The main purpose of this research was to see whether age and attention influence social decision making. There was previous research on how age could potentially influence social decisions however this was conducted in the previous decade and the aspect of attention had not been involved. To investigate this effect, an online experiment based on the social card game by Fredrickson & Carstensen (1990) was conducted, with the addition of hidden information boxes which enabled the tracking of the participants’ attention throughout the experiment.
Participants were divided into one of two groups: young adults or older adults and could choose between three choices about who to spend time with. These choices included a familiar category and two novel categories across two different scenarios, an open-ended scenario and a time-limited scenario. Further, half of the rounds only contained information about the person they could spend time with, and the other half included additional
information associated with that specific person.
Exploring how the different age groups made decisions through the whole
experiment, we found that most of the participants chose the familiar option across most of the scenarios (Table 1). This was the case for both age groups and for the person open-ended, person time-limited and additional info time-limited scenarios. The trend towards choosing the familiar option was similar in these scenarios and not significantly different across age groups. However, in the final scenario, additional info open-ended, young adults mostly went for the familiar option while the majority of the older adults went for the novel option. There was a significant difference at a 5% level in both a chi-square test and logistic regressions, with gender and additional variables being controlled for. This indicates that there was an association between the age of the participant and the choice they made. This is an interesting result that differs from previous literature that found the opposite result for the open-ended scenario with younger adults making more novel choices compared to older adults
(Fredrickson & Carstensen, 1990; Carstensen, Isaacowitz & Charles, 1999; Carstensen, 2006, Sharot & Sunstein, 2020). It is also contradictory to what we expected and was stated in the first hypothesis, that older adults would go for the choice with an emotional goal more compared to the younger adults in the open-ended scenario. One possible explanation could be due to the current social environment we are in regarding COVID-19 pandemic. The COVID-19 pandemic has put social contact into perspective, and it could have influenced how people value their social contact compared to the social environment in which the previous experiments were conducted. Additionally, it could also be due to specifics of the experimental design. In the final scenario, participants had been introduced to both scenarios and had made multiple social decisions in the previous scenarios so had thought carefully about who they would want to spend time. In addition, the additional information regarding the person they could spend time with may have impacted the emotional vs. information-seeking motivations. In the open-ended scenarios, which were scenario 1 and 4, we see that the additional information did change the majority for the older adults to the novel choice. In the time-limited scenarios, which were scenario 2 and 3, the majority for both age groups was still the familiar choice however, especially for the older adults, we do see a large group of participants change their preference to the novel choice. For the older adults the familiar choice was chosen 97% of the time in the second scenario while in the third scenario the familiar choice was only chosen 56%. Looking at figure 9 and 10 this change also becomes visually evident. The main difference between these two scenarios is that in the third scenario additional information was added. Even though the result was not significant for the third scenario it does seem like choice was influenced to some extent. Based on the results the first hypothesis is rejected as we did not find a significant result for the person open-ended
scenario and in fact found the opposite for the additional info open-ended scenario. In the
additional info scenario, the older adults made more novel decisions in the open-ended context compared to the younger adults. For the second hypothesis we did not find a
significant result for either of the time-limited scenarios. However, it is clear from the results, even though not significantly different across age groups, that familiar was the preferred option here regardless of age.
Besides looking at how age influenced the choice of the participant we also looked at how age influenced the attention of the participant. The hypotheses for this were: the third hypothesis, that we expect older participants attention to be more towards the familiar choices, and the fourth hypothesis, that we expect older participants to focus on the person boxes and younger participants to focus on the information boxes. For attention the focus was on the frequency of attending the different boxes of information as the results of this were more reliable compared to dwell time, which can be found in Appendix B. The first two scenarios only presented boxes of information regarding the people they could choose, so the frequency of the familiar option and the frequency of the novel familiar options are
measured. Therefore, the overall conclusion for the first two scenarios is that age did not have a significant effect on the participants attention (table 6, table 7). In the third and fourth scenario participants were given additional information regarding the people they could choose between, namely learning value, that was also hidden under boxes. For both
scenarios, no significant results were found regarding the effect of age on attention. There is not one box of information that received more attention than the other. We do see in figure 12-15 that both young and older adults paid more attention to the novel boxes compared to the familiar boxes. We can therefore conclude that age does not have a significant influence on where the attention of the participant went, and we reject the third hypothesis. Similarly, there is no clear difference between whether young or old participants looked more at information regarding the person of additional information as the mean ranks for these variables are all very similar, which in turn rejects the fourth hypothesis.
Lastly, we investigated the effect attention had on the choice of the participant.
According to the fifth hypothesis we expect that more attention towards the specific boxes will guide the participant into choosing that option. In the first scenario, person open-ended, we found highly significant results at the 1% level for both the familiar and novel attention variable in predicting choice (table 11). Specifically, attention towards the familiar box influences the choice toward the familiar option, and attention towards the novel boxes significantly led towards the novel choice. In model 2, adding age and gender showed that male participants were significantly more likely to choose the novel option. That attention has influenced the choice is in line with previous literature what attention is a key factor that influences choice (Orquin and Mueller Loose, 2013;Krajbich, Armel & Rangel, 2010;
Krajbich, Camerer & Rangel, 2012)). In the second scenario, we see the same pattern as the first scenario but is not significant across models.
Proceeding to the third scenario. In this scenario there are additional variables
regarding attention to the extra information provided for each of the people they could choose between. In model 1, that only contains the attentional variables, we see that the only
significant variable is the one regarding the extra information of the familiar option. As the variable has a negative beta which would suggest that attention to this box of information significantly leads to the participant going for the familiar option, as the familiar choice is the baseline. The familiar person box also has a negative beta however is not significant. The novel options, both person and information, have positive betas suggesting that the choice would be directed towards the novel option, but this is not significant. What is interesting to see is that in model 2, where the variables age group and gender have been added, the variable regarding attention to the novel information becomes significant. This suggests that when the demographics of the participants are controlled, the attention towards the novel
information does significantly direct the choice of the participant towards the novel choice. It is interesting to see that only the additional info is significant and not the person boxes. This could be because this is the first scenario that has additional information, and the participant is more curious as to what additional information each of the people have. Also, seeing as attention towards these boxes is significant, and therefore had an effect on the choice, the participants must have found this information relevant when deciding who to spend time with. Attention to the novel information was not significant in the first model but did become significant when control variables age and gender were added. This would suggest that, in general, attention towards novel information did not have an effect on the choice but there must be a significant difference in how either age or gender paid attention to these boxes as when these variables are added, it does become significant. As older participants and male participants are the alternatives in both these variables, it would suggest that these
characteristics significantly influenced the attention to these boxes. However, these variables are not significant in the model as a whole. Another intriguing aspect of this scenario is when the attentional variable regarding time is added to this regression, as can be seen in table 13 model 3, the age group of the participant also becomes significant and so does the time spent on the novel additional information. For age group the beta is negative and marginally
significant at the 10% level suggesting that if the participant is male, they are steered towards the more familiar option in a social scenario, which is the opposite of what we saw in
scenario 1. The difference between these scenarios is the additional information in scenario 3 and the fact that scenario 1 is open-ended while scenario 3 is time-limited. This would suggest that male participants go more for the familiar option in time-limited scenarios and more novel options in open-ended scenarios. When looking back at table 2 and table 4 and the logistic record performed with the focus on the effect of age on choice, we see a
significant similar effect in the first scenario however not in the third scenario. Therefore, in scenario 1 it seems that gender significantly guides the choice here on its own, yet in scenario 3 does need the additional variable of attention in the regression to become significant. In the final scenario, additional info open-ended, we once again see that the attention variables regarding the familiar option are both negative, therefore leaning towards the familiar option, and the attention variables regarding the novel option are both positive, therefore leaning towards the novel option. However, in all the models none of the variables are significant implying that attention did not significantly affect or play an important role in the eventual choice of the participant. So overall there are mixed results. In the person open-ended and additional info time-limited we found significant effects for attention on choice. In the person time-limited and the additional info open-ended we did not find significant results. According to the fifth hypothesis we expected that attention towards the familiar boxes will lead to choosing the familiar person and similarly for attention towards the novel boxes will lead to choosing the novel person. In scenario 1 and 3 this would be the case however in scenario 2 and 4 it is not. It is therefore not possible to conclude that this is always the case and we have to reject the hypothesis.