimuli teach, rather than having value simply because they are part of an emergent pattern.

Acknowledgement

Attila Gere wishes to acknowledge and thank the Premium Postdoctoral Research Program of the Hungarian Academy of Sciences.

References

1. Moskowitz HR, Gofman A, (2007) Selling blue elephants: How to make great products that people want before they even know they want them. Pearson Education.
2. Box GE, Hunter WG, Hunter JS (1978) Statistics for experimenters, New York, John Wiley.
3. Boring EG (1929) A History of experimental psychology. The Century Company, New York.
4. Fisher TD, Davis CM, Yarber WL (2013) Handbook of sexuality-related measures. Routledge.
5. Montesi JL, Conner BT, Gordon EA, Fauber RL, Ki KH, et al. (2013) On the relationship among social anxiety, intimacy, sexual communication, and sexual satisfaction in young couples. Archives of Sexual Behavior 42: 81–91. [Crossref]
6. Stephenson KR, Meston CM (2010) When are sexual difficulties distressing for women? The selective protective value of intimate relationships. The Journal of Sexual Medicine 7: 3683–3694. [Crossref]
7. Harvey SM, Washburn I, Oakley L, Warren J, Sanchez D (2017) Competing priorities: Partner-specific relationship characteristics and motives for condom use ang at-risk young adults. The Journal of Sex Research 54: 665–676. [Crossref]
8. Katz BP, Fortenberry JD, Zimet GD, Blythe MJ, Orr DP (2000) Partner-specific relationship characteristics and condom use among young people with sexually transmitted diseases. Journal of Sex Research 37: 69–75.
9. Peplau LA, Rubin Z, Hill CT (1977) Sexual intimacy in dating relationships. Journal of Social Issues 33: 86–109.
10. Widman L, Welsh DP, McNulty JK, Little KC (2006) Sexual communication and contraceptive use in adolescent dating couples. Journal of Adolescent Health 39: 893–899. [Crossref]
11. Gofman A, Moskowitz H (2010) Isomorphic permuted experimental designs and their application in conjoint analysis. Journal of Sensory Studies 25: 127–145.
12. Green PE, Srinivasan V (1990) Conjoint analysis in marketing: New developments with implications for research and practice. The Journal of Marketing 54: 3–19.
13. Kahneman D (2011) Thinking fast and slow. Macmillan.
14. Dubes R, Jain AK (1980) Clustering methodologies in exploratory data analysis. Advances in Computers 19: 113–238.
15. Van der Meulen E, Durisin EM, Love V (2013) Selling sex: Experience, advocacy, and research on sex workers in canada. (eds.). UBC Press.
16. Kempadoo K, Doezema J (eds.), (2018) Global sex workers: Rights, resistance, and redefinition. Routledge.
17. Bekteshi V, Gjermeni E, Van Hook M (2012) Modern day slavery: Sex trafficking in Albania. International Journal of Sociology and Social Policy 32: 480–494.
18. McClain NM, Garrity SE (2011) Sex trafficking and the exploitation of adolescents. Journal of Obstetric, Gynecologic & Neonatal Nursing 40: 243–252. [Crossref]

Abstract

Two studies were run to understand the driving factors for intangibles, the first study dealing with attendance at an academic/business conference, the second dealing with the likelihood of voting for a candidate promoting specific values. In each study, groups of US respondents, varying in age and gender, each evaluated unique sets of 24 vignettes, comprising 16 different messages with the vignettes created by experimental design, following the Mind Genomics paradigm.Noticeable and occasionally significant age and gender differences emerged in the set of elements driving positive responses, but the group differences did not tell a coherent story. Only when the respondents were divided into mind-sets, based upon the pattern of their responses did a coherent story emerge, both for the first experiment on conferences, and the second experiment on candidates. Focusing analysis on age and gender may hinder the search for more profound difference among people, one based upon mind-set. With mind-sets, inter-individual variation in thinking about a topic becomes to become more interpretable and meaningful.

Introduction

Convincing others to do something may occupy a great deal of time. Whether the convincing is to have a child eat or behave, convincing children to study, convincing another to become romantically involved, purchase, and so forth, the normal life of a person in society is grounded in the act of persuading.  A great deal of a nation’s literature, a great deal of psychology and sociology, not to mention economics, deals with the various aspects of attempts to convince.

Convincing individuals varies by the nature of the topic. Thousands of years ago the Greeks, masters of rhetoric, realized that it was both the substance of the argument and the form of the argument which were important.  Yet, to make the topic simple, we may summarize the process by first observing the problem, second by proposing solutions, third explaining how the solutions will work, and fourth, appealing to the individual interests of the audience.  Beyond that, the rest is method and content, respectively, for those wanting to do the convincing, and by the need states and susceptibilities of those who are to be convinced. The literature of decision making is vast and cannot be dealt in a simple ‘methods paper.’ Rather, the objective of this paper to present a new, alternative approach, Mind Genomics, which emerges from experimental psychology and disciplined behavioral science [1]. Mind Genomics at its heart comprises experiments which identify ‘causation’ when messages are used to convince a decision maker.

This paper introduces a relatively new approach to study the art of ‘convincing.’ Mind Genomicshas been used in the form of conjoint measurement to understand what messages one to use in order to convince. Mind Genomics focuses in the application of conjoint measurement to the decisions, and the decision rules, of the everyday experience [2, 3]. The paper use Mind Genomics to compare two types of ‘convincing,’ one to ‘sell a professional conference,’ the other to sell a ‘candidate’ for an election. Historically, conjoint measurement has been used to identify the relative importance of factors in a considered decision, such as insurance selection, or health benefits. The respondent is provided with pairs of stimuli, whose composition is known for each alternative in the pair. The respondent must select one of the two stimuli. The pattern of selections can be processed by an accepted computation scheme to generate the ‘utilities’ or ‘impact’ of each element of the set of elements.  The methods can be tedious, but have found use in large-scale, expensive, but critical decisions, such as the choice of medical and so forth [4].

Mind Genomics – foundations and processes

The scientific method teaches that variables should be separated and studied in ‘splendid isolation.’  For most variables the isolation works, but not necessarily in messages. Typically, messages come to people in combinations, with the different messages complementing each other, suppressing each other, or even synergizing with each other so that the whole, the combination, often is far more impactful than would be by the sum of individual impacts.   In normal life, we do not encounter single messages, except perhaps for those signaling ‘emergency’ or ‘danger.’  When we are exposed to single messages in a research context or in public opinion polling, we focus unduly on the topic, and give biased answers by trying to ‘guess’ the correct answer, or answer in the way that the researcher expects.

The Mind Genomics approach differs dramatically from the conventional one-at-a-time approach espoused by traditional research. The researcher presents the respondent with different combinations of messages, instructing the respondent to ‘vote’ for the combination.  The approach seems a bit odd, based upon the traditional method of ‘one at a time.’   When compared to conventional approaches, we can say that that the Mind Genomics approach is more Socratic, holistic, yet systematic, and oriented towards creating combinatorial models for persuasion and communication:

Holistic: The test stimuli are combinations of messages, not single messages alone. The holistic approach simulates what we see in real, daily life. For the most part, we deal with mixtures of stimuli coming at us all the time. When we talk, walk, drive, read, eat, and so forth, we do not pay attention to one variable, except perhaps for a very short time to examine it more closely. We live in the moment, the moment comprising a kaleidoscope of changing combinations.

Socratic: The Socratic method comprises a question and answer dialogue. Carried out effectively, the dialogue reveals the underlying structure of the topic. Socratic dialogue is not the pure science to which we are accustomed (nomothetic), for it is intensely individual rather than general (idiographic).  Yet, for topics studying the act of judgment itself, the Socratic method can generate the necessary test material by which the researcher uncovers some individual ‘rules’ of decision.  

Systematic: The abovementioned answers to questions are ‘elements,’ namely simple, easy to comprehend statements, almost factoids. These elements or answers to questions are combined by the discipline of experimental design into small, easy to read combinations. [2]. Mind Genomics is based upon the belief that when making a decision the respondent ‘grazes’ for information, rather than ingests, chews, and digests, respectively. The Mind Genomics research process recognizes grazing, and is designed to be fast and minimally intellectual, metaphorically similar to grazing at a superficial level. No effort is made to combine the different elements into a flowing paragraph.

Models:  Mind Genomics develops mathematical models showing how each element or answer ‘drives’ the response. The response may be a rating (would not attend to would attend, would vote for candidate to would vote for candidate), the selection of an emotion or feeling from a set of several alternatives (happy, sad, curious, excited, etc.), the selection of a price, the selection of an end use, etc.  The models show the linkage between the different elements and the rating.

Steps in the Process – Creation of raw materials through a Socratic process: The researcher selects a topic. The researcher then asks four questions which ‘tell a story.’ Asking the four questions can be hard, and forces creative and critical thinking. Most people are not educated to ask questions in a systematic way, in contrast to a reporter or writer who does so by habit when creating a coherent story. Once the four questions are asked, it become very easy to provide four simple answers to each question. The concern is often raised as to whether the questions are truly the ‘correct questions’ to ask, and in turn, whether the four answers to each question suffice, as well as whether or not they are the proper answer. It takes a while to disabuse the novice of the reality that there are no correct questions nor answers, but rather report ‘blocks’ at this stage, because they either cannot think of questions, and freeze up, or they take the instruction literally, and cannot ‘tell a story.’  With practice, however, they realize that the narrative can tell a story, but not the polished story to which they have been accustomed.

Steps in the Process –Creating the vignettes using experimental design to specify combinations: Mind Genomics traces back to the evaluation of combinations of messages, with the combinations prescribed by experimental design, or metaphorically by a set of recipes which combine the individual messages into known combinations. For the Mind Genomics studies run here, with the four questions and four answers per question, each respondent evaluated a unique set of 24 vignettes or combinations. Each vignette comprised 2, 3 or 4 answers, at most one answer from each question. The answers are coded 0 when absent from a vignette, and 1 when presented in a vignette. The experimental design ensures that each respondent evaluates the 16 answers in different combinations, and that the answers or elements are statistically independent of each other.  The statistical independence will allow the researcher to create individual-level equations, one for each respondent, relating the presence/absence of the 16 elements either to the binary transformed rating (0/100) or to the consideration time (CT).

Table 1 shows the schematic for eight vignettes from Respondent #1. The respondent evaluated the eight vignettes in sequence. The combination is defined by the experimental design. A ‘0’ represents the fact that the element was absent from the vignette. A ‘1’ represents the fact that the element was present in the vignette. The respondent rating on a 9-point scale was recorded, along with CT, consideration time, the number of seconds elapsing between the presentation of the vignette on the screen and the rating. The CT is recorded to the nearest tenth of a second.

Table 1. Eight vignettes from the conference study.

Study 1 – ‘Selling a conference’

Study 1 focused on how one ‘sells’ or at least advertises a conference about the evolving area of data analytics, when the audience comprises random people.

Conferences are important as a key venue for academics. It is important to market conferences, to communicate what the conference provides for the attendee [5]. Beyond the conference as an academic product to be marketed, the conference is a topic of interested in itself, The conference is a contained environment where relevant interpersonal behaviors are strongly demonstrated. Researchers have investigated conferences from the outside, from the benefits of the respondent [6].  One of the key benefits is making connections [7].

Conferences themselves are a venue for sociological and psychological research. The conference is a specific venue, offering the chance to observe a variety of different behaviors. For example, one research avenue is to study behavior at conferences in terms of the behaviors of males versus females. An anthropological approach might look at the conference as a venue wherein certain attitudes are manifest in behaviors, in the so-called ‘lived experience’ [8]. There are a variety of dimensions to conferences, dimensions which can serve as the foundation of research to understand the mind and motivations of those who attend the conference. These dimensions range from the conference as a venue of information to be disseminated and learned, involving different groups, such as academics versus practitioners, respectively [9, 10], as well as networking vs knowledge [11–13].  Then there is the ever-present dimension of the conference as a venue to introduce students, and to let the students interact with senior professionals [14, 15].

Table 2 presents the four questions and the 16 answers.   The actual questions were recorded, along with the answers, and then slightly edited to ensure proper English.  Note that the answers are simple, with three dots (…) replacing some connectives, to make reading easier.It is important to note that the elements or messages, i.e., the answers to the questions, are simple. They are descriptive, and generally feature a single idea. They will be combined in a simple way, as a set of phrases, centered, on the screen, one atop the other, with no effort to connect them. Although it seems quite ‘stark’ and unreal to have a paragraph or concept comprise a block of phrases with no connectives, the reality is that this structure makes the task easy for the respondent, who really ‘grazes’ for information, rather than reading the entire concept in depth.  When the same task is implemented with paragraphs created in better English style, as grammatically correct paragraphs, the task becomes onerous and boring. Mind Genomics studies are generally executed on the Internet, often with respondents recruited by a panel provider specializing in the process of panel creation and deployment for research studies. For this study, the researchers entered the questions, answers, and rating scale in a program designed to run these studies. The program, BimiLeap (short for Big Mind Learning Application), mixes the answers according to an experimental design, presenting 24 combinations of elements to each respondent who participates. The entire process takes 3–5 minutes for a respondent.

Table 2. Conference, list of elements.

The respondents were 50 Americans, 18 years or older. The respondents were provided by a panel company specializing in providing anonymous respondents for these types of studies (Luc.id, Inc.)   The actual elements were created at a conference of professors and students. Each respondent evaluate a unique set of 24 vignettes, created by experimental design. The experimental design ensures that all 16 elements are statistically independent of each other, permitting the use of OLS, ordinary least-squares linear regression, to relate the presence or absence of the element to the rating.  An algorithm permuted or modified the specific combinations, maintaining the underlying experimental design, but ensuring that the specific combinations different from one respondent to another [16]. The research benefit of permutation is to generate a more representative and thus a more valid model because the researchtests more of the potential mixtures of elements. That is, rather than reducing variability by testing the same limited set of combinationsmany times, and suppressing variability by averaging it out, Mind Genomics deals with variability by covering a wider array of potential test combination. Mind Genomics is statistically powerful, and conservative by design, measuring many stimuli rather than imputing from a less noisy, far less representative sample of possible vignettes.

Ratings, transformations, and averages: The original ratings were assigned on a anchored 9-point scale (1=Do not choose … 9=Choose.) The practice of Mind Genomics is to divide the rating scale into two parts. We did this division two times. The first time was ‘Choose to attend’ (1–6 transformed to 0 to denote not choose to attend; 7–9 transformed to 100 to denote choose to attend).  The second time was ‘Reject’, (1–3 transformed to 100 to denote reject; 4–9 transformed to 0 to denote not reject.) The program further recorded the ‘consideration time’ (CT), operationally defined as the number of seconds between the appearance of the vignette on the screen and the respondent’s rating of the vignette on the 9-point scale.

Table 2 shows the mean ratings for the three dependent variables by key subgroups. These subgroups are total, gender, age, and the two mind-sets or clusters of respondents, with respondents in the same cluster showing similar patterns of response coefficients (see below).  It is clear from Table 2 that the subgroups differ from each other in their ratings.

Table 2A. Means of the dependent variables (Accept, Reject, Consideration Time) for key subgroups.

Males are much more likely to say ‘I will attend’ than are females (57 versus 34, meaning that 57% of the responses of males to the vignettes are 7–9, whereas only 34% of the responses of females are 7–9).

Older respondents are more likely to say I will attend than do younger respondents (51 vs 43)

Dividing the respondents into groups based upon the pattern of how elements drive ‘attend’ (i.e., mind-sets) suggest no difference in frequency of responding ‘attend,’ but as we will see, strong differences in the elements which drive them to say ‘attend.’

The differences in ‘reject’ can be interpreted in the same way.

When we measure the consideration time, we see that women take longer to respond, that older take longer to respond, and that Mind-Set 2A takes longer to respond

Thus far, all that the data has revealed is the average rating and the response time. Those measures provide some idea of the differences between groups. Furthermore, Mind Genomics provides far deeper information for the simple reason that the elements themselves are cognitively rich, having deep meaning.  It is not simply the stimulus, but the fact that the stimulus can be understand in and of itself.

Modeling to show causality: The next step, this time for deeper understanding, creates a simple model or equation, relating the presence/absence of the 16 elements to the binary ratings. The regression modeling, OLS regression (ordinary least-squares) works with the full data set of respondents in the subgroup. The output is a simple linear expression relating the presence/absence of the 16 elements to the rating, after the binary transformation.  The regression model lacks an additive constant, for the simple reason that in the absence of elements the respondent is not likely to either accept or reject the conference.  This is called ‘regression through the origin.’

We express the equation as: Binary transformed rating = k₁(A1) + k₂(A2) …k₁₆(D4)

Deep learning – total panel, age and gender:  It is from the coefficients and their commonality that we learn the most about what drives ‘accept’ the conference, i.e., expect to attend.  Table 3 shows the strong performing elements presented in shaded cells, and bold font.  Strong performing is based upon the fact that in previous studies these coefficients are both statistically significant (from inferential statistics), and meaningful in terms of ‘real-word’ situations. When we look at the commonality of strong performing elements across elements and subgroups, we see different sets of strong-forming elements.  If we had to hazard a guess about which elements are consistently strong performers, we would say that the answers to Question D (What is interesting about the conference beyond the topic?).  That finding may be correct at the superficial level, but it leaves out the world of people who attend conferences to become stronger in their profession.

Table 3. Coefficients of the models relating the presence/absence of the elements to the ‘attend’ rating, after recoding

Beyond the discovery of ever-present individual differences, variation in the criteria of judgment, is the postulation by Mind Genomics that for every topic of experience, no matter how ‘micro’, there are a limited number of different groups, mind-sets, metaphorically alleles or variations of genes. These mind genomes do not need to covary with the typical groupings to which we have become accustomed, e.g., age, gender, and nor even behavior and attitude, such as attending conferences.

The comparison of Mind Genomes to the science of biological genomics is, to stress the point, metaphorical. In the biological science of Genomics, the belief is that there are actual alleles that can be manipulated and reinserted into cells to change their behavior. There is the belief that these alleles have actual physical reality. In the world of Mind Genomics, the mental alleles are hypothetical constructs, patterns of decision criteria which emerge from the statistical method of clustering, a procedure in the mathematics of numerical analysis. That is, there is no belief in the physical reality of the mind genome, the mental allele, but just a convenient, and sensible group of ideas which float together.

These mind genomes or mind-sets emerge from the pattern of coefficients for the different elements, with the pattern uncovered by experimentation (our respondent study with the 50 respondents), the creation of individual-level models (made possible by the experimental design), and then the clustering individuals by the pattern of their coefficients (application of clustering, a method in numerical analysis.)

When we follow the procedure of experimentation, modeling, clustering, afterwards extracting meaningful sets of ideas or clusters, we end up with three different groups.Clustering simply places the objects (here respondents) into a set of complementary, non-overlapping groups, using mathematical criteria. The objective to minimize the number of clusters (parsimony), as well as ensure that each cluster or mind-set makes sense (interpretability).  Table 4 shows the performance of all 16 elements by total, and by each of the two emergent mind-sets, i.e., emergent clusters of respondents based on the pattern of coefficients. It is clear from Table 4 that separating the mind-sets allows the strong performing elements to do far better than they do when the data from all 50 the respondents are combined to create the one group, total panel. Mind-set segmentation through clusteringremoves much of the suppression of element performance attributable to the opposing patterns of responses of different mind-sets to the same element.  The countervailing forces emerge, and can be separated from each other, placed by the researcher into the different mind-sets (clusters), with the result being radically different patterns of coefficients released by the suppressing, mutually cancelling effect by the opposite mind-set.

Table 4. Performance of elements driving Choosing a Conference. Data based on the total panel and the two mind-sets.

Engagement – Measurement of consideration time (CT) for conference elements: In order to identify the existence of mental processing of stimulus input, such as our elements, experimental psychologists introduced the notion of reaction time, later called response time, and now in this stage of Mind Genomics called ‘consideration time.’ The underlying notion is that longer consideration times signal that more complicated mental processing is occurring.The original measures of reaction time were done when the respondent was instructed to observe a test stimulus (see, feel, hear, taste, smell), and then report when the respondent could detect the stimulus (i.e. the stimulus was present), or report when the respondent could recognize the nature of the stimulus. The right-most column of Table 2 above presents the average consideration times (CT) for the 24 vignettes rated by each respondent in the relevant subgroups. Table 2 suggests that, on average, the time to read and rate a vignette is approximately 3.5 seconds.  The younger respondents read and rate the vignettes far more quickly than do the older respondents (2.9 seconds vs 3.8 seconds). Males read and rate the vignettes far more quickly than do females (3.0 seconds vs 4.2 seconds). Finally, Mind-Set 2B (Conferences for professional development) reads and rates the vignettes far more quickly than does Mind-Set 2A (Conferences for fun), specifically 2.7 seconds versus 4.0

Knowing the consideration time tells us something about the general speed of reading and decision- making but does not tell us anything about the consideration time given to the individual elements. That consideration time is a measure of engagement of the respondent with the message. The engagement may be short or long for a variety of reasons, such as length and complexity of the message, basic ‘stickiness’ of the message to keep the respondent focused, and so forth. The respondent cannot tell the researcher which particular element in a vignette ‘engages’ attention, but through experimental design and modeling, along with a measure of response time to the entire vignette, the researcher can estimate the number of seconds that is most likely taken up by the specific element, such as a particularly provocative phrase. Systematic design reveals just what just what phrases are ‘sticky’, when they are ‘sticky,’ and with whom.

The strategy is the same as used to develop the models relating the presence/absence of the 16 elements to the rating. The analysis uses OLS (ordinary least-squares) regression to relate the presence/absence of the elements to the consideration time, measured to the nearest 10^thof a second.  The equation is the same, except for the dependent variable: Consideration Time (Time interval from presentation to rating) = k₁(A1) + k₂(A2) …k₁₆(D4)

Table 5 suggests a different story for the commonality among the longest consideration times for the group:

Total panel –serious aspects such as workshops and mentors

Younger –mentors and growth

Older respondents – learning new technology easily and with fun

Males – workshops and mentors

Females – learn new technology, learn at the start of the career

Mind-Set 2A (Conferences are for fun) – learning new skills, then many of the professional growth elements

Mind-Set 2 B (Conferences are for professional development) – no elements show unusually long engagement. Equal attention is paid to all elements

Table 5. Consideration time for all elements by total panel and key subgroups (conference).Element coefficients of 1.2 seconds or higher are shown in shaded cells.

Study 2 – ‘Selling a political candidate’

If the topic of conferences is of interest to academics and to those sponsoring conferences, in contrast, the topic of political candidates and their messaging is of interest to virtually everyone, or almost everyone, especially in elections where two or more sides, radically opposite, vie for power.   Furthermore, election and the messaging of the candidates must address the many different dimensions on which a candidate can appeal to her or his audience, and the many different facets, the granularity of each dimension, that must somehow be considered

More than 80 years ago, the mind of the voter was already of interest [17], but of course one could go back centuries to Machiavelli, to Aristotle, and to Plato for even older points of view. These philosophers talked a great deal about citizens and their leaders. Many of their points, including appeal to emotion, hold today.  One need only read Machiavelli’s ‘Prince,’ Aristotle’s ‘Politics’ or Plato’s ‘Dialogue’ to see the politics of today presented by the eminent thinkers of the past. Today’s world works with tools taken from marketing, attempting to persuade people to vote in the same way one might persuade people to buy toothpaste [18].  There is a great deal of effort put in by consultants, polling organizations, and so forth to identify messages which at once most strongly resonate with the electorate, as well as being appropriate, realistic, and believable. Despite the best efforts of marketers to provide honest data, perhaps somewhat copy-edited (‘massaged’), today’s political messaging is believed a lot less than was the case years and decades before [19].

Marketing theory has also entered political messaging and polling. The notion of inward vs outward orientation in the mind of a consumer has been applied to an Australian election, revealing the application of this construct to election messaging [20].This inward versus outward orientation more clearly focuses on what affects the voter, and moves beyond the more tradition of description of one’s behavior, such as mudslinging, defined both as allegations about the candidate’s family, but also references to an opponent’s voting record, broken campaign promises, rumors on health and financial dealings, and the use of harsh language.

More recent approaches to studying political communication focus on how to legitimize one’s point of view, and not just to convince the voter based upon one or two key points. Legitimizing one’s point of view is akin to building one’s brand, again recognizing the mind of marketing, as it enters the political arena [21] discussed the political communication as exemplified by George Bush and by Barack Obama, when they had already won the election, and were trying to convince the electorate about their efforts of the war on terror, in 2007 and 2009. In Reye’s words, ‘strategies of legitimization can be used individually or in combination with others and justify social practices through: (1) emotions (particularly fear), (2) a hypothetical future, (3) rationality, (4) voices of expertise and (5) altruism.’By 2010, the marketing concepthad entered the world of communication. The five strategies, or motivations for message, just above, would be quite familiar to today’s marketer. The final aspect making a study of political messaging interesting is the increasing importance of social media on the political process. Research published almost a decade ago suggest that in the early years of social media the interplay of social media and political viewpoint was not particularly strong [22]. Kim’s words of a decade ago can be contrasted with the emergence of political messaging in the form both of real news and of fake news.  It is worthwhile quoting Kim’s now-passe language, quite important in 2011, and probably based upon research conducted the year or two before. It would hard to substantiate Kim’s words today, as of this writing.

The increasing popularity of social network sites (SNSs) has raised questions about the role of social network media in the democratic process. This study explores how use of SNSs influences individuals’ exposure to political difference. The findings show a positive and significant relationship between SNSs and exposure to challenging viewpoints, supporting the idea that SNSs contribute to individuals’ exposure to cross-cutting political points of view. Partisanship was not found to interact with SNS use, suggesting that SNSs contribute to expanding exposure to dissimilar political views across individuals’ partisanship. Online political messaging also has a direct effect on exposure to dissimilar viewpoints, and it mediates the association between SNSs and exposure to cross-cutting political views.  (Bold added for emphasis)

Specifics of the candidate study: The principles underlying the Mind Genomics studies remain the same, no matter what the topic.  The second study, done around the same time concerned a political candidate, of an unnamed political party. The respondents were US adults, recruited by the same company as the respondents in Study 1 on ‘selling a conference.’

The key differences in the two studies were the topic, the elements (Table 6), and the use of a 5-point scale, rather than a 9-point scale for the scale. For the rating of ‘win’, the 5-point scale was transformed to the binary values of 0 (ratings 1–3), and 100 (ratings 4–5). For the rating of ‘lose,’ the 50point scale was transformed the binary value of 100 (ratings 1–2), and 0 (ratings 3–5). All modeling was done using the binary scale, not the original scale.

Table 6. Candidate – List of elements

Table 7 give a sense of the response patterns for the different vignettes, across the different groups. What is most interesting is that when the topic is political, something serious and relevant to the respondents, the consideration time is a second longer than the consideration time for the conference (3.5 seconds for the conference, 4.4 seconds for the candidate.) The experimental design is the same, the elements are approximately of the same size, but the respondents spend more time reading.  This pattern, longer consideration times for important topics, has continued to emerge again and again in experiments by author Moskowitz (unpublished data)

Table 7. Means of the dependent variables (Accept, Reject, Consideration Time) for key subgroups

Table 8 shows the results for Total, Age and Gender, respectively.

Table 8. Performance of elements driving Choosing a Conference. Data based on the total panel, age and gender, respectively.

The key drivers for winning are the personal characteristics of the candidate, especially the care about the people and being a role model.

He/she concerned about people well-being

He/she has a vision to develop the country

He/she is going to be the people’s voice in government

He/she concerns about people well-being

He/she has a vision to develop the country

He/she is going to be the people’s voice in government

Some key differences emerge, mostly in terms of degree

Men are concerned about the situation in the country

Women are concerned about the candidate ‘being involved’

Younger respondents do not like a boastful, dominating person who tells others what to do. In contrast, older respondents don’t care.  This is a subtle but an importance difference between different age cohorts, representing an emerging sensitivity to ‘authenticity’

Applying the clustering approach to the 50 coefficients generates two clearly different, and interpretable mind-sets, shown in Table 9. Mind-Set 1 responds to the candidate as a leader in the unstable times. Mind-Set 2 responds to the candidate as a nation builder.

Table 9. Performance of elements driving voting for a candidate. Data based on the total panel and the two mind-sets.

We finish the detailed analyses of the by looking at the consideration time attributable to each element. Recall from the previous analysis of conferences that the form of the model for consideration time comprised a simple linear model, without an additive constant.  The experimental design for this study of a candidate is precisely the same as the experimental design for the study of a conference, namely 24 vignettes comprising 2–4 elements per vignette. When we deconstruct the contribution of each element to consideration time (Table 10) we find that virtually all but three of the consideration times are 1.0 second or longer, several twice as long at 2.0 and 2.1 seconds. Thus, the topic itself, is a major driver of consideration time, a subject to be explored more fully.  There is no clear pattern of covariation between the response time and who the respondent is, except that the younger respondents show somewhat shorter consideration times, very much shorter for descriptions of the candidate’s personal behavior (e.g., C1 and C4.)

Table 10. Consideration time for all elements by total panel and key subgroups (conference)

Who belongs to these mind-sets, and how to discover them

The mind-sets for both the conference and the candidate make sense. Yet, a standard cross tabulation of membership in the mind-set versus the standard classifications of gender and age suggest that the mind-sets do not divide simply across easy-to-measure subgroups based upon who a person IS. Table 11 shows the cross tabulation of mind-set membership versus age and gender. There is no clear relation. Indeed from author Moskowitz’s experience, except for the most obvious of cases (e.g., age versus concern with problem of dying), the relation between the way a person thinks and who the person IS appears to be tenuous at best.  Furthermore, even asking a person about general thoughts regarding a topic does not suffice to place a person into a mind-set

Table 11. Two-way table showing the relation between membership in a mind-seg (column) and both age and gender, respectively.

A new way be developed to probe membership in a group defined by the specifics or granular aspects of the way a person thinks about a topic. Conferences and candidates are large subjects. The mind-sets which emerge are limited to the topic revolving around questions and answers investigated in the Mind Genomics study. It may well be that the easiest way to discover the membership of a person in a mind-set segment is to accept the fact that the mind-set segment is granular at best. That ‘best’ may be to assign a new person to the granular-based mind-set uncovered in the Mind Genomics experiment. Authors Gere and Moskowitz have created an algorithm based on the separation of the mind-sets across the 16 elements. Using a Monte-Carlo simulation, they identified a set of six elements, the pattern of binary answers to which, suggest membership in one mind-set or the other.   Figure 1 shows the PVI, the personal viewpoint identifier, emerging from this exercise.

Figure 1. The PVI (personal viewpoint identifier), comprising six questions for each topic. The pattern of answers assigns a respondent to one of the two mind-sets.

Discussion

The typical study of a topic involves a few stimuli, rarely varied systematically, but evaluated by many people, respondents in the world of public opinion polling and consumer research, subjects or observers in the world of psychology.  The objective of these studies is typically to confirm a hypothesis. The use of large numbers of respondents has become sacrosanct in many areas of science, for the simple reason that with these large number of respondents the sampling distribution of ratings is more precise, with smaller standard errors. Mind Genomics as presented here provides the researcher with a different strategy. Rather than being developed within the constraints and world-view of the traditional world of the ‘hypothetico-deductive,’ Mind Genomics approaches the topic by exploring a wide, albeit feasible, range of alternative aspects, evaluated by the respondent in formats, vignettes, simulating a more typical way that nature presents information to people, namely in the form of  mixtures.  The systematic variation of the composition of these mixtures by experimental design allow the researcher to pick out the operative variables to which the respondent attends.

As we review the process of the two studies, we come upon the following key factors which differentiate Mind Genomics studies from other studies of the same topic:

Mind Genomics studies focus on the mind of the respondent, weaving a story, but without having the respondent elaborate and tell the story. Qualitative research focuses on the mind of the respondent as well but requires that the respondent participate in a dialog. The experienced researcher, like an experienced therapist, may pull out underlying motives, thoughts, defenses, and biases, but the researcher should be experienced must shunt aside presuppositions. In contrast, Mind Genomics, attempting the same outcome, works with responses to cognitively rich expressions, the elements, not chosen by the respondent, but by the researcher. Mind Genomics studies can be executed more rapidly, more generally, and more cost-effectively.  What Mind Genomics lacks, however, is the skilled interpretation, when such skill exists. Mind Genomics studies can be likened to the MRI of the Mind.  Each individual Mind Genomics study creates 24 vignettes for each respondent, with the vignettes differing from respondent to respondent. Thus, in one Mind Genomics study with 30 respondents, we deal with 720 different snapshots of the same problem. One need not know the ‘correct’ or best combinations to test. Mind Genomics studies create, metaphorically, a realistic ‘picture’ of the topic from which one can discover new things or reaffirm hypotheses and conjectures which seem simplistic after the fact, but hard to confirm ahead of time.

We have illustrated two different studies and show slightly different dynamics of each. The speed and ease of a Mind Genomics study makes it possible to execute one or two studies a day and create a rich library of knowledge about any topic involving the decision of a respondent when faced with various pieces of information. A science of such decision rules, appropriate indeed and archives, may constitute a new direction for sciences of the mind, and of society.

Acknowledgement 

Attila Gere wishes to acknowledge and thank the Premium Postdoctoral Research Program of the Hungarian Academy of Sciences.

References

1. Box GEP, Hunter WP, Hunter JS (1978) Statistics for experimenters, New York, John Wiley.
2. Moskowitz HR (2012) ‘Mind genomics’: The experimental, inductive science of the ordinary, & its application to aspects of food & feeding. Physiology & Behavior 107: 606–613. [Crossref]
3. Moskowitz HR, Gofman A, Beckley J, Ashman H (2006) Founding a new science: Mind Genomics. Journal of Sensory Studies 21: 266–307.
4. Ryan M, Farrar S (2000) Using conjoint analysis to elicit preferences for health care. Bmj 320: 1530–1533. [Crossref]
5. Stevens R, Bressler M, Silver L (2016) Challenges in marketing academic conferences: a pilot study. Services Marketing Quarterly 37: 200–207.
6. Parker BJ (2007) What makes a professional conference worth attending? Strategic Finance 88: 13.
7. Cherrstrom CA (2012) Making connections: Attending professional conferences. Adult Learning 23: 148–152.
8. Mair J, Frew E (2018) Academic conferences: a female duo-ethnography. Current issues in Tourism 21: 2152–2172.
9. November P (2004) Seven reasons why marketing practitioners should ignore marketing academic research. Australasian Marketing Journal 12: 39–50.
10. Nyilasy G, Reid LN (2007) The academician–practitioner gap in advertising. International Journal of Advertising 26: 425–445.
11. Anderson L, Anderson T (2010) Online professional development conferences: An effective, economical & eco-friendly option. Canadian Journal of Learning & Technology 11: 35.
12. Rogers T, Davidson R (2015) Marketing destinations & venues for conferences, conventions & business events. Routledge.
13. Hughes T, Bence D, Grisoni L, O’regan N, Wornham D (2011) Scholarship that matters: Academic–practitioner engagement in business & management. Academy of Management Learning & Education 10: 40–57.
14. Gardner SK, Barnes BJ (2007) Graduate student involvement: Socialization for the professional role. Journal of College Student Development 48: 369–387.
15. Mata H, Latham TP, Ransome Y (2010) Benefits of professional organization membership & participation in national conferences: Considerations for students & new professionals. Health Promotion practice 11: 450–453.
16. Gofman A, Moskowitz H (2010) Isomorphic permuted experimental designs & their application in conjoint analysis. Journal of Sensory Studies 25: 127–145.
17. Herring EP (1938) How does the voter make up his mind? Public Opinion Quarterly 2: 24–35.
18. Harris P, Lock A (2010) “Mind the gap”: the rise of political marketing & a perspective on its future agenda. European Journal of Marketing 44: 297–307.
19. Brants K, Voltmer K (2011) Introduction: Mediatization & de-centralization of political communication. In: Political Communication in Postmodern Democracy, 1–16 Palgrave Macmillan, London.
20. O’Cass A (2001) The internal-external marketing orientation of a political party: social implications of political party marketing orientation. Journal of Public Affairs: An International Journal 1: 136–152.
21. Reyes A (2011) Strategies of legitimization in political discourse: From words to actions. Discourse & Society 22: 781–807.
22. Kim Y (2011) The contribution of social network sites to exposure to political difference: The relationships among SNSs, online political messaging, and exposure to cross-cutting perspectives. Computers in Human Behavior 27: 971–977.
23. Stewart CJ (1975) Voter perception of mud-slinging in political communication. Communication Studies 26: 279–286.