Influencing the Emotions of the Decision Maker by Adding a Decoy- or Superior Option to
the Choice Set, After the Decision Has Been Made
Milou Ottink 10676651
University of Amsterdam
Faculty of Economics and Business
MSc Economics, specialization in Behavioral Economics & Game Theory
Supervisor: prof. dr. J.H. Sonnemans Second examiner:
Abstract
A behavioral experiment looking at the relation between the adding of a decoy- or superior
option to a choice set after a decision has been made and the emotional response of the
decision maker has not been done before. However, the potentially large impact of emotions
on decision-making has been recognized (Rick & Loewenstein, 2008). An online survey
consisting of 80 participants was used to investigate the happiness of a decision maker after
the adding of a decoy- or superior option to the previously chosen- or non-chosen option of
the decision maker. Happiness was analyzed by looking at the different levels of satisfaction,
disappointment and regret experienced by the decision maker. The results showed that, as
predicted, the adding of decoy options raised satisfaction and decreased disappointment and
regret of the decision maker, while the adding of superior options decreased satisfaction and
increased disappointment and regret. The adding of a decoy option thus seemed to increase
happiness and a superior option decreased happiness. Furthermore, there was no difference
between the adding of a decoy to the chosen- or non-chosen option. This is in contrast to the
adding of a superior option, where the attributes seemed more important. After the adding of a
superior option to the chosen option instead of the non-chosen option, the results were
significantly stronger in the predicted direction. In addition, the participants seemed to change
their preferences towards the dimension that was most important in the option they choose
and away from the dimension that was least important. These results support the idea that the
emotional responses after the adding of certain product to a market can be predicted and this
can help in understanding the underlying causes of decision-making.
Table of contents Abstract……… 2 1. Introduction………. 4 2. Literature review………. 6 2.1 Decoy effect………... 6 2.2 Choice models………. 7 2.3 Non-dominated alternative……….. 8 2.4 Explanations……… 8
2.5 Applicability decoy effect………... 11
2.6 New insights……… 12
2.7 Superior options………... 14
2.8 Emotions in decision-making……….. 16
2.9 Valuation………. 17
2.10 Different types of emotions in decision making……… 18
3. Hypotheses……….. 20 4. Design……….. 22 4.1 Participants……….. 22 4.2 Measures……….. 23 4.2.1 Satisfaction……….. 23 4.2.2 Disappointment……… 24 4.2.3 Regret……….. 24
4.2.4 Comparison third option……….. 24
4.2.5 Change in preferences………. 25
4.3 Procedure………. 25
4.3.1 Overview components of survey………. 25
4.3.2 Type of answer possibilities……… 27
4.3.3 Product choice………. 27
4.3.4 Location and dimensions of decoy- and superior options... 27
4.3.5 Division of third option………... 28
4.3.6 Hypothetical incentives………... 29 5. Results………. 29 5.1 Hypothesis 1……… 33 5.2 Hypothesis 2……… 34 5.3 Hypothesis 3……… 35 5.4 Hypothesis 4……… 36 5.5 Hypothesis 5……… 37
5.6 Familiarity and frequency of use influence………. 39
5.7 Comparison of third option to made decision………. 40
6. Discussion……… 41
References………. 45
Appendix A: Example of survey………... 50
1. Introduction
The influence of emotions on economic behavior has become an increasing field of
interest since up rise of behavioral economics (Rick & Loewenstein, 2004). It has been
recognized that many rational choice models do not reflect reality and that understanding the
underlying causes of decision-making can be of great interest (Rottenstreich & Shu, 2004).
Within the field of marketing the influence of adding a new product to the market on the
choice share of the other products it is an important subject. It has been a long established fact
that the adding of a new option with specific characteristics can cause the decoy effect to
occur (Huber, Payne & Puto, 1982). The decoy effect is the phenomenon that when an
asymmetrically dominated alternative is added to a choice set, the preferences of the decision
maker will change in favor of the item that dominates this added decoy option (Huber et al.,
1982). For example, Huber and colleagues (1982) compare two restaurants that are
supposedly equally desirable because of differences in driving time and star rating. One
restaurant has five-star rating and a driving time of 25 minutes, while the other restaurant has
a three-star rating and a driving time of 5 minutes. When a decoy restaurant is then added to
the choice set, which is relatively worse so undesirable, the two restaurants of the original
choice set are not equally desirable anymore. If a restaurant with a four-star rating and 35
minutes driving time is added, the preferences move toward the restaurant with the five-star
rating and 25 minutes driving time. However, when a restaurant with a two-star rating and 15
minutes driving time is added, there is an increase in preferences for the restaurant with a
three-star rating and 5 minutes driving time. Much research has been done on the decoy
effect. Several explanations for the decoy effect have been proposed and it has been tested on
multiple different products and choice tasks (e.g. Huber et al., 1982; Huber & Puto, 1983;
Ratneshwar, Shocker & Stewart, 1987; Simonson, 1989, Simonson & Tversky 1992, Tversky
emotions on decision-making is addressed (Hsee & Rottenstreich, 2004; Kahneman &
Knetsch, 1992; Damasio, 1994; Pham, 1998; Baron, 1992; Frederick, 2002; Finucane,
Alhakami, Slovic & Johnson, 2000; Schwarz & Clore, 1996) and the malleability and
importance of emotions in the decision-making process has been recognized (Rottenstreich &
Shu, 2004). In this study emotions and decision-making are looked upon in a different
direction. Not the way in which emotions influence decisions is investigated, but the emotions
that occur when a decision has already been made. This is done by analyzing the effect of
adding a third option on the decision maker’s emotions. With the current literature available
about the decoy effect and the literature about emotions in decision-making, the relationship
between these two seems like a logical next step. Rubaltelli and Agnoli (2011) already looked
at this relationship by incorporating the decoy effect in their study about the emotional cost of
charitable donations. This study is more closely related to emotions and the decoy effect.
However, it uses the decoy effect only to counteract between two conflicting goals, namely,
moral institutions and costs. It thus does not look at what adding a decoy option does to the
emotions of people in itself. The only study that seems to relate the adding of a decoy option
to emotional responses are Hedgcock and Rao (2009). In this study a fMRI experiment is used
which shows that the activation of brain areas related to the experience of negative feelings
decreased after the addition of a third dominated alternative. This suggests that the addition of
a decoy option leads to a decrease of negative emotions. Possibly it could thus make people
happier. Furthermore, the adding of a superior- instead of a decoy option will also be
addressed in this study. The adding of an option that is superior to the choice set and the
former made decision could decrease happiness and increase disappointment and regret. A
behavioral experiment looking at the relation between the adding of a decoy- or superior
option and the emotional response of the decision maker about its choice does not seem to
The behavioral experiment used will be in the form of an online survey. In this survey,
the participants will be confronted with four products with different attributes. They are
asked questions about their preferences concerning the attributes of the products and their
familiarity and use of these products. After this they need to make a decision between two
options of each product with different attributes. Then a third option, which is either a decoy-
or superior option to either of the two options in the core set, will be shown. The subjects are
then asked about their level of satisfaction, disappointment and regret with their former made
choice, knowing that this option is now also available in the market. They will also be asked
how good they believe their previously chosen option is compared to the newly added option.
In the end, the questions about the subjects’ preferences towards certain attributes are asked
again to measure a possible change in preferences.
The results show that the emotional responses after the adding of a third decoy- or
superior option move in the predicted direction. The adding of a decoy option increases
satisfaction and decreases disappointment and regret, while the adding of a superior option
decreases satisfaction and increases disappointment and regret. The attributes of the chosen
option seem important with the adding of a superior option and not with the decoy option,
since the emotional responses are stronger when the superior option is added to the chosen
option compared to the non-chosen option and there is no difference in emotional responses
between the adding of a decoy option to the chosen- or non-chosen option. In addition, the
results show that decision makers change their preferences towards the dimension that was
most important in their chosen option and away from the one that was least important.
2. Literature review
2.1 Decoy effect
The decoy effect is a long established phenomenon in the literature about consumer
dominated alternative (decoy option). This added option will increase the probability that the
dominating (target) item in this choice set is preferred. An asymmetrically dominated
alternative is asymmetric in the sense that it is only dominated by one other option in the
choice set (the target) and not by the other (the competitor) (Huber et al., 1982). The decoy
effect, also called asymmetric dominance- or attraction effect was first discovered by Huber
and colleagues (1982). It was discovered by the conduction of an experiment that compared
how many times participants preferred the target option compared to the competitor option,
with the decoy option present and without it. The results then showed that with the decoy
option present, the target option was chosen significantly more than without it.
2.2 Choice models
Before the discovery of the decoy effect (Huber et al., 1982) choice models commonly
used the idea that adding a new brand into an existing market would make the new brand gain
share proportionately. This means that a new product would take share proportionately to the
original shares of the existing products (Luce, 1959). However, at this time it was already
proven that proportionality often fails (Debreu, 1960). More focus was then laid on the
assumption of similarity. The similarity hypothesis states that products which are similar to a
new product entering the market will lose more share to this new product than less similar
products (Tversky, 1972). Both proportionality and the similarity hypothesis share that
regularity is required for them to hold. Regularity means that when a new alternative is added
to the choice set the probability of choosing any alternative of the original set should not
increase. This is a minimal condition for many choice models (Luce, 1977). However, the
decoy effect is a clear violation of proportionality, the similarity hypothesis and regularity.
The study of Huber and colleagues (1982) thus proved the failing of these three important
2.3 Non-dominated alternatives
Even though Huber et al. (1982) proved the decoy effect, there is no consensus about
the possible explanations and underlying causes of the decoy effect. Huber and Puto (1983)
did extend the finding of the decoy effect from dominated alternatives and proved that it is
also present in non-dominated alternatives, which are relatively inferior in the choice set.
Relative inferiority means that the option cannot just be stated to be inferior, because the
utility function of an individual is not known. However, a relative inferior option is less
desirable in the sense that the tradeoff ratio between the dimensions is worse than the tradeoff
ratio from the options in the core set, such as choice options positioned in the light grey area
of figure 1.
Figure 1. Possible location of dominated alternatives and non-dominated alternatives, which are relatively inferior, compared to a competitor and target option (Huber et al., 1982).
2.4 Explanations
After the decoy effect was established, several explanations were tested. Ratneshwar
familiarity aspects. It is hypothesized that lack of meaningfulness and familiarity of the
products that were used in the experiments of Huber and colleagues (1982) caused the
observed decoy effect found in the experiments. The experiments show that the hypothesis
about the lack of meaningfulness of the products is supported. When extensive explanations
about the used products are added, the decoy effect is less strong. However, the results are
less supportive about the familiarity hypothesis (Ratneshwar et al., 1987).
In the following years, the explanations that were mostly analyzed center around two
lines of theory. Within the first line of theory the explanations for the decoy effect are related
to the idea of dimensional weighting and suggest that subjects’ perception of the given stimuli
is changed by adding the decoy option (Huber et al. 1982, Huber & Puto, 1983, Simonson &
Tversky 1992; Tversky & Simonson, 1993).Tversky and Simonson (1993; Simonson &
Tversky, 1992) developed a model of ‘’extremeness aversion’’ in which both loss aversion
(Tversky & Kahneman, 1991) and dimensional range are used to explain the observed choice
behavior. In the model overall value is replaced by local contrasts and then the concept of loss
aversion is added. With loss aversion options are compared to a reference point, in which
perceived losses loom larger than gains (Tversky & Kahneman, 1991). These different factors
then add up to a rational choice model that can explain the decoy effect. The dominating
option can, for example, seem to be more attractive after adding the decoy option, if the
dimension on which the dominating option is weaker than its competing option looms
smaller, because the added decoy stretches this dimension. This seemingly decreases the
weakness of the dominating option compared to its competitor and makes it more attractive,
leading to the decoy effect.
The other line of theory is based on different heuristics that people use which can
cause the decoy effect. Huber and Puto (1983) already found that when subjects were asked
was a good compromise. The first theory that went further on this compromise explanation
was Simonson (1989) who talked about the ‘’compromise effect’’ as a reason for the decoy effect to occur. This idea is based on the added value approach, which means that the
preference of an option can change because of other reasons than its dimensional value. In the
study of Simonson (1989) ‘’justifiability’’ is shown to be a reason for the decoy effect. When people were informed that they had to explain their decision afterwards, the dominating
option was chosen more often, thus increasing the decoy effect. This is called the
‘’compromise effect’’ since, the decoy option makes the dominating option look like a good and justifiable compromise. The heuristic here is that people can use the strategy of ‘’always choose the compromise option’’ as a rule of thumb.
Who build further on this discovery of the decoy effect (Huber et al., 1982, Huber &
Puto, 1983) and the theories developed afterwards (Simonson, 1989; Simonson & Tversky
1992; Tversky & Simonson, 1993) were Ariely and Wallsten (1995). Ariely and Wallsten
(1995) developed a theory that combined the two lines of explanations used before, namely
dimensional weighting (Simonson & Tversky 1992; Tversky & Simonson, 1993) and the
added value approach (Simonson, 1989). The fact that a product’s value and relationship
towards the other products is important is taken from the dimensional weighting approach.
From the added value approach the insight is used that it is important to realize that people
have limited processing capability and that a there is a strong relationship between the
structure of the choice problem and the choice process itself. The goal of the study was to
form a theory that could be used as an explanation for the decoy effect and deliver a more
general framework for understanding choice behavior. The theory bases itself on dominance
seeking, where the decision maker actively looks for ways in which he can simplify his task.
Ariely and Wallsten (1995) assumed that a decision maker will try to reduce the difficulty of
between the choice items is formed. This subjective dominance can help, because dimensions
that are important to help distinguish among different items, will gain more weight in the
decision-making process. A decision maker can establish subjective dominance by ignoring
the differences on certain product attributes, while enlarging differences on other attributes.
This means that the attributes can be compared on fewer dimensions, the weights of the
attributes are then determined by their local context. The weights and subjective value will
depend on the attributes of the decoy option. This theory of dynamic choice reconstruction is
based on the results found that show that decoy options effect the influence of context on
elicited values and choice (Ariely & Wallsten, 1995). However, these weight-change models
which say that the decoy effect occurs because of changes in relative weighting or importance
of certain dimensions, do not gain much support. As Ariely and Wallsten (1995) already
mention themselves, the results of the experiment do not seem to proof that the weighted
additive model explains the decoy effect.
2.5 Applicability decoy effect
All previously mentioned experiments on the decoy effect are about consumer choice,
which is where most of the literature focuses on (Huber et al., 1982; Huber & Puto, 1983;
Ratneshwar et al,. 1987; Simonson, 1989, Simonson & Tversky 1992, Tversky & Simonson;
1993, Ariely & Wallsten, 1995). However, it is good to realize that the decoy effect can
account for many more situations. Namely, the effect has also been observed in job candidates
(Highhouse, 1996), choice among monetary gambles (Wedell, 1991), political candidates
(Pan, O’Curry, Pitts, 1995), political issues (Herne, 1997), real in-store purchases (Doyle, O’Connor, Reynolds & Bottomley, 1999) and within the medical area (Rubaltelli et al., 2008). These experiments show the wide range of applicability of the decoy effect. Almost all
experiments in the decoy effect literature use hypothetical choice tasks, except for Simonson
particular decision. Herne (1999) investigated if there would be any change in the decoy
effect if real incentives are used instead of hypothetical ones. She used monetary gambles just
like Wedell (1991) and the results show that the decoy effect is still present with the use of
real incentives in the same manner as when hypothetical incentives are used.
2.6 New insights
More recent studies have looked at the decoy effect from different angles. Recently it
has been related to the new field of neuroeconomics. Hedgcock and Rao (2009) test the
explanation for the decoy effect in the area of trade-off aversion. Namely, Hedgcock and Rao
(2009) state that people are often confronted with choices involving attributes that can be
equally (un)attractive on different aspects. This can make a choice between them difficult.
The adding of a decoy option can help to make a choice between these two options, because it
can reduce the trade-off difficulty experienced by a decision maker (Hedgcock & Rao, 2009).
In addition, the study of Luce, Bettman and Payne (1999) explains that trade-off choices can
become emotionally inconvenient. This makes that decision makers can become trade-off
averse. Hedgcock and Rao (2009) show in their functional magnetic resonance study that the
activation of brain areas related to the experience of negative feelings decreased after the
addition of a third decoy option. This study thus clearly shows the importance of emotions in
trade-off choices, which is important in the search for the underlying mechanisms of the
decoy effect. It is concluded that the adding of a decoy option can be the reason to decrease
trade-off aversion and consequently decrease the experience of negative feelings related to it.
Mao & Oppewal (2011) relate the decoy effect to the way in which people think. They
distinguish between people who rely more or who rely less on intuitive reasoning in judgment
and decision-making and between people who rely more or who rely less on rational thinking.
The results show that the decoy effect is stronger when people rely more on intuitive
personal characteristics can be influential in how strong the decoy effect turns out to be.
Supporting this importance of individual differences is the study of Pocheptsova, Amir, Dhar
and Baumeister (2009). This study shows that when subjects’ cognitive resources are
depleted, their intuitive and automatic information processing is increased and that this
magnifies the decoy effect.
Another new angle for looking at the decoy effect has recently been in the subject of
‘’saliency’’. Bordalo, Gennaioli and Shleifer (2013) developed a model of context-dependent preferences. In this model it is theorized that the attention of a decision maker will be drawn
to attributes of the product that appear to be more salient. An attribute will be more salient if it
stands out compared to the other attributes relative to the attribute’s average level, where the average level is referred to as the ‘’reference good’’. The salient attributes will then gain more
weight in the decision-making process. This context- dependent preference model
incorporating saliency, can also account for the decoy effect. The reason is that when a decoy
option is added, the reference good changes and another attribute can become salient. This
can make the target option seem like the better choice. Bordalo and colleagues (2013)
furthermore suggest that there is an asymmetry present in the decoy effect. However, Bordalo
and colleagues (2013) were not the first ones that suggested this. The experiments of Heath
and Chatterjee (1995) also seem to have achieved the same conclusion. These experiments
show that lower-quality competitors lose a larger share than higher-quality competitors when
a decoy option is added. Bordalo and colleagues (2013) conclude that there is asymmetry in
the way that choices will be influenced towards goods with a higher quality/price ratio. The
asymmetry in the decoy effect makes the models of Bordalo and colleagues (1995) and Heath
and Chatterjee (1995) differ significantly from previous models about context dependence and
try to minimize his losses on all attributes, which would lead to decision preferences being
more on the middle options, without any asymmetry present.
The study of Bordalo and colleagues (2013) is partly related to the study of Koszegi
and Szeidl (2013) who develop a theoretical model about focusing. In this study, focus is
higher on attributes in which the decision possibilities differ more and decision makers
consequently overweigh these attributes in the decision-making process. This means that there
is a ‘’bias toward concentration’’ present, because a large advantage gets much more focus than multiple small disadvantages. A decision maker will thus be more likely to choose an
option which has advantages concentrated in fewer attributes. This bias towards concentration
can be linked to the decoy effect, since decision makers would then focus more on the single
attribute in which the dominant target option is better than the decoy option, instead of
focusing on the multiple differences that exist between the target- and the competitor option.
2.7 Superior options
That the adding of a decoy option changes the preferences of a decision maker has
been addressed in the previous part. Before the study of Huber and colleagues (1982) the fact
that adding a (relatively) inferior decoy option changes preferences did not seem like a logical
process. However, that the adding of a superior option changes preferences towards this better
option seems predictable. If an option clearly dominates on all dimensions, it will be most
desirable. Just like relative inferiority (Huber & Puto, 1983) an option can also be relatively
superior. In which relative superiority means that an option cannot just be stated superior,
because the utility function of an individual is not known. This option will be more desirable
in the sense that the tradeoff ratio between the dimensions is better than the tradeoff ratio
Figure 2. Possible location of dominating alternatives and non-dominating alternatives, which are relatively superior, compared to a competitor and target option.
Simonson (1989) already suggested that the decoy effect could be the result of
justifiability of decisions. Choosing a superior option can increase both utility and
justifiability. There does not seem to be a reason why a decision maker would not choose an
option that is superior on all dimensions. However, if an option is only relatively superior,
justifiability can be a good reason to choose the option. Slovic (1975) did four experiments in
which the subject had to decide between two options that they perceived as equally desirable.
One of the options was then made superior on one important dimension, but inferior on the
other dimension, so as to cancel the advantage. Most of the subjects still chose this option
compared to the other equally desirable one, supporting the idea that choices are determined
by reasons that are easily justifiable to oneself and others. An option that is relatively superior
is not even equally desirable compared with the target option, but it is relatively even more
desirable. The adding of a (relatively) superior option, should thus lead to a change in
of Shafir, Simonson and Tversky (1993) who explain that when having to decide between two
options, the pros and cons need to be given a certain weigh to be able to make a decision.
They state that the pros of a certain option will be given larger weight when choosing and that
the cons will be given larger weight when rejecting.
2. 8 Emotions in decision-making
The effect of emotions on decision-making can be substantial and should not be
neglected (Rick & Loewenstein, 2008). The psychology that lies behind decision-making has
been explained by both normative and descriptive thoughts. Experiments are often set up to
improve on the, relatively conservative, normative accounts like expected utility theory. The
descriptive accounts regularly try not to deviate too much from the normative theories,
leading to a conservative bias (Rottenstreich & Shu, 2004). This conservative bias can be seen
as a positive aspect in decision-making research, since a theory with small differences
between descriptive and normative theories, like prospect theory (Kahneman & Tversky,
1979), can be very influential. However, this conservative bias made the incorporation of
‘’affect’’ in the models get neglected. According to Rottenstreich and Shu (2004) this is the case, because emotional experience generally does not really fit into the normative models. In
contrast, Rick and Loewenstein (2008) state that incorporating emotions into the
decision-making process can be completely in line with these models. For example, utility
maximization does not rule out in any way that a prediction of emotions plays part in the
maximization of someone’s utility. Whether emotions fit in the conservative models or not, the effect of emotions on decision-making has not been denied anymore. Since the up rise of
behavioral economics the understanding of emotions on decision-making had been of great
2.9 Valuation
There are several different ways for decision makers to form their valuation and
emotions can be of great influence in some of them. Valuation by calculation and valuation by
feeling are two different psychological ways that are distinguished by Hsee and Rottenstreich
(2004) for people to form their subjective value. Subjective value is a way of valuation that
can be influenced by emotions. To illustrate subjective value, one can look at the study of
Kahneman and Knetsch (1992). Their study showed that moral satisfaction of giving to
charity can influence the (subjective) value assigned to this. Valuation by calculation and
valuation by feeling, as the names suggest, differ in their way of valuing an option by either
rationally calculating the value using an algorithm, or by letting feelings influence how you
value a product (Hsee & Rottenstreich, 2004). This can be compared to valuing an option
because of ‘’reason’’ or because of ‘’feeling’’ (Damasio, 1994; Pham, 1998); because of logical rules or anticipated emotional reactions (Baron, 1992) and because of deliberate- or
automatic valuation processes (Frederick, 2002).
The influence of emotions on decision-making also comes in to place in the study of
Finucane et al. (2000), who explain the ‘’affect heuristic.’’ People often rank goods towards
which they have negative feelings as high risk and low benefit and goods towards which they
have positive feelings as low risk and high benefit, however risk and benefit are generally
positively correlated. This leads to a bias caused by the ‘’affect’’ towards a product. It has also been suggested by Schwarz and Clore (1996) that affect is often used as a source of
information by their notion of ‘’feelings-as-information’’. It is stated that people often make a
choice by asking themselves how they feel about a certain option. Furthermore, this feeling is
most probably also influenced by the pre-existing mood in which people are when they have
feeling (Hsee & Rottenstreich, 2004) and that this valuation process can contain systematic
errors, because of the influence of a decision maker’s mood.
The different studies mentioned before suggest that the preferences of decision makers
can actually be controlled or influenced by taking note of these different valuation processes
and their characteristics. This can be done by looking at the underlying psychological
processes a decision maker has and this malleability of preferences can be seen as a central
finding in the decision-making literature (Rottenstreich & Shu, 2004).
2.10 Different types of emotions in decision-making
There are several different types of affective experiences that can be distinguished in
decision-making. Rottenstreich and Shu (2004) explain that these can be separated by looking
at different emotions depending on when in the decision-making process they occur.
Rottenstreich and Shu (2004) distinguish between emotions at decision time, emotions that
occur after someone has made a decision, but when he does not experience the outcome of his
decision yet and emotions that arise after the consequences of the decision are present. Two
different kinds of emotions are present at decision time, namely emotions that one anticipates
to arise because of a certain decision and emotions related to the process of making a decision
itself (Rottenstreich & Shu, 2004). Rick and Loewenstein (2008) distinguish between
expected- and immediate emotions, in which immediate emotions are subdivided into
‘’integral’’ emotions and ‘’incidental’’ emotions. Expected emotions are the emotions that do not occur yet at the time of decision-making, but only when the outcome of a decision is
experienced. In contrast, immediate emotions are experienced during decision-making.
Within immediate emotions, integral emotions are about the expectations that you have about
the outcome of you choices, but these emotions are felt at the moment of decision-making.
Integral emotions are thus elicited by the target or features of the target. However, the features
imagination about the target features (Cohen, Pham & Andrade, 2007). Incidental emotions
also arise when making a decision, but these are emotions unrelated to the decision in itself
and are for example related to the emotions caused by you external environment at the
moment of choice (Rick & Loewenstein, 2008).
Zeelenberg, Van Dijk, Manstead and van der Pligt (2000) explain that decision makers
are afraid to experience regret or to be disappointed by their decision after they have made it.
This fear can already influence their decision before they make it. People thus make their
decisions, by taking their post-decision feelings into account. In addition, it has been
investigated how people handle the negative emotions that they feel when having to make a
decision (Luce, 1998; Luce, Bettman & Payne, 1997; Luce Payne & Bettman, 1999). During
a decision-making process a decision maker will generally try to minimize these negative
emotions. Decision makers will try to use avoidant responses (e.g. using the status quo) as a
way to minimize these negative emotions. Related to this, Botti and Iyengar (2003) found that
when people are not given a choice, but are just assigned to an outcome they can sometimes
be happier with this than when they get the option to make the decision by themselves.
The previously mentioned studies show the importance of emotions with respect to
decisions and the decision-making process. The link between these influences of affect and
the adding of decoy- or superior options can be logically formed. Fear to experience regret or
to be disappointed with your decision afterwards, can be of large influence on the option a
decision maker prefers (Zeelenberg et al., 2000). The adding of a decoy option afterwards can
influence regret and disappointment, since it makes the target option seem more attractive.
Also, the adding of a superior option after the decision has been made will influence these
emotions, since people will generally prefer the superior choice. Furthermore, minimization
of the negative emotions related to decision-making in itself (Luce, 1998; Luce et al., 1997;
make the decision-making process easier, by for example the compromise effect (Simonson,
1989) or the decrease of trade-off aversion (Hedgcock & Rao, 2009).
3. Hypotheses
Based on the literature review several hypotheses can be formed. The fMRI study of
Hedgcock and Rao (2009) showed that the activation of brain areas related to the experience
of negative feelings decreased after the addition of a third dominated alternative. This
suggests that the addition of a decoy option decreases trade-off aversion and because of this
the negative emotions related to trade-off aversion from the subjects. Trade-off aversion
decreases, because the adding of a decoy option makes it easier to choose the target.
Furthermore, the adding of a decoy option increases justifiability, because of the compromise
effect (Simonson, 1989). Moreover, the target option now appears to be the more salient
option (Bordalo et al., 2013) and the focus will be laid on its pros compared to the decoy
option, making it seem more attractive (Koszegi & Szeidl, 2013). Also, the integral emotions
that a decision maker has about the outcome of his choice will become more positive, since
the option now shown verifies that the outcome of his made decision will appear to be better
(Rick & Loewenstein, 2008). In addition, the fear to experience regret or to be disappointment
with the outcome of the decision maker’s choice will decrease (Zeelenberg et al., 2000). The
adding of the decoy option to the option the decision maker chose, thus verifies that it was a
good decision. This leads to the first hypothesis that a decision maker becomes happier after
the addition of a decoy option to its chosen option.
Hypothesis 1
The adding of a decoy option to the option the decision maker chose should make him happier
with his former made decision.
A (relatively) superior option is highly justifiable (Simonson, 1989) and preferences
decision has been made the decision maker’s integral emotions (Rick & Loewenstein, 2008)
will become more negative and the level of fear to experience regret or to be disappointment
will increase (Zeelenberg et al., 2000) since the decision maker will be disappointed that he
was not able to choose this option beforehand. In addition, economic intuition can tell us that
the emotional responses will be specifically strong when the superior option is superior to the
option the decision maker chose earlier, since this option then has the characteristics the
decision maker is looking for.
Hypothesis 2
The adding of a superior option to the option the decision maker chose should make him less
happy with its former made decision.
There is a distinction between a decoy- or superior option added to the choice option
the decision maker made or to the choice option he did not make. It could be hypothesized
that when an option is added to the non-made choice the decision maker’s happiness is not
influenced, because he did not want a product with such attributes to begin with. However, it
can also turn out that the effect of an added decoy to a non-made choice still elicits more
positive emotions for reasons mentioned in hypothesis 1.
Hypothesis 3
The adding of a decoy option to the option the decision maker did not choose should make
him not care too much or make him happier anyway.
Adding a superior option to the option the decision maker did not chose can, equally
as with the decoy option, not give any large emotional responses, because the decision maker
did not want a product with such attributes. However, it can also still elicit emotions of
disappointment and regret, since it remains an option with relatively superior characteristics.
The adding of a superior option to the option the decision maker did not chose should make
him not care too much or make him less pleased anyway.
Having to make a decision between two options of a product with different attributes
is a difficult and unpleasant process (Hedgcock & Rao, 2009). The preferences concerning the
attributes of a product can change from before seeing the decision possibilities and after the
decision has been made. Decision makers can experience a change in preferences towards the
dimension that is most important in the product they have chosen and away from the
dimension that is least important (important meaning that this dimension is the more attractive
one in a certain product option compared to the other product option). This change in
preferences increases the justifiability (Simonson, 1989) both to oneself and to others.
Furthermore, this change in preferences can make it feel that they will experience less regret
or disappointment with their made decision (Zeelenberg et al., 2000).
Hypothesis 5
The preferences of a decision maker will change towards the dimension that is most important
in the product option that a decision maker chose and away from the least important
dimension, when comparing the preferences before the decision maker has seen the choice
options to the ones after the decision has been made.
4. Design 4.1 Participants
The sample consisted of 80 participants. From these 80 subjects, 6 did not finish the
survey till the end. Subsequently, these 6 participants were deleted from the analyses. The
majority of the 74 participants that are left is female, namely the sample consists of 42
females and 32 males. The age of the participants ranges from 19 till 59 years. In which a
4.2 Measures
To test the hypotheses the data about the level of satisfaction, regret and
disappointment after the adding of the third option will be analyzed. This data needs to be
linked to if the decoy- or superior option of the chosen or non-chosen option was shown.
Also, it is analyzed if the participants perceive the decoy- and superior option as they were
intended. Furthermore, it can be investigated if people try to justify their level satisfaction,
regret and disappointment to themselves or others, by looking if there is a change in
preferences towards the dimension that is important in their chosen option when they are
again asked about the importance of the dimensions at the end of the survey.
4.2.1 Satisfaction. As the first measurement of emotional response, the level of
satisfaction is asked. The level of satisfaction from the subjects with their former made
decision, taking into account the third option that has entered the market, is measured on a
seven-point Likert scale. The answer possibilities range from ‘’Completely dissatisfied’’ to
‘’Completely satisfied’’ and in between at option 4 there is the neutral level stating ‘’Neither satisfied or dissatisfied’’. Hypothesis 1 leads us to predict that after the adding of a decoy option to the chosen option, a decision maker will be more satisfied with its former made
decision. According to hypothesis 2, the adding of a superior option to the chosen option,
should made a decision maker less satisfied. The level of satisfaction after the adding of a
decoy option to the non-chosen option is also investigated to see if the subjects respond the
same to a decoy option to their chosen- or non-chosen option or if they might not care too
much about a product with such attributes, as stated in hypothesis 3. The same procedure goes
for testing hypothesis 4, thus also analyzing the level of satisfaction after the adding of a
superior option to the non-chosen option. Moreover, the differences between these four
groups: decoy after chosen option, decoy after non-chosen option, superior after chosen
conclusions if the level of satisfaction is higher/lower in one group than the others, in the way
predicted by the hypotheses.
4.2.2 Disappointment. The level of disappointment does not really have a neutral
value like satisfaction. The first option is ‘’Not disappointed at all’’ and from there on
disappointment increases. This is why we would expect that after a decoy option to the chosen
option the subjects are the least disappointed since hypothesis 1 states that the adding of a
decoy option to the chosen option actually increases the happiness of the decision maker.
After the adding of a superior option to the chosen option, the happiness of the decision maker
with its former choice will decrease according to hypothesis 2, this is why we measure if the
level of disappointment is higher after the adding of a superior option. To test hypothesis 3
and 4 it is investigated if the direction of adding a decoy- or superior option to the non-chosen
option makes the level of disappointment go in the same direction as predicted in hypothesis 1
and 2.
4.2.3 Regret. After the adding of a decoy option to the chosen option the decision
maker should be happier with its former made decision according to hypothesis 1, this means
the level of regret should be low. When a superior option is added to the chosen option the
level of regret is hypothesized to be high, as stated in hypothesis 2. To test hypothesis 3 and 4
it is again checked if the direction of adding a decoy- or superior option to the non-chosen
option makes the level of regret go in the same direction as predicted in hypothesis 1 and 2.
4.2.4 Comparison third option. The participants were asked how they felt their
chosen option was compared to the third added option (either a superior option or decoy
option to one of the options in the core set). In the answer possibilities number 3 was ‘’About the same’’ and everything below was that their option was worse and above better. It would thus be predicted that they feel their chosen option is better compared to a decoy option and
4.2.5 Change in preferences. To test hypothesis 5 it needs to be analyzed if the
participants change their preferences towards the dimension that is most important in the
product option that a decision maker chose and away from the dimension that was least
important. This is done by comparing the preferences before the decision maker has seen the
choice options (preferences 1) to the ones after the decision has been made (preferences 2).
To test if the participants change their preferences, the difference between the preferences
measure before the decision-making process and the preferences measure afterwards are
calculated by subtracting the latter from the former. If the preferences are about the dimension
that was most important in the decision option chosen it would thus be predicted that the later
asked preferences are larger than the previously asked preferences, leading to a negative
number as a difference. If the preferences are about the dimension that was least important it
would be predicted that the later asked preferences are smaller than the first, leading to a
positive number as the difference.
4.3 Procedure
4.3.1 Overview components of survey. To answer the research question an online
survey was used. This survey consists of different components. Each of the four products
receives two preference questions, one on each dimension of the product. Also, the subjects’
familiarity with the product and frequency of use of the product are asked for all the products.
The familiarity and frequency of use questions served more as a filler task between the
preference and decision-making questions, to try to cover up the main goal of the study.1
After the preference, familiarity and frequency of use questions the subjects needed to
make a decision between two options of a product, which differ in their attributes. These two
options (A and B) will be the core set (competitor and target option). Subsequently, a decoy-
or superior option of one of the options in the core set will be added to the choice set.
1 The measurement of familiarity with products was also used as a filler task in Simonson
Afterwards, the participants are asked questions about their level of: satisfaction, regret, and
disappointment with their former made decision, taking into account that this third option is
now also available on the market. Finally, the questions about the participants’ preferences
about the product attributes are asked again. All these steps will be repeated for all four
different products. At the end of the survey the participants are asked to fill in their gender
and age. In appendix A all questions from one version of the survey can be found.
Table 1
Components of survey Preference dimension 1
Preference dimension 2
Familiarity product
Frequency of use product
Decision-making between the two options of the core set (option A and option B)
Adding of a third option (decoy- or superior option of option A or B)
Level of satisfaction
Level of disappointment
Level of regret
Preference dimension 1 (second time)
Preference dimension 2 (second time)
Gender
4.3.2 Type of answer possibilities. In the survey participants are asked to answer all
questions according to their own preferences.2 In all answer possibilities, except gender and
age, five- or seven-level Likert scales are used. The use of Likert scales gives the subjects the
possibility to answer quite neutrally if they want to. The questions used in the survey with
their answer possibilities can be found in appendix A.
4.3.3 Product choice. Four different product categories are used in the experiment.
These products have been chosen, because they have already repeatedly been used in
experiments that proved the decoy effect. The products used are: cars3, beer4, TVs5 and
restaurants.6
4.3.4 Location and dimensions of decoy- and superior options. The values of the
core set and decoy options are derived from values previously used. However, they are
adjusted to current prices of the products and converted to euros and kilometers when
necessary. Furthermore, the relationship to the superior option has been derived from the
distance between the options in the core set and the decoy. This was done to prevail that the
decoy- and superior option will differ largely in their distance to the core set, such that this
difference could influence results about the decision maker’s emotions about the adding of the
third option. The fixed distances between the core set and the decoy- and superior option can
be found in appendix B. The decoy- and superior options used for the products in the survey
2
Ariely and Wallsten (1995) tested if results were stronger if people had to decide in a way that reflected the preferences of the population, contrary to deciding according to their own preferences. This was done to prevent participants from deciding they personally ‘’did not care’’ about a certain dimension to ease the decision making process. The results show that letting people decide in a way that reflects the preferences of the population does not seem to give an advantage over letting participants decide according to their own preferences. 3
Used by Huber and colleagues (1982); Huber and Puto (1983); Simonson (1980); Pan and Lehmann (1993); Lehmann and Pan (1994).
4
Used by Huber and colleagues (1982); Huber and Puto (1983); Simonson (1989). 5
Used by Huber and colleagues (1982); Simonson (1980); Ratneshwar and colleagues (1987); Pan and Lehmann (1993); Lehmann and Pan (1994).
6
balance each other out in whether they are dominated or dominant on the range dimension (on
which the target is inferior) or frequency dimension (on which the target is superior) so as to
not give decoy- or superior options a (dis)advantage for stronger results because of the
location of the option.7
The products in this survey were explained on two dimensions.8 The dimensions used
were ride quality and liter per 100 km for the car; quality and price for the six-pack of beer;
durability and price for the TV and food quality and driving time for the restaurant. The
decoy- and superior options of each product, accompanied by a graph showing their location,
can be found in appendix B.
4.3.5 Division of third option. Each time a third option is added this is either a decoy-
or superior option to one of the possibilities in the core set (option A and B). The division of
the decoy- or superior options from option A or B over the four surveys is such that each
survey has one decoy- and superior of option A and of option B. Furthermore, for each
product the decoy- and superior options of option A and B are used in total one time over all
surveys.
Table 2
Third option division over the four surveys
Survey 1 Survey 2 Survey 3 Survey 4
Car A superior B decoy B superior A decoy
Beer A decoy A superior B decoy B superior
TV B decoy B superior A decoy A superior
Restaurant B superior A decoy A superior B decoy
7
Huber and colleagues (1982) show that the decoy effect is strongest when the decoy is dominated on one dimension compared to two dimensions. More specifically, it is stronger when the decoy option is dominated on the range dimension rather than the frequency dimension. The study concludes that the placement of the decoy is an important factor in the robustness of the decoy effect, compared to just incorporating a dominance relationship. 8
Ariely and Wallsten (1995) compared if the decoy effect was stronger when products were specified on three instead of two dimensions, this was done because decision makers could then less easily compare and trade-off pair ratios, but the results show that using three dimensions did not increases the strength of the decoy effect. This is in line with the results found by Huber and Puto (1983).
4.3.6 Hypothetical incentives. In the survey the choice task was completely hypothetical,
such that the survey could be spread around online to help increase the number of
participants.9
5. Results
In order to analyze the difference in the levels of satisfaction, disappointment and regret the
average level of these emotional responses per person from all products together are
compared for the four categories: decoy after the chosen option, superior after the chosen
option, decoy after the non-chosen option and a superior after the non-chosen option. These
categories are compared by the use of a two-tailed paired samples t test. A two-tailed paired
samples t test was also used to investigate the change in preferences. This was done by
comparing the first- and second asked preferences of the most and least important dimension
for the average per person of the different products per survey, as well as for the average per
person for all products together.
9
Herne (1999) investigated if there would be any change in the decoy effect if real incentives are used instead of hypothetical ones. The results show that there is no advantage in using real- compared to hypothetical choice tasks.
Table 3
Levels of satisfaction after the adding of a third option
Decoy after chosen (DAC) Superior after chosen (SAC) Decoy after non-chosen (DANC) Superior after non-chosen (SANC) Product Mean (SD) N Product Mean (SD) N Product Mean (SD) N Product Mean (SD) N Beer 5.40 1.26 10 Car 2.57 1.62 7 Beer 5.38 2.26 8 Car 3.64 1.57 11 TV 5.20 2.05 5 Restaurant 2.50 1.55 16 TV 5.38 1.85 13 Restaurant 4.00 1.41 2 Car 5.60 1.76 15 Beer 3.88 2.36 8 Car 6.25 1.50 4 Beer 5.27 2.15 11 Restaurant 7.00 n.a 1 TV 2.50 1.93 8 Restaurant 6.83 0.71 18 TV 3.36 2.42 11 Beer 5.75 1.16 8 Car 5.40 1.55 15 Beer 4.82 2.40 11 Car 6.00 1.41 4 TV 6.00 2.00 9 Restaurant n.a n.a 0 TV 6.10 1.91 10 Restaurant 5.32 2.08 19 Car 4.20 1.79 5 Beer 3.50 1.77 8 Car 4.85 1.99 13 Beer 4.70 2.06 10 Restaurant 5.82 1.74 17 TV 2.55 1.97 11 Restaurant 4.00 n.a 1 TV 4.86 1.35 7 All 5.59 1.67 70 73 3.37 2.06 73 All 5.67 1.90 78 All 4.65 2.06 75
DAC vs. SAC DAC vs. DANC DAC vs. SANC DANC vs. SAC
Mean DAC Mean SAC Difference N Mean DAC Mean DANC Difference N Mean DAC Mean SANC Difference N Mean DANC Mean SAC Difference N 5.59 3.40 2.19*** 70 5.59 5.79 0.20 70 5.59 4.66 0.93*** 70 5.73 3.37 2.36*** 73
DANC vs. SANC SAC vs. SANC
Mean DANC Mean SANC Difference N Mean SAC Mean SANC Difference N 5.65 4.65 1.00*** 75 3.37 4.66 1.29*** 73 *, **, *** indicate p<0.1, <0.05, <0.01 in the two-tailed paired t test
Table 4
Levels of disappointment after the adding of a third option
Decoy after chosen (DAC) Superior after chosen (SAC) Decoy after non-chosen (DANC) Superior after non-chosen (SANC) Product Mean (SD) N Product Mean (SD) N Product Mean (SD) N Product Mean (SD) N Beer 1.30 0.67 10 Car 2.57 1.40 7 Beer 1.75 1.49 8 Car 2.36 1.03 11 TV 2.00 1.41 5 Restaurant 3.25 1.29 16 TV 1.38 0.77 13 Restaurant 1.00 0.00 2 Car 1.13 0.35 15 Beer 2.25 1.04 8 Car 1.50 1.00 4 Beer 1.45 0.93 11 Restaurant 1.00 n.a 1 TV 3.5 1.60 8 Restaurant 1.00 0.00 18 TV 2.00 1.55 11 Beer 1.25 0.71 8 Car 1.47 0.64 15 Beer 1.27 0.65 11 Car 1.25 0.50 4 TV 1.33 1.00 9 Restaurant n.a n.a 0 TV 1.10 0.32 10 Restaurant 1.79 1.13 19 Car 1.80 1.10 5 Beer 2.38 0.92 8 Car 1.38 0.77 13 Beer 1.30 0.67 10 Restaurant 1.35 1.06 17 TV 2.91 1.64 11 Restaurant 3.00 n.a 1 TV 1.43 1.13 7 All 1.36 0.87 70 73 2.59 1.37 73 All 1.31 0.76 78 All 1.71 1.10 75
DAC vs. SAC DAC vs. DANC DAC vs. SANC DANC vs. SAC
Mean DAC Mean SAC Difference N Mean DAC Mean DANC Difference N Mean DAC Mean SANC Difference N Mean DANC Mean SAC Difference N 1.36 2.57 1.21*** 70 1.36 1.31 0.04 70 1.36 1.71 0.36*** 70 1.30 2.59 1.29*** 73
DANC vs. SANC SAC vs. SANC
Mean DANC Mean SANC Difference N Mean SAC Mean SANC Difference N 1.29 1.71 0.41** 75 2.59 1.73 0.86*** 73 *, **, *** indicate p<0.1, <0.05, <0.01 in the two-tailed paired t test
Table 5
Levels of regret after the adding of a third option
Decoy after chosen (DAC) Superior after chosen (SAC) Decoy after non-chosen (DANC) Superior after non-chosen (SANC) Product Mean (SD) N Product Mean (SD) N Product Mean (SD) N Product Mean (SD) N Beer 1.50 0.85 10 Car 2.71 1.38 7 Beer 1.63 0.92 8 Car 2.45 1.13 11 TV 2.00 1.00 5 Restaurant 3.31 1.30 16 TV 1.15 0.38 13 Restaurant 1.00 0.00 2 Car 1.33 0.72 15 Beer 1.88 0.83 8 Car 2.00 1.15 4 Beer 1.73 1.10 11 Restaurant 1.00 n.a 1 TV 3.50 1.60 8 Restaurant 1.11 0.47 18 TV 2.00 1.55 11 Beer 1.13 0.35 8 Car 1.60 0.91 15 Beer 1.45 1.04 11 Car 1.50 1.00 4 TV 1.22 0.67 9 Restaurant n.a n.a 0 TV 1.00 0.00 10 Restaurant 1.79 1.18 19 Car 1.60 0.89 5 Beer 2.25 1.28 8 Car 1.31 0.63 13 Beer 1.50 0.85 10 Restaurant 1.29 0.99 17 TV 3.00 1.55 11 Restaurant 3.00 n.a 1 TV 1.71 1.25 7 All 1.37 0.80 70 All 2.60 1.41 73 All 1.31 0.71 78 All 1.83 1.17 75
DAC vs. SAC DAC vs. DANC DAC vs. SANC DANC vs. SAC
Mean DAC Mean SAC Difference N Mean DAC Mean DANC Difference N Mean DAC Mean SANC Difference N Mean DANC Mean SAC Difference N 1.37 2.59 1.21*** 70 1.37 1.29 0.09 70 1.37 1.81 0.44** 70 1.27 2.60 1.33*** 73
DANC vs. SANC SAC vs. SANC
Mean DANC Mean SANC Difference N Mean SAC Mean SANC Difference N 1.29 1.83 0.53*** 75 2.60 1.82 0.78*** 73 *, **, *** indicate p<0.1, <0.05, <0.01 in the two-tailed paired t test
5.1 Hypothesis 1
The adding of a decoy option to the option the decision maker chose should make him
happier with his former made decision.
After the adding of a decoy option to the chosen option the average level of
satisfaction for all products is larger than the neutral level of 4 with a mean for all products
together of 5.59. The level of disappointment is on average equal to 1.36, this is quite close to
the lowest available level of 1 which means ‘’Not disappointed at al’’. Regret levels have an
average of 1.37, which is close to ‘’No regret’’.
The level of satisfaction after the adding of a decoy-option to the chosen option is
significantly larger than both the level of satisfaction after the adding of a superior option to
the chosen option and to the non-chosen option.
There is no significant difference in satisfaction levels when comparing the level after
the adding of a decoy to the chosen option and a decoy to the non-chosen option.
The level of disappointment after the adding of a decoy option to the chosen option is
significantly lower comparing it to when a superior option is added to the chosen option and
lower when comparing it to a superior option added to the non-chosen option.
Comparing the level of disappointment after the adding of a decoy option to the
chosen option to the level after the decoy option was added to the non-chosen option shows
no significant difference.
After the adding of a decoy to the chosen option, the level of regret is significantly
smaller compared to the level of regret after both the adding of a superior option to the chosen
option and non-chosen option.
Also with the level of regret there is no significant difference between the adding of a
5.2 Hypothesis 2
The adding of a superior option to the option the decision maker chose should make
him less happy with its former made decision.
When a superior option is added to the chosen option the average level of satisfaction
is smaller than the neutral level of 4 and equal to 3.37 close to ‘Somewhat dissatisfied’’. The level of disappointment has a mean of 2.59, close to ‘’Somewhat disappointed’’. The average level of regret is equal to 2.60, above ‘’Minor regret’’.
The levels of satisfaction are significantly lower after the adding of a superior option
to the chosen option compared to when a decoy option is added to the chosen option and
when a decoy option is added to the non-chosen option.
If the level of satisfaction is compared between the adding of a superior option to the
chosen- and the non-chosen option, the level of satisfaction is significantly lower when the
superior option is added to the chosen option.
The level of disappointment is significantly higher when a superior option is added to
the chosen option, comparing it to the levels of disappointment after the adding of a decoy
option to the chosen option or the non-chosen option.
Just as with the level of satisfaction, the results are stronger when a superior option is
added to the chosen option compared to the non-chosen option. This means that the level of
disappointment is significantly larger when a superior option is added to the chosen option.
After the adding of a superior option to the chosen option, the level of regret is
significantly higher than after the adding of a decoy option to the chosen option or a decoy
option to the non-chosen option.
Similar to the situation with the levels of satisfaction and disappointment, the level of
regret is significantly higher after the adding of a superior option to the chosen option
5.3 Hypothesis 3
The adding of a decoy option to the option the decision maker did not chose should
make him not care too much or make him happier anyway.
The average level of satisfaction after the adding of a decoy-option to the non-chosen
option is 5.67 close to ‘’Mostly satisfied’’. The level of disappointment has a mean close to
‘’Not disappointed at all’’ at 1.31. The average level of regret is also equal to 1.31, close to ‘’No regret’’.
The levels of satisfaction after adding a decoy option to the non-chosen option are
significantly higher when comparing it to a superior option added to either the chosen- or
non-chosen option.
As noted earlier, the levels of satisfaction of a decoy added to the chosen option and
non-chosen option do not differ significantly.
Comparing the adding of a decoy to the non-chosen option with the adding of a
superior option to the chosen- and non-chosen option shows that the level of disappointment
is in both cases significantly lower after the adding of a decoy to the non-chosen option.
As explained in hypothesis 1, there is no significant difference in levels of
disappointment when comparing the level after the adding of a decoy option to the chosen
option to the level after the adding of a decoy option to the non-chosen option.
Looking at the level of regret after the adding of a decoy option to the non-chosen
option shows that this level is significantly smaller compared to the regret levels after the
adding of a superior option to the chosen and non-chosen option.
The level of regret does not differ significantly between the level after adding a decoy
to the chosen option and the level after adding a decoy to the non-chosen option, as noticed
5.4 Hypothesis 4
The adding of a superior option to the option the decision maker did not chose should
make him not care too much or make him less pleased anyway.
After the adding of a superior option to the non-chosen option the average level of
satisfaction is 4.65, a little below ‘’Somewhat satisfied’’. The level of disappointment
averages at 1.71, close to ‘’Slightly disappointed’’. And the level of regret has a mean of 1.83,
just below ‘’Minor regret’’.
The levels of satisfaction after the adding of a superior option to the non-chosen option
are significantly smaller than the levels of satisfaction after adding a decoy option to the
chosen option or a decoy option to the non-chosen option.
When the levels of satisfaction after the adding of a superior option to the chosen
option and the non-chosen option are compared, the results show that the level of satisfaction
is significantly lower when the superior option is added to the chosen option versus the adding
of a superior option to the non-chosen option.
The level of disappointment is significantly higher after the adding of superior option
to the non-chosen option, when comparing it to the levels after the adding of a decoy to the
chosen or non-chosen option.
As already concluded in hypothesis 2, similar to the level of satisfaction, the level of
disappointment is significantly larger when a superior option is added to the chosen option
comparing it to the level after the adding of a superior option to the non-chosen option.
After the adding of a superior option to the non-chosen option, the level of regret is
significantly higher comparing it to the level of regret after the adding of a decoy to either the
The difference between the adding of a superior option to the chosen- and non-chosen
option shows that the level of regret is significantly larger after the adding of a superior option
to the chosen option, as stated in hypothesis 2.
5.5 Hypothesis 5
The preferences of a decision maker will change towards the dimension that is most
important in the product option that a decision maker chose and away from the least important
dimension, when comparing the preferences before the decision maker has seen the choice
Table 6
Change in preferences of most and least important dimension
Most important dimension Least important dimension Product Dimension Preferences 1 Preferences 2 Difference N Preferences 1 Preferences 2 Difference N
Car Liter 5.86 6.00 −0.14 7 5.09 4.64 0.45 11 Car Quality 5.36 5.27 0.09 11 4.71 4.71 0.00 7 Beer Price 4.20 4.50 −0.30 10 4.50 4.38 0.13 8 Beer Quality 6.38 6.00 0.38 8 5.00 4.50 0.50 10 TV Price 4.92 5.00 −0.08 13 5.60 5.80 −0.20 5 TV Durability 5.00 5.20 −0.20 5 5.31 5.23 0.08 13
Restaurant Driving time 5.00 5.50 −0.50 2 4,68 0.70 0.88*** 16 Restaurant Quality 6.50 6.19 0.31** 16 7.00 7.00 0.00 2 Car Liter 6.00 6.25 −0.25 4 5.27 5.27 0.00 15 Car Quality 5.40 5.27 0.13 15 5.25 5.25 0.00 4 Beer Price 4.00 4.25 −0.25 8 4.82 4.45 0.36 11 Beer Quality 5.45 5.27 0.18 11 4.50 4.38 0.13 8 TV Price 4.72 5.27 −0.55* 11 5.13 5.50 0.38* 8 TV Durability 5.88 5.88 0.00 8 4.91 4.73 0.18 11
Restaurant Driving time 5.00 5.00 0.00 1 4.50 3.83 0.67** 18 Restaurant Quality 6.17 6.22 −0.06 18 6.00 6.00 0.00 1 Car Liter 5.00 5.75 −0.75* 4 5.47 4.93 0.53** 15 Car Quality 5.33 5.40 −0.07 15 5.00 4.00 1.00 4 Beer Price 4.64 5.82 −1.18* 11 5.125 5.125 0.00 8 Beer Quality 5.63 6.00 −0.38 8 4.81 4.18 0.64 11 TV Price 4.56 5.56 −1.00* 9 5.20 5.10 −0.10 10 TV Durability 5.40 5.80 −0.40 10 5.56 4.89 0.67** 9 Restaurant Driving time n.a n.a n.a 0 4.68 4.05 0.63** 19 Restaurant Quality 6.37 6.16 0.21 19 n.a n.a n.a 0
Car Liter 5.80 5.40 0.40 5 4.92 4.62 0.31 13 Car Quality 5.54 5.54 0.00 13 5.40 4.40 1.00* 5 Beer Price 3.90 5.20 −1.30** 10 5.13 4.75 0.38 8 Beer Quality 5.63 5.63 0.00 8 5.10 5.00 0.10 10 TV Price 5.64 5.64 0.00 11 5.00 4.57 −0.43 7 TV Durability 5.00 5.29 −0.29 7 4.72 4.09 0.64 11
Restaurant Driving time 5.00 5.00 0.00 1 4.29 4.11 0.18 17 Restaurant Quality 6.18 6.24 −0.06 17 6.00 5.00 1.00 1 All − 5.46 5.61 −0.16*** 296 4.95 4.59 0.35*** 296 *, **, *** indicate p<0.1, <0.05, <0.01 in the two-tailed paired t test
