Altering Intention to Mudik during COVID-19 Pandemic: A Salient Cue and Simple Reminder Nudge

Abstract

Mudik is a unique exodus in Indonesia where people travel to hometown during Eid festive season. It posed a challenge for the government to cope with the COVID-19 pandemic. Several reports have identified that people still stick with their plan to Mudik amidst the latest Mudik ban. Thus, a scientific nudge campaign is pivotal to influence Mudik behaviour. The present study designed and tested five visual campaigns to identify key elements which can reduce Mudik intention. A pre-post field experiment used six between-subject groups designed on Qualtrics collected 767 responses measuring intention to Mudik. We asked participants respective Mudik ban institution in anticipating the latest nation-wide Mudik ban. Mother visual and a lethal virus for elderly reminder are of which sufficient reducing Mudik intention. The treatment effect holds in certain people conditions. The study yielded an initially worked nudge campaign to support government Mudik ban policy.

Keywords

Nudge Mudik treatment COVID-19 pandemic behaviour change reminder campaign

In 2020, the novel coronavirus (COVID-19) impacted people’s lives worldwide. Anxiety and uncertainty became prominent daily feeling, as the local transmission increased, and no effective vaccine has been confirmed till date.

Travelling, in some places, is strictly prohibited, while in other areas several precautions have been mandated. In Indonesia, the yearly mass travelling (close to an exodus) known as Mudik would be a catastrophe to spread the virus on a nationwide level. Though people are aware of the dangers of travelling, many still attempt to do the Mudik. At the time of national Mudik ban announced, 7% of the national population had travelled to their hometown already, while 24% people were still planning to do so (Ghaliya, 2020). Despite the awareness of a hazard that lies ahead, people still keep making attempts to Mudik.

The pandemic has led to an uncertain situation, where an individual’s optimal decision-making might be challenged. Simon’s bounded rationality (1972) and Kahneman’s dual process theory (2011) have agreed that human beings have limited capacity to process information, and it gets more challenging during this pandemic where fear and anxiety govern individual emotion and cognition (Van Bavel et al., 2020). Interestingly, the uncertainty of whether an individual’s attempt will elicit a desired outcome harms prosocial behaviour (Kappes et al., 2018). The communication of Mudik that harms others welfare might be less salient, making its social importance at that time much more prominent.

Uncertainty causes anxiety, which exhausts the cognitive capacity. When it is scarce, it increases the tendency to become passive (Altmann et al., 2018). People in this situation pay less attention and following others (intuition) seems the most apt way. In this situation of scarce cognitive capacity, their optimal decision-making capability is impaired, and they decide more intuitively (Kahneman, 2011). This mechanism might underlie the reason why people still commit to Mudik; in part because they want to avoid losing their long-established habit.

Mudik has been an annual practice in Indonesian culture for centuries. Society perceives Mudik as the ritual to maintain their traditional roots, take a break and relax in their home village (Yulianto, 2011). It remains an ‘obligation’ across social status, and no ordeal can prevent people from doing so (Soebyakto, 2011; Arribathi & Aini, 2018). Given the perceived moral consequences of Mudik, it has become a social Mudik since, for some people, not doing so will be considered as disrespectful to the parents. The reason is that Mudik ritual is the occasion in which people ask for forgiveness (sungkem) to their parents (Yulianto, 2011).

Mudik has become the status quo and a default routine activity. A default sticks because it lets people act in a routine manner without spending any cognitive resources (Sunstein, 2017); changing the routine means allowing additional attentional resource which is challenging in a situation where anxiety and uncertainty engulf human emotion and cognition. Moreover, a default establishes a reference point where loss or gain is determined (Sunstein, 2017). Even a simple default is difficult to change since people adhere to current status, and any change is costly (Beshears et al., 2009; Benartzi & Thaler, 2007). People tend to stick to current ritual such as Mudik since any change of ritual would cause uncertainty and any uncertainty would raise unappealing risk, even in the case where it is not an optimum decision to choose (Kahneman & Tversky, 2013). The status quo condition and the social norm pressure make changing the Mudik ritual a challenging effort.

Cancelling Mudik might violate the norm and cause a cultural consequence. It is because Mudik is psychologically (i.e., meeting with family as a stress reliever), spiritually (i.e., visiting ancestral tombs) and socially (i.e., communicating success story to the family and locals) rooted in the society (Iriyanto, 2012). Meeting with family becomes one of the crucial experience people seek (Iriany et al., 2019) and enjoying a good time together is a yearly refreshment. Such activity is not exclusively tied to Muslims only but also to the majority of non-Muslims (Yulianto, 2011; Prasetyo & Warsono, 2017) because of the substantial value of togetherness emphasised during the festive. We can see that making a rational decision is challenging under such strong cultural pressure.

During this particular time, daily reporting of infected and fatal cases is the common way to make people aware of the pandemic. However, as argued by Van Bavel et al. (2020), it might not educate the public well on how to comprehend the situation and act accordingly; instead, it creates negative public sentiments. It might further facilitate people’s anxiety and make them unable to think thoroughly and confidently. They further suggested that a message that induces fear will only be successful when the individual feels that it delivers a sense of efficacy as well, if not, then it will lead to individual paranoid response. Public needs a clear and direct message on what to do during this hard time.

Reducing people’s intention to Mudik might require intervention that informs clear benefit or drawback about the consequences following the behaviour. Eventually, society can commit to a prosocial behaviour to keep themselves at home amidst the temptation to Mudik. One way to leverage prosocial behaviour is by informing about a good or bad consequence of an individual’s action in uncertain conditions (Kappes et al., 2018). Thus, it might be beneficial to remind about the risk of further infection (i.e., family in the hometown) just to ensure people remain at home. This reminder is in line with the nudge paradigm (Thaler & Sunstein, 2009) to ensure the clarity of the message and what kind of benefit an individual can get by following the nudged message.

Generally, nudge does not permit or forbid; instead, it gives a subtle push to elicit the intended behaviour (Thaler & Sunstein, 2009). Nudge is fundamental when people face alternatives which (a) do not yield direct benefits, (b) are difficult options and it is uncertain if the outcomes will be favourable or (c) the immediate feedback is not present. One of the strategies to nudge is creating salient features of the message as it ensures a smooth encoding process—since a crucial process of attention is to ensure that stimulus is presented in the individual’s awareness. Salient stimulus stimulates a semantic information processing (Weiten, 2013) and links it to the already-available schema in the mind. Halpern (2015) supported this statement by emphasising that any intervention made should be simple, relevant, resonate with the respective society’s views and well written in order to achieve clarity and certainty. Moreover, a more emotional and personal message gives rise to conscious cognitive processing (Posner et al., 2004) and increase the recall (a cue of a stimulus getting into the processing system).

Nudge intervention may target dual processing System 1, 2 or both. There are debates on which system (1 or 2, see Kahneman, 2011) stimuli need to target to elicit the expected behaviour. For example, when it comes to effectiveness, System 1 might be more favourable. In studies on breaking habits, Hohle (2014) designed intervention nudging meat lovers to choose vegetarian food in a café by presenting the food in an attractive (i.e., attractive name) and salient (i.e., as a default dish of the day) way. However, people still prefer System 2 for nudges, as it maintains people agency (Sunstein, 2016). Interestingly, Renaud and Zimmermann (2019) put this to test where they nudge people to create a strong password by using just visual cues or adding an expiration reminder for the password. These nudges were created to ensure that the participants try to secure their online accounts. They found that the latter targeting both System 1 and 2 nudged more people to make a stronger password.

We can see from the above explanation that in a condition of uncertainty with lots of options and less prompt feedback, nudge approach might be promising to alter people attempt to Mudik by achieving an individual’s feeling of certainty. In particular, the clarity of reminders and warnings seem to add the sense of efficacy by augmenting people’s knowledge and capacities and making relevant facts salient.

The current research aims to investigate whether a similar implication of the nudge approach alters a strong cultural habit. As Sunstein (2016) noted, the principle of nudge is being used to fine-tune the best intervention specific to different contexts. In the present study, the nudge is tested in the unique context of Mudik as a strong cultural habit in Indonesia amidst the pandemic threat. Options of posters were developed involving saliency of fact and a reminder to not Mudik. Also, following the findings by Posner et al. (2004), an emotional and personal message was developed by presenting relevant figure which increases saliency and relevancy.

The present study used a pre-post experiment to measure the intention to Mudik and to see how the posters affect the sense of efficacy. Additionally, it also assessed how the perception of coronavirus (i.e., how dangerous it is perceived to be by the people) and perception of Mudik (i.e., how costly Mudik is—financially and psychologically) changed upon seeing the posters.

Practically, the findings might be valuable for the authority to apply the nudge approach in redirecting people behaviour, particularly amidst the uncertainty caused by the COVID-19 pandemic or any other plague alike.

Method

Data collection conducted from 23 March to 16 April 2020 during which only people working in government institutions were forbidden to Mudik (government-issued national Mudik ban on 21 April). This different participant condition might be a problem in drawing the experimental effect. Thus, we asked whether any Mudik ban occurred in the institution to which participants were affiliated (applicable vs not applicable, ‘Is there any Mudik ban applicable in your environment/institution/organisation’). This way, we were also able to compare the effectiveness of the posters between those two conditions.

Participants

Data collection included 767 respondents (initially, there was 778, of which 11 were excluded due to incomplete submission) joining via online pre-post field experiment designed on Qualtrics.

Of the total respondents, 52% were female, married (49%) and Muslim (65%). The latter characteristics reflect the actual country proportion for faith believers. Most respondents (49%) were born between 1981 and 2000, who are considered millennials (Hammill, 2017), followed by Gen Z (26%) and Xers (24%). Mudik ban was applicable for 59% of participants’ institutions across the groups.

Table 1.

Demographical Characteristics of Respondents in Each Group Condition.

	T1(n = 137)		T2(n = 115)		T3(n = 120)		T4(n = 142)		T5(n = 131)		T6(n = 122)
	n	%	n	%	n	%	n	%	n	%	n	%
Parent: Father
Alive	98	72	80	70	96	80	110	77	93	71	100	82
Not alive/away	39	28	35	30	24	20	32	23	38	29	22	18
Parent: Mother
Alive	121	88	101	88	107	89	123	87	113	86	110	90
Not alive	16	12	14	12	13	11	19	13	18	14	12	10
Mudik status
Mudik banned	83	61	78	68	68	57	84	59	67	51	76	62
Mudik not banned	54	39	37	32	52	43	58	41	64	49	46	38
Current residence
Banten	11	8	9	8	7	6	13	9	12	9	6	5
Bali & Nusa Tenggara	9	7	1	1	7	6	8	6	5	4	6	5
DIY Yogyakarta	2	1	3	3	2	2	7	5	3	2	2	2
DKI Jakarta	26	19	25	22	29	24	32	23	25	19	20	16
West Java	27	20	21	18	15	13	16	11	24	18	28	23
Central Java	9	7	1	1	3	3	6	4	2	2	1	1
East Java	40	29	37	32	48	40	44	31	40	31	32	26
Kalimantan	3	2	8	7	5	4	8	6	3	2	5	4
Sulawesi & Maluku	3	2	4	3	2	2	2	1	4	3	5	4
Sumatera	7	5	6	5	2	2	6	4	13	10	16	13
Origins
Banten	4	3	1	1	1	1	0	0	2	2	2	2
Bali & Nusa Tenggara	13	9	2	2	6	5	9	6	8	6	6	5
DIY Yogyakarta	9	7	7	6	7	6	13	9	9	7	4	3
DKI Jakarta	7	5	12	10	13	11	8	6	5	4	9	7
West Java	19	14	16	14	8	7	21	15	23	18	27	22
Central Java	25	18	18	16	20	17	27	19	14	11	11	9
East Java	41	30	42	37	40	33	42	30	46	35	39	32
Kalimantan	3	2	7	6	9	8	7	5	3	2	5	4
Sulawesi & Maluku	5	4	4	3	6	5	3	2	3	2	4	3
Sumatera	11	8	6	5	9	8	11	8	17	13	14	11
Latest education
High school	23	17	27	23	31	26	25	18	26	20	25	20
Undergraduate	74	54	58	50	70	58	73	51	71	54	52	43
Postgraduate	40	29	30	26	19	16	44	31	34	26	45	37

Source: Authors.

Research Design

The present study used a pre-post field experiment and a between-subject design. The three dependent variables (intention to Mudik, perception of coronavirus, perception of Mudik) were assessed before and after the poster exposure (in T6, these variables were reassessed amidst the absence of poster exposure).

The survey was designed on Qualtrics and sent through Indonesia Behavioural Economics Forum (IBEF) and other student and Indonesian online networks. The experiment was done anonymously, voluntarily and no reward was given. The randomisation designed on the Qualtrics divided participants into six groups. Each group was exposed to only one poster. One group served as a control group and, thus, no poster exposure was given. Table 1 showed the number of participants and demographic characteristics in each group.

Stimulus

Five posters were developed manipulating three nudges: personal message (language, message), saliency (virus visual, parent figure), reminder (reminder text). Table 2 and Figure 1 shows how the manipulation was designed.

Table 2.

The Presence (Manipulation) of the Experimental Variables.

	Personal message		Saliency		Reminder
	Language	Message	Virus visual	Parent figure	Reminder Text
Group 1 (T1)	Bahasa	Short	–	Mother	–
Group 2 (T2)	Bahasa	Long	–	Mother	–
Group 3 (T3)	Bahasa	Short	–	Mother	Present
Group 4 (T4)	Bahasa	Short	Present	Father	–
Group 5 (T5)	Local	Short	Present	Mother	–
Group 6 (T6) – control	–	–	–	–	–

Source: Authors.

All posters used mother figure (except T4), COVID-19 visual and typical virus transmission condition (except T2) and personal message (except T2). The personal message (long and short) states parent’s advice, feeling and acceptance during the virus pandemic (refer to Appendix 1 for the English translation). Reminder text was only present in T3. In T4 and T5, a virus visual was added. Finally, in T5, local languages (those which respondents use most often in everyday communication: Bahasa/Javanese/Sundanese/Madura/Batak) replaced Bahasa language. Immediately before the poster exposure participants were asked to choose which language they use most often.

Figure 1.

Source: Authors.

Procedure

The experiment was conducted in three parts (see Figure 2 for detail experimental procedure). The Qualtrics randomly assigned the respondents to six different groups which ensured that all demographic characteristics were distributed evenly across the groups.

In the first part, the participants were presented with a random number between 1 to 6 on their screen. Each number had the same probability of appearing. After seeing the number, they were asked to select a number that corresponds to the number that appeared previously. Later, this number determines the poster the participant would see (i.e., if they see number 1, they would be allocated in Group 1 and see poster T1). This procedure was adopted to randomly allocate participants across the six groups.

Then the participants were asked statements on intention to Mudik (‘I will travel home (Mudik) in this upcoming Lebaran 2020’), perception of coronavirus (‘coronavirus is dangerous’), and perceived cost of Mudik (‘Mudik is so tough—far and expensive’). The response was recorded using 4-point Likert-type format, where 1 stands for ‘strongly disagree’ and 4 for ‘strongly agree’. Once they finished, they needed to click the next button and were shown the poster (see Figure 1). Subsequently, participants were reassessed on the same initial three statements. In the last sections, demographic questions, questions regarding Mudik status (banned vs not banned) and motivation to Mudik were presented Mudik.

Figure 2.

Source: Authors.

Result

All data were managed in Microsoft Excel, and statistical analyses was done using R. Before any poster exposure, participants had shown their reluctancy to Mudik indicated by their average mean of below 2 (out of 4, the higher the score, the more they were willing to Mudik). There was also no difference of Mudik intention between people working in an institution which forbade and did not forbade their employees to Mudik. Peoples’ intention to Mudik was slightly higher for those whose mothers were still alive compared to those whose had passed away. Table 3 showcased these means briefly.

Table 3.

The Initial Intention to Mudik Before Treatment Exposure Across the Group Condition.

	Intention to Mudik
	n	M	SD
Mudik status
Mudik banned	456	1.82	0.79
Mudik not banned	311	1.86	0.83
Parent: Father
Alive	577	1.83	0.80
Not alive	190	1.87	0.83
Parent: Mother
Alive	675	1.85	0.82
Not alive	92	1.71	0.73
Religion
Muslim	497	1.92	0.83
Non-Muslim	270	1.69	0.74

Source: Authors.

Note: Intention to Mudik was measured by response rating to what extent does the respondent agree with the statement ‘I will travel home (Mudik) in this upcoming Lebaran 2020. The response was recorded using 4-point Likert-type format, where 1 is for ‘strongly disagree’ and 4 for ‘strongly agree’.

However, all participants agreed that gathering around with families were the reason they might undergo Mudik, despite the pandemic condition. Averagely, all scored above 3 (for ‘important’) for these particular reasons (see Table 4).

Table 4.

The Difference in Mudik Motivation Regardless of Treatment Condition.

	Motivation to Mudik
	M	SD
To meet mother	3.37	0.80
To meet father	3.22	0.87
To meet relatives	3.13	0.76
Miss the hometown	2.68	0.87
To meet friends	2.60	0.79
For vacation	2.51	0.93

Source: Authors.

Note: Motivation to Mudik was measured by response rating to what extent does the respondent considers each of item important ‘The reason I will travel home is to meet father/mother/relatives/friends/miss the hometown/for vacation’. The response was recorded using 4-point Likert-type format, where 1 is for ‘strongly not important’, and 4 for ‘strongly important’.

These initial numbers showed that Mudik was an important activity where all related to connecting with families/relatives might some of its notable triggers.

Intention to Mudik

Our primary research question was that which treatment has the best effect in reducing Mudik intention. Figure 3 showed how the six groups differed.

Figure 3.

Source: Authors.

Table 5.

Poster Effectiveness in Reducing Intention to Mudik.

	Pre-exposure		Post-exposure		t(df)	p	95% CI
Group	M	SD	M	SD	LL	UL
T1	1,89	0,81	1,78	0,82	2,05 (136)	.04	0.00	0.22
T2	1,83	0,77	1,77	0,79	1,11 (114)	.27	–0.05	0.19
T3	1,89	0,88	1,78	0,81	2,38 (119)	.02	0.18	0.20
T4	1,85	0,81	1,77	0,82	1,25 (141)	.21	–0.04	0.18
T5	1,77	0,77	1,73	0,72	0,76 (130)	.45	–0.07	0.16
T6	1,79	0,82	1,73	0,73	1,54 (121)	.13	–0.02	0.13

Source: Authors.

Notes. CI: confidence interval; LL: lower limit; UL: upper limit.

Using paired t-test analysis, exposure of T1 and T3 significantly reduced intention to Mudik. On average, the respondents in T1 showed a significantly reduced Mudik intention from before (M = 1.89, SD = 0.81) to after (M = 1.78, SD = 0.82), t(136) = 2.05, p < .05, r = .17, 95% CI [0, 0.22] seeing the poster. Similarly, T3 also significantly changed the intention prior to exposure (M = 1.89, SD = 0.88) and after seeing the poster (M = 1.78, SD = 0.81), t(119) = 2.38, p < .05, r = .21, 95% CI [0.02, 0.20]. The higher effect size (r value) indicates T3’s better capability to alter the behaviour, though both values still indicate a small effect (Cohen, 1992).

Furthermore, we investigated how the treatment effect of T1 and T3 influence a specific group of participants. Tables 6 and 7 show which group of people showed reduced Mudik intention after T1 and T3 exposure.

Table 6.

Change of Intention to Mudik in Subgroups of Group 1 (T1).

	Pre-exposure		Post-exposure		t(df)	p	95% CI
	M	SD	M	SD	LL	UL
Intention to Mudik … group
In Muslim	2.02	0.88	1.86	0.85	2.40 (86)	.02	0.03	0.29
In non-Muslim	1,66	0,63	1,64	0,75	0.23 (49)	.82	–0.16	0.20
In no-ban Mudik	1,98	0,79	1,98	0,86	0 (53)	1	–0.13	0.13
In ban-Mudik	1,83	0,82	1,65	0,77	2.35 (82)	.02	0.03	0.33
In male	1,94	0,92	1,94	0,94	0 (68)	1	–0.11	0.11
In female	1,84	0,68	1,62	0,65	2.43 (67)	.02	0.04	0.40
In low education	1,91	0,67	1,87	0,69	0.27 (22)	.79	–0.29	0.38
In high education	1,89	0,84	1,76	0,84	2.20 (113)	.03	0.01	0.23

Source: Authors.

Note: Lower-educational group is high school and below. The higher-educational group is above high school.

Table 7.

Change of Intention to Mudik in Subgroups of Group 3 (T3).

	Pre-exposure		Post-exposure		t(df)	p	95% CI
	M	SD	M	SD	LL	UL
Intention to Mudik … group
In Muslim	2,06	0,87	1,90	0,78	2,26 (69)	.02	0.02	0.30
In non-Muslim	1,66	0,85	1,62	0,83	0,81 (49)	.42	–0.06	0.14
In no-ban Mudik	1,94	0,92	1,87	0,86	1,16 (51)	.25	–0.06	0.21
In ban-Mudik	1,85	0,85	1,72	0,77	2,12 (67)	.04	0.01	0.26
In male	1,96	1	1,81	0,88	1,84 (56)	.07	–0.01	0.33
In female	1,83	0,75	1,76	0,76	1,66 (62)	.10	–0.01	0.14
In low education	1,87	0,92	1,77	0,88	1,36 (30)	.18	–0.05	0.24
In high education	1,9	0,87	1,79	0,84	2 (88)	.05	0.00	0.22

Source: Authors.

Note: Lower-educational group is high school and below. The higher-educational group is above high school.

We found here that the treatment effect of T1 and T3 only hold for some groups of participants. For example, both treatments did not affect the Mudik intention of the non-Muslim group, no-ban group, male group and lower educational group. Equally important, we found that no treatment could lower the perception of Mudik intention of those participants who have lower education (did not go to college). Practically speaking this group of people is one of the majorities of Indonesian who had gone to Mudik (Mulyanto, 2020).

Interestingly, in the two conditions (T1 and T3), there is no indication that such reduced intention to Mudik occurred in an institution where Mudik ban is not applicable. The 0 t-value (and the respective p-value) in the no-ban group (T1 condition, Table 6), for example, further showed that the exposure did not change the intention at all, which is also confirmed by both pre- and post-exposure similar mean values. The 0 value is also observed in the male group but not in the female group, indicating its significance only in the female group amidst the absence of reminder message (see the Method section). Higher female sensitivity to an emotional message could be associated with this finding.

Eventually, for T3, we computed the t-test for ‘intention to Mudik’ and filtered the response by the motivation to Mudik and the both parent’s condition in the hometown (alive or not). Table 8 presents the significantly reduced Mudik intention when meeting both the parents is important as well as when the mother is still in good health and present in the hometown. All the three significant effects (exclude ‘when father is still alive’) have medium-size effects (r ≥ .3).

Table 8.

Change of Intention to Mudik in T3 Condition Filtered by the Motivation to Mudik and Parent Condition.

	Pre-exposure		Post-exposure		t(df)	p	95% CI
Condition	M	SD	M	SD	LL	UL
Intention to Mudik
When to meet father is important	2.08	0.99	1.86	0.93	2.85 (49)	< .01	0.06	0.38
When to meet mother is important	2.07	0.99	1.88	0.96	2.65 (57)	.01	0.05	0.33
When father is still alive	1.91	0.87	1.81	0.79	1.90 (95)	.06	–0.00	0.19
When mother is still alive	1.92	0.87	1.79	0.8	2.63(106)	< .01	0.03	0.23

Source: Authors.

Perception of Corona

All mean scores were very close to 4, meaning that people believed in the hazard of corona even before the exposure. As we can see from Table 9, any posters exposure did not change how hazardous people think the virus is. The mean score before and after the poster exposure support these insignificant effects with all t-values nearly close to zero.

Table 9.

Perception of Corona in Each Group Condition.

	Pre-exposure		Post-exposure		t(df)	p	95% CI
Group	M	SD	M	SD	LL	UL
T1	3.47	0.69	3.50	0.64	–0.85 (136)	.40	–0.15	0.06
T2	3.49	0.71	3.55	0.6	–1.26 (114)	.21	–0.16	0.04
T3	3.51	0.65	3.50	0.65	0.16 (119)	.87	–0.09	0.11
T4	3.56	0.55	3.57	0.61	–0.19 (141)	.85	–0.08	0.07
T5	3.61	0.6	3.61	0.63	0 (130)	1	–0.09	0.09
T6	3.37	0.77	3.35	0.76	0.40 (121)	.70	–0.07	0.10

Source: Authors.

Perception of Mudik

Table 10 showed the t coefficient in negative value, which means that the mean value increased after the poster exposure. Thus, the posters seemed to enhance the cost of Mudik; however, only T3 group shows significant effect. It made the perception of Mudik becomes more challenging to do, t(119) = –2, p ≤ .05, r = 0.18, 95% CI –0.18, 0]. Additionally, the significant effect remained particularly for those having lower education (high school and below) (pre-exposure, M = 2.35, SD = 0.8; post-exposure, M = 2.55, SD = 0.85) with medium effect, t(30) = –2.26, p < .05, r = 0.38, 95% CI –0.37, –0.02].

Table 10.

Change in Perception of Mudik in Each Group Condition.

	Pre–exposure		Post–exposure		t(df)	p	95% CI
Group	M	SD	M	SD	LL	UL
T1	2.18	0.80	2.20	0.83	–0.43(136)	.67	–0.08	0.05
T2	2.31	0.72	2.27	0.79	0.84(114)	.40	–0.06	0.15
T3	2.26	0.78	2.35	0.84	–2.00(119)	.05	–0.18	–0.00
T4	2.22	0.72	2.20	0.76	0.47(141)	.64	–0.05	0.07
T5	2.27	0.83	2.31	0.90	–1.23(130)	.22	–0.12	0.03
T6	2.11	0.80	2.16	0.85	–1.10(121)	.28	–0.14	0.04

Source: Authors.

Discussion

Before starting any discussion, readers need to know that the time the article was written, the government has issued a national ban. However, as we noted earlier (see Method section), we prompted this situation when collecting the data by asking whether any currently applicable Mudik ban occurred at the time of participating in the survey. In fact, the initial intention to Mudik was not that different between ‘Mudik is banned’ and ‘Mudik is not banned’ (1.82 and 1.86 respectively, see Table 3).

The present study aimed at supporting the behaviour change strategy for a strong cultural habit of Mudik. By manipulating visual saliency, personal message and reminder, we developed five posters aiming to reduce the intention to Mudik, which could lead to catastrophe amidst the global outbreak. The next sections explain how the posters reduced the intention to Mudik and discuss its similarity with previous research.

The present study showed that by manipulating visual saliency combined with personal message and reminder, one might reconsider their plan to Mudik despite it being a strong and important cultural habit. Table 4 facilitates our understanding of importance of Mudik and gathering with families during the Eid season. Among the five developed visuals, T3 seems the most promising poster to alter the Mudik intention. Aside from having personal message and saliency, T3 was embedded with reminder text to amplify the learning of the conscious thought of the hazards of elderly susceptibility to transmission.

The close-up mother visual in the poster is vivid and salient. Regardless being single or married (shown by the two highest proportion of demographic status, see Table 1) depiction of mother here represents a respected significant other, that any individual respects and obeys in a typical collectivistic culture (Grotevant, 1998, in Tamis-LeMonda et al., 2008). The visual was also complemented with a personal message amplifying the depiction of mother as a figure, one needs to consider and respect. The reminder text ‘elderlies are more susceptible to infection’, printed in red, serves as a clear and direct reminder. Sunstein (2016) argued that reminder is a powerful nudge tool to intensify the already available knowledge by making relevant event prominent. In line with this argument, the text intensifies the virus risk for the elderly and that they are the most vulnerable group. The visuals emphasise the impact of virus spread to the loved ones if the targeted behaviour (that is, to not Mudik during the COVID-19 pandemic) is not achieved.

By using personal message, visual saliency and reminder text, theoretically, T3 has used System 1 and System 2 to design an impactful nudge approach. Since it was introduced, nudge has been heavily known to change behaviour by taking advantage of heuristics and rules of thumb (Thaler & Sunstein, 2009), which are the products of System 1 thinking. However, the effectiveness of using Systems 1 and 2 to change behaviour have become an interesting debate. By using saliency and default, Hohle (2014) succeeded to alter the habit of meat lovers to eat more vegetables. Similarly, Thorndike et al., (2014) created a simple nudge intervention in which they put a colour code of green (healthy), yellow (less healthy), and red (unhealthy) to sets of food sold in a cafeteria, resulting in the increase of sales of healthy items (from 41% to 46%) and decrease of sales of red items (from 24% to 20%) in two years. All these behaviours are redirected by designing a System 1 thinking nudge.

On the other hand, Sunstein (2016), for example, argued that people might not want to be entirely steered and prefer to preserve their preference, particularly amidst the uncertainty and anxiety situation. Moreover, in US society, he showed that such a mechanism is more salient in the case of health and environmental protection. Thus, System 2 might work better as it lets people maintain their personal agency. The effectiveness of reminder (which mostly play around in activating System 2 thinking) could be linked with this rationale. Our results are supported by the findings of Renaud and Zimmermann (2019) who found that adding a reminder of the urgency to create strong password (along with visual cues to create one, i.e., long dachshund, pair of eyes) yielded more people to make strong password. However, as noted by Renaud and Zimmermann (2019), it was difficult to assess each variable’s effect individually. We can conclude in this study that, amidst the strong cultural habit and uncertainty caused by the pandemic, the intention to Mudik could be reduced by using multiple nudge tools and the combined thinking of both systems.

Measuring intention could be one source of bias. It is widely known that intention is not always followed by behaviour and eliciting behaviour might require habit (Triandis, 1989) and external factors (facilitating condition). The intention could be influenced by perceived value and subjective norms (Ajzen, 1991), something that takes into account the work of System 2. However, many of our behaviours are the result of System 1 working alone (Kahneman, 2011).

Aside from how far the stimulus has been developed by merging both systems, it is interesting to reassess the effectiveness of the treatment by measuring actual behaviour. That way, we can be sure that the reflection that might occur (the work of System 2) end up as an actual behaviour—something that needs to confirm for readers aiming to extend the findings for future research.

These considerations might be in line with the observed effect size value (of most the T3 significant effect) which in average fall in medium effect size range (r ≥ 0.3, see Cohen, 1992). Possibly, directly observing the behaviour (compared to self-report) could contrast the difference one might elicit between pre- and post-poster exposure; and, thus, increase the effect size.

Also, we would like to note that the time and place of data collection could have an initial influence on people having a low intention to Mudik before any treatment exposure (see Table 2) regardless of the presence of Mudik ban. Furthermore, if we look at the changes in mean values, we can see that the presence of the advert further lowers down the intention. Thus, the stimulus not necessarily alters the behaviour; however, it could magnify and intensify the reluctance to travel home (Mudik). We would like to note as well in reading the results, readers should pay attention to the number of cases particularly when we want to know the treatment effect on specific group (e.g., people with low education background, people in no-ban Mudik situation). It should also be considered that the analysis was run using filtered response which reduced the number of cases. A low number of observations is an issue when we want to get a robust and more generalisable findings as it will not reflect the shape of normal distribution (Field et al., 2012, p. 47). In the current study, the treatment effect analysis on specific group included a relatively small number of observations (20–50 participants in each group). Future research focusing on these specific sections of the society needs to concentrate on acquiring responses from them.

Finally, a nudge could be promising in altering people’s habit, particularly when the presence of a strong culture hinders people from changing. In the middle of uncertainty and anxiety, communicating policy using the nudge approach is promising. In particular, using the combination of System 1 and System 2 nudges could be more impactful compared to using System 1 alone. Practically, our study also contributes to the accumulating implications of nudge approach in public policy and how its effectiveness and implication can have a positive impact on people’s life. It showed how a simple reminder with visually salient images and personal message could be beneficial to subtly influence people to achieve a common goal, particularly during the current COVID-19 pandemic.

Footnotes

Appendix 1 Declaration of Conflicting Interests

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

References

Ajzen

(1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179–211.

Altmann

S.,

Grunewald

A., &

Radbruch

(2018). Cognitive foundations of passive choices [Mimeo].

Arribathi

A. H.,

& Aini

(2018). Mudik Dalam Perspektif Budaya Dan Agama (Kajian Realistis Perilaku Sumber Daya Manusia) [Mudik from the lens of cultural and religion perspectives (study of human resource)]. Cyberpreneurship Innovative and Creative Exact and Social Science, 4(1), 45–52.

Benartzi

S.,

& Thaler

(2007). Heuristics and biases in retirement savings behavior. Journal of Economic Perspectives, 21(3), 81–104.

Beshears

J.,

Choi

J. J.,

Laibson

D., &

Madrian

B. C.

(2009). The importance of default options for retirement saving outcomes: Evidence from the United States. In Social security policy in a changing environment (pp. 167–195). University of Chicago Press.

Cohen

(1992). A power primer. Psychological Bulletin, 112(1), 155.

Ghaliya

(2020 April 21). ‘It’s too late’: Lawmakers criticize timing of Jokowi’s ‘Mudik’ ban. The Jakarta Post .

Field

, Miles

J.,

& Field

(2012). Discovering statistics using R. SAGE Publications.

Halpern

(2015). Inside the nudge unit.WH Allen.

10.

Hammill

(2017). Mixing and managing four generations of employees. FDU Magazine Online, Winter/Spring 2005 .

11.

Hohle

S. M.

(2014). Nudging sustainable food choices. The role of defaults, frames, habits and nature relatedness (Master’s thesis). University of Oslo.

12.

Iriany

I. S.,

Pasciana

R., &

Ramdhani

(2019). Eid homecoming ‘Mudik’ tradition as a conventional pattern in the global era. Journal of Advanced Research in Social Sciences and Humanities , 4(3).

13.

Iriyanto

A. M.

(2012). Mudik dan keretakan budaya. Humanika , 15(9).

14.

Kahneman

(2011). Thinking fast and slow. Penguin.

15.

Kahneman

D., &

Tversky

(2013). Choices, values, and frames. In Handbook of the fundamentals of financial decision making: Part I (pp. 269–278).

16.

Kappes

, Nussberger

A. M.

, Faber

N. S.

, Kahane

, Savulescu

, & Crockett

M. J.

(2018). Uncertainty about the impact of social decisions increases prosocial behaviour. Nature Human Behaviour, 2(8), 573–580.

17.

Mulyanto

(2020 April 21). Indonesia bans end of Ramadan exodus amid coronavirus pandemic. Aljazeera .

18.

Posner

M. I.

, Snyder

C. R.

, & Solso

(2004). Attention and cognitive control. Cognitive Psychology: Key Readings , 205.

19.

Prasetyo

, & Warsono

(2017 October). Specific migration in Indonesia Mudik and Balik Lebaran. In Social sciences, humanities and economics conference (SoSHEC 2017). Atlantis Press.

20.

Renaud

, & Zimmermann

(2019). Nudging folks towards stronger password choices: Providing certainty is the key. Behavioural Public Policy, 3(2), 228–258.

21.

Simon

H. A.

(1972). Theories of bounded rationality. Decision and Organization, 1(1), 161–176.

22.

Soebyakto

B. B.

(2011). Mudik Lebaran: studi kualitatif. [Mudik during Ied day: A qualitative study] Jurnal Ekonomi Pembangunan [Journal of Developmental Economy], 9(2), 62–67.

23.

Suherdjoko.

(2020 March 26). COVID-19: ‘Mudik’ begins in Java while government mulls over travel ban. The Jakarta Post .

24.

Sunstein

C. R.

(2016). People prefer system 2 nudges (kind of). Duke LJ, 66, 121.

25.

Sunstein

C. R.

(2017). Default rules are better than active choosing (often). Trends in Cognitive Sciences, 21(8), 600–606.

26.

Tamis-LeMonda

C. S.

, Way

, Hughes

, Yoshikawa

, Kalman

R. K.

, & Niwa

E. Y.

(2008). Parents’ goals for children: The dynamic coexistence of individualism and collectivism in cultures and individuals. Social Development, 17(1), 183–209.

27.

Thaler

R. H.

, & Sunstein

C. R.

(2009). Nudge: Improving decisions about health, wealth, and happiness. Penguin.

28.

Thorndike

A. N.

, Riis

, Sonnenberg

L. M.

, & Levy

D. E.

(2014). Traffic-light labels and choice architecture: Promoting healthy food choices. American Journal of Preventive Medicine, 46(2), 143–149.

29.

Triandis

H. C.

(1989). The self and social behavior in differing cultural contexts. Psychological Review, 96(3), 506.

30.

Van Bavel

J.J.

, Baicker

, Boggio

, Capraro

, Cichocka

, Cikara

, Crockett

M.J.

, Crum

A. J.

, Douglas

K. M.

, Druckman

J. N.

, Drury

(2020). Using social and behavioural science to support COVID-19 pandemic response. Nature Human Behaviour, 4(5), 460–471.

31.

Yulianto

V. I.

(2011). Is the past another country? A case study of rural-urban affinity on Mudik Lebaran in Middle Java. Journal of Indonesian Social Sciences and Humanities, 4, 49–66.

32.

Weiten

(2013). Psychology: Themes and variations. Wadsworth.