A slice of hygiene: justification and consequence in the persuasiveness of prescriptive and proscriptive signs

Abstract

Studies concerning sign effectiveness highlight the use of justifications and information about consequences of non-compliance. Assuming that the most persuasive messages contain both types of information, we compared the effectiveness of eight different signs (that encouraged supermarket shoppers to hygienically handle bread). The independent variables were (a) sign content (justification vs. consequence vs. justification and consequence vs. neutral) and (b) form of request (prescriptive vs. proscriptive). After assessing the subjective level of the signs’ persuasiveness through a survey (N = 240), we observed people’s compliance in a natural experiment (N = 1.440). Signs containing both a justification and information about consequences achieved the highest rating and actual compliance.

Keywords:

• Sign
• persuasion
• norms
• justification
• consequence

Introduction

To communicate established social norms, legislators and private owners commonly use regulatory and warning signs due to the low costs of their application (Dwyer et al., 1993; Geller, Winnett, & Everett, 1982; Stern & Oskamp, 1987). For example, posting a ‘no littering’ sign may be less expensive than installing additional trash cans in parks. But are such interventions effective?

According to Goal-framing theory (Lindenberg & Steg, 2007) the influence of norms is determined by three goals: hedonic, gain, and normative. Each goal can evoke certain attitudes and behavior scripts in reference to specific situations. Within a gain goal, people focus on guarding and improving their resources. For example, if particular environmentally harmful behavior (i.e., littering) is sanctioned, a person in a gain goal frame could choose to act environmentally in order to avoid the negative consequences (i.e., paying fine). On the other hand, within a hedonic goal, people focus on immediate gratification, which makes them sensitive to what increase or decrease their mood state. The sight of litter in a public park may evoke negative mood state, which can be reduced by cleaning it up. However, it is important how much effort certain behavior costs. The thought of cleaning up the whole public park (which requires high effort) can dampen the way one feels in comparison to the thought of throwing only one soda can into the trash bin (which requires low effort). Finally, a normative goal is focused on what is generally considered appropriate, therefore people pursuing this goal are more sensitive to external information regarding norms. For example, visiting a national park for the first time would induce a person to search the environment for cues (i.e., signs concerning campfire permissions) to answer the question: what behavior would be appropriate in this place or situation?

Therefore, we conclude that an effective regulatory sign should combine cues that would activate at least one of the three goals (hedonic, gain, and normative) in the recipient. Nevertheless, it is crucial to have properly worded and visually well designed sign to avoid counterproductive effects (Winter, Sagarin, Rhoads, Barrett, & Cialdini, 2000).

Researchers have established that effective signs should (a) attract attention, (b) be understandable, and (c) motivate people to comply (Laughery, 2006; Wogalter, Begley, Scancorelli, & Brelsford, 1997). To maximize these features, previous findings suggest the use of proper visual sign design (e.g., adding color, enlarging print and using bold font, including pictorials) and explicit information in the text about the posted regulation (Braun, Mine, & Silver, 1995; Laughery, 2006; Schell, 2009; Wogalter et al., 1997; Wogalter, Conzola, & Smith-Jackson, 2002; Wogalter, Kalsher, & Rashid, 1999). These guidelines closely relate with model of persuasion process introduced by William McGuire (1985), which states that in order to persuade, constructed message should be (a) presented in attractive and visible way to the recipient, (b) perspicuous in the content, (c) perceived as useful, and (d) encouraging to compliance. Hence, we will address this dimension further in relation to construction of our survey used in the first study. Currently, the research focus is on words and language that can enhance the persuasiveness of signage (e.g., Burger & Shelton, 2011; Guéguen & Lamy, 2011; Lawrence, 2015; Pascual, Felonneau, Guéguen, & Lafaille, 2014; Updegraff, Emanuel, Gallagher, & Steinman, 2011).

Prescription and proscription

Focus Theory of Normative Conduct (Cialdini, Kallgren, & Reno, 1991; Reno, Cialdni, & Kallgren, 1993) states that normative information can describe either typical human behavior (descriptive norm, i.e., ‘Most people don’t clean up after their dog’) or desirable behavior in a particular situation (injunctive norm, i.e., ‘Please clean up after your dog’). In changing undesired behavior it is recommended to use injunctive norms over descriptive ones (Bator, Tabanico, Walton, & Shultz, 2014; Cialdini, 2003). For example, in a study concerning off-trial hiking (Winter, 2006), posting a sign with descriptive norm increased the undesired behavior whereas injunctive norm decreased it. This effect seems to correspond with goal-framing theory. Inducing and supporting of normative goal, that would evoke behavior change, requires direct and explicit cues. Injunctive norms seem to meet this conditions, hence our focus on this type of social norms.

It is important to notice that injunctions can be framed positively (prescriptive) or negatively (proscriptive) (Winter, 2006). For example, injunctive norm can encourage certain behavior (prescriptive – ‘Please clean litter before you leave the park’) or discourage undesirable one (proscriptive – ‘Please don’t litter during your stay in the park’). An important question arises as to what form of request an effective sign should use – prescriptive or proscriptive? The results of previous studies are contradictory (Cialdini et al., 2006; Winter, Cialdini, Batar, Rhoads, & Sagarin, 1998; Winter et al., 2000), showing that even if positively framed messages are perceived as more efficient, the negatively framed ones happen to be more effective in practical use. Due to these inconsistencies, we have decided to compare the effectiveness of both request forms – prescriptive and proscriptive – that would give us the opportunity to discuss the reasons for this discrepancy.

Consequence and justification

It is worth noticing that wording of most signs in public use does not simply consist only of injunctive requests. For example, ‘No smoking! Individual violation: $50 fine’ sign includes also the information about sanction/consequence on breaking the regulation. On the other hand, sign ‘Please do not feed birds. It encourages vermin and is a health risk’ includes the justification of posted regulation. We are interested in both types of this information (consequence and justification) since they can activate certain goals (within the goal-framing theory) that would lead to actual compliance.

Including information about the consequences of non-compliance with a stated rule (e.g., ‘Campfires are not permitted. Violators are subject to a $250.00 fine.’; Duncan & Martin, 2002, p. 22) positively impacts the effectiveness of a regulation (Duncan & Martin, 2002; Dwyer et al., 1993). As normative behavior theory (Blake & Davis, 1964) states, sanctions should do the following: (a) clearly and directly express one’s will (Shelby, Vaske, & Donnelly, 1996), (b) increase willingness to complete the desired action (Gramann, Bonifield, & Kim, 1995), and (c) help in setting objectives for compliance (O’Reilly & Puffer, 1989). For example, Reiter and Samuel (1980) limited the problem of littering in public parking lots through the use of signs containing information about possible sanctions. Gramann, Bonifield, and Kim (1995) concluded that sanctions create greater likelihood that people will refrain from breaking a given norm. From the perspective of goal-framing theory, exposing sanctions (or consequences) will induce in a recipient a gain goal, that would further lead to compliance (since recipient in a gain goal would want to avoid possible negative consequences). However, messages based solely on requests and threatened punishments may elicit negative reactions, therefore reduce their potential to influence behavior and, in some cases even lead to counterproductive effects (Pennebaker & Sanders, 1976; Winter et al., 2000).

In order to reduce this possibility it could be worth including additionally the justification of a norm. For example, researchers (Wogalter et al., 1997) have compared the effectiveness of elevator service signs encouraging riders to use the elevator only for longer distances. The result was that signs containing pictures, explicit wording and justification (e.g., ‘If you do this, we’ll have better elevator service.’; Wogalter et al., 1997, p. 183) were the most effective. Researchers agree that including a justification is more effective than posting a simple message (Duncan & Martin, 2002; Gramann et al., 1995; Ham, 1992; Widner & Roggenbruck, 2000). The elaboration likelihood model (Petty, Cacioppo, & Goldman, 1981) points out that personal significance (of message content to the recipient) enhances the motivated reception of the message; therefore, justification is important as it influences the understanding and acceptance of norms (Christensen & Dustin, 1989). In conclusion, justification to a regulation could counterweight possible negative reactions created by previously mentioned sanctions or consequences, therefore enhancing the overall persuasion of the written message.

Research aims

However, there has been no research evaluating the persuasiveness of signs that include both consequences and a justification even though these signs are in public use (i.e., ‘Do not feed the waterfowl! It creates dependence, the spread of disease, and upsets the natural cycles. They foul our shores, pollute our waters, and drive away timid species. Issued by the Bucks County Department of Parks & Recreation Ordinance #95, Section 38.b. Fines up to $300 will be issued.’)

Based on previous findings and conception of goal-framing theory, we hypothesized that a regulatory sign including both consequences and a justification (in comparison to signs with one or neither of these elements) would (a) be perceived as being the most persuasive and (b) enhance compliance. Also, due to inconsistencies in findings concerning request form, we asked the following research question: Which form of the sign is more effective – prescriptive or proscriptive? In the first study, we surveyed respondents to establish the perceived persuasiveness of each sign. In the second study, we placed these signs in real-life conditions and observed people’s behavior (to measure compliance to the regulation).

Situational context

One of the disobeyed norms in self-service stores is handling and selecting bread through a plastic bag that fits over the hand. Therefore, we decided to test the effectiveness of signs encouraging hygienic bread handling. For this purpose, we cooperated with four supermarkets in Warsaw, where management board members agreed to have the studies conducted. Both studies were approved by the ethics committee from the University of Social Sciences and Humanities in Warsaw.

Method of first study

We used the survey method to assess clients’ subjective rating of the constructed signs which promoted hygienic bread handling.

Sample and place

The survey involved 240 customers (equally distributed between sexes) of supermarkets in Warsaw. Respondents were divided into five age groups: 18–29 years old (n = 97); 30–39 years old (n = 66); 40–49 years old (n = 39); 50–59 years old (n = 35); and 60 years old and older (n = 3). The survey took place in stores next to the bread shelves where we placed the signs. Data collection lasted for 10 days in a constant time interval between 3:00 pm and 5:00 pm

Variables

The study adopted two independent variables: (a) form of request (prescriptive vs. proscriptive) and (b) sign content (justification vs. consequence vs. justification and consequence vs. neutral). The dependent variable was the persuasiveness of the sign (mean of the ratings given by respondents in the survey).

Materials

Signs

We constructed eight signs (example in Appendix 1) that differed in content and form of request (in order to exclude the impact of message length). Each sign contained three elements: (a) a request (prescriptive or proscriptive), (b) a justification or neutral content, and (c) a consequence or neutral content. We decided to incorporate neutral sign content in order to sustain similar quantity of written information between messages, since it is well established that it can be an important factor in recipient compliance (Cole, Hammond, & McCool, 1996).

The following statements appeared on the signs (translation from Polish):

•	Prescriptive form: ‘Touch bread by hand through the plastic bag!’
•	Proscriptive form: ‘Do not touch the bread by hand without the plastic bag!’
•	Justification: ‘By doing this: – You won’t spread germs; – The bread will stay fresher longer.’
•	Consequence: ‘Bread touched without the use of plastic bag is considered bought.’
•	Neutral content 1 (instead of a justification): ‘On the shelf you will find bread and ciabattas.’
•	Neutral content 2 (instead of a consequence): ‘Bread prices are located on the shelves.’

Based on these elements, we created the research conditions (see Table 1).

Table 1. Written content type of the signs in eight research conditions.

Condition number	Prescription	Proscription	Justification	Consequence	Neutral 1	Neutral 2
1	+		+	+
2	+		+			+
3	+			+	+
4	+				+	+
5		+	+	+
6		+	+			+
7		+		+	+
8		+			+	+

The content and appearance of the signs were discussed with the management of the market network. The signs had a height of 11 cm and a width of 16.5 cm.

Survey

The survey (Cronbach’s α = .75) contained six dimensions of sign evaluation based on the seven-point Thurstone scale, ranging from −3 (very negative) to +3 (very positive). Dimensions were extracted on the basis of McGuire’s persuasion model (1985) and guidelines about sign evaluation from other researchers (Frantz, Rhoades, & Lehto, 2005). Scales concerned the signs’ (a) visibility, (b) attractiveness, (c) perspicuity, (d) usefulness, (e) compliance encouragement, and (f) elicited reaction (negative or positive). These dimensions allowed us to calculate message persuasiveness (by averaging the ratings from each scale given by respondents into one variable). We additionally placed demographical boxes (age and sex) at the end of the survey.

Procedure

While in the bread section of the supermarket, we asked nearby clients to participate in a short survey. After stating the purpose of the study and obtaining a verbal agreement, we pointed at the sign placed on the shelf (one randomly selected version of the eight created) and asked the respondent to rate it through the survey. The whole procedure took approximately 2 minutes.

Results

A two-way between-subjects ANOVA of the form of request (prescriptive vs. proscriptive) and sign content (justification vs. consequence vs. justification and consequence vs. neutral) on the persuasiveness of the sign (measured by the survey) was conducted. Although the main effect of the sign’s content was statistically significant, F(3, 232) = 4.51, p < .05, η² = .05 (the main effect of the form of request was not, F(1, 232) = .39, p = .52), there was a significant interaction between both of the independent variables, F(3, 232) = 3.07, p < .05, η² = .04. Table 2 shows means and standard deviations across conditions.

Table 2. Means and standard deviations across experimental conditions.

Experimental condition	M	SD
Prescriptive with justification	5.51	1.02
Prescriptive with consequence	4.95	1.30
Prescriptive with justification and consequence	6.07	.89
Prescriptive with neutral content	5.16	.90
Proscriptive with justification	5.50	1.06
Proscriptive with consequence	5.22	.90
Proscriptive with justification and consequence	5.33	.92
Proscriptive with neutral content	5.32	.80

Simple effects analysis of the variable of sign content showed statistically significant differences between the prescriptive signs, F(3, 232) = 7.06, p < .05, η² = .09. The prescription that contained both a justification and a consequence was rated as being more persuasive (n = 30, M = 6.07, SD = .89) than the prescription with only a consequence (n = 30, M = 4.95, SD = 1.30) or only neutral content (n = 30, M = 5.16, SD = .90). However, they did not differ in a significant way from the prescription that contained only a justification (n = 30, M = 5.51, SD = 1.02). Analysis failed to provide the statistically significant differences between the proscriptive signs, F(3, 232) = .43, p = .73.

Discussion

The primary goal of this study was to determine possible levels of persuasiveness of experimental signs concerning hygienic bread handling. We partially confirmed our hypothesis, which stated that signs containing both a justification and a consequence of the request would be rated as being the most persuasive. However, the prescriptive sign containing only a justification was rated as highly as one with both a justification and a consequence. This result seems to confirm previous findings about the role of justification in persuasion (Langer, Blank, & Chanowitz, 1978; Wogalter et al., 1997). On the other hand, our hypothesis was not confirmed in the context of proscriptive signs. Nevertheless, presented results should be interpreted with precautions, since the effect sizes were small (η² = .05 for main effect of sign content; η² = .04 for interaction of independent variables). In answering our research question, the results failed to provide any significant differences between request forms.

In a study of human attitudes, it is crucial to distinguish declaration from real behavior (Ajzen, 1991; Ajzen & Fishbain, 2005). Therefore, the presented study is not sufficient to address our hypothesis and research question. That is why we have conducted another study using signs in real-life conditions (to rate their actual persuasiveness).

Method of second study

In the second study, we observed customer compliance to the norm of hygienic bread handling, which was expressed by different sign variations.

Sample and place

The study included 1440 observations in four supermarkets (in Warsaw) in order to exclude observation of the same group of people twice. As a part of the same network, each store had a similar assortment of goods, area size, visual identity, and organization of space. The observations took place in the bread sections. To ensure good visibility, we mounted signs next to the holes for bread selection (Appendix 2). Each shelf had two signs (a total of eight signs on four shelves). Data collection lasted for 3 days in a constant time interval between 3:00 pm and 5:00 pm.

Variables and research conditions

Variables and research conditions were similar to the pre-study. The only difference concerned the indicator of the dependent variable (sign persuasiveness), which was the actual behavior of customers in relation to the posted signs. In the control condition, there was no sign.

Materials

For the purpose of the natural experiment, we used the same set of signs as in first study. We also created a form for making notations (observation number) and classifying observed behavior (compliance or non-compliance).

Procedure

We used security cameras (mounted above the bread sections) for gathering observational data. On the first day, we collected data for the control group (no signs) from four locations. At the beginning of the second and third days, we placed randomly assigned sets of signs in each location and then collected data. In conclusion, each of the four supermarkets had two randomly assigned experimental conditions. At the end of each day, the signs were removed. After collecting the data, we analyzed the security camera recordings from every supermarket. We focused on situations where only one person approached the bread shelf. To categorize customers’ behavior as being obedient to a sign, the subject had to put a plastic bag on his or her hand and use it for handling the bread. Due to possible errors, we did not categorize subjects by age and sex.

Results

In the first step, the differences between experimental locations were examined with the use of the chi-square test for one variable. The value of the statistics was non-significant, χ² (3, 480) = .21, p = .98. Therefore, the number of behaviors consistent and inconsistent with the norm in each location was not different in a statistically significant way. In the next step, the frequency of unhygienic bread handling was assessed. The proportion of customers who touched bread through the plastic bag (n = 96) and without the plastic bag (n = 384) was compared within the control group. The value of the statistics was significant, χ² (1, n = 480) = 172.80, p < .001, showing that significantly more people handled bread in an unhygienic way.

To verify our hypothesis, logistic regression analysis was conducted to predict sign persuasiveness using sign content and form of request as predictors. In the first step, sign content was entered to a model; second step included form of request; and third step analyzed the interaction of two predictors. Dependent variable – sign persuasion – was equal to 1 if customer complied to a regulation in a sign, and 0 if not. Table 3 shows coefficients of built models.

Table 3. Coefficients of two built models in logistic regression.

	B	S.E.	Wald χ2	p	Exp(B)	95% Confidence interval for EXP(B)
						Lower	Upper
Model 1
Sign content			81.11	<.001
Justification	.99	.18	31.95	<.001	2.69	1.91	3.79
Consequence	.99	.18	31.95	<.001	2.69	1.91	3.79
Justification + consequence	1.45	.17	69.92	<.001	4.25	3.03	5.96
Neutral content	.50	.18	7.60	.006	1.66	1.16	2.37
Constant (control)	−1.41	.12	150.83	<.001	.24
Model 2
Sign content			81.11	<.001
Justification	.99	.18	31.95	<.001	2.69	1.91	3.79
Consequence	.99	.18	31.95	<.001	2.69	1.91	3.79
Justification + consequence	1.45	.17	69.92	<.001	4.25	3.03	5.96
Neutral content	.50	.18	7.60	.006	1.66	1.16	2.37
Form of request			2.79	.095
Prescriptive	−.22	.13	2.79	.095	.80	.62	1.04
Constant (proscriptive)	−1.41	.12	150.83	<.001	.24

The results from Model 1 indicated that sign content reliably distinguished between obedience and disobedience to the signs – χ² (4) = 86.33; p < .001. The Wald criterion demonstrated significant contribution of sign content – χ² (4) = 81.11; p < .001. Statistics of Cox and Snell’s (R² = 0058) and Nagelkerke’s (R² = 0081) revealed low relationship between prediction and grouping, whereas overall prediction success was 67.3% (25.7% for obedience and 87.8% for disobedience). Analysis of odds ratio (OR) for each experimental condition (in comparison to control group) indicated that sign including both justification and consequence (OR = 4.25) generated four times more likely compliance behavior with the written norm. Posting sign with only consequence (OR = 2.69) or justification (OR = 2.69) caused over two more times higher compliance rate. On the other hand, using sign with only neutral content can potentially increase compliance by 66% with comparison to control condition.

The results from Model 2 revealed that the form of request fits the data on the level of statistical tendency – χ² (1) = 2.79; p = .095. Coefficients of pseudo R-squared statistics raised to the level of R² = .06 (Cox and Snell) and R² = .084 (Nagelkerke). However, the Hosmer-Lemeshow test was highly non-significant, therefore showing a good fit of model to the data – χ² (3) = 2.03; p = .845. Odds ratio for proscriptive form (OR = .8; after reversal OR = 1.25) indicates that prescriptive form is associated with 25% higher chance for customer compliance.

In the last step, the interaction between sign content and form of request was added to the model, showing that both predictors did not interact with each other – χ² (3) = 206; p = .560.

Since sign content significantly distinguished between customer obedience and disobedience to the signs – series of chi-square test for two variables were conducted between experimental conditions. Sign including both justification and consequence, differed from sign with neutral content [χ² (1, 480) = 24.51, p < .05, φ = .23] with potential of improving customers compliance by two and a half times (OR = 2.57). Furthermore, sign with both justification and consequence differed from sign with only justification [χ² (1, 480) = 5.69, p < .05, φ = .11] with 55% higher chance of inducing customers compliance (OR = 1.55). In comparison to sign with only consequence, sign with both justification and consequence differed significantly [χ² (1, 480) = 6.14, p < .05, φ = .11] with 58% higher chance of inducing customers compliance (OR = 1.58). Results did not provide statistically significant difference between sign including only justification or only consequence – χ² (1, 480) = .009, p = .93. However, both of these conditions significantly differed from neutral sign condition – consequence vs neutral: χ² (1, 480) = 6.29, p < .05, φ = .11; and justification vs neutral: χ² (1, 480) = 6.77, p < .05, φ = .12. Odds ratio of sign with consequence (OR = 1.62) and sign with justification (OR = 1.65) showed that in comparison to neutral sign condition, they were associated with 62 and 65% greater chance of customer compliance. Figure 1 presents the frequencies of shoppers compliant and noncompliant behaviors with the norm of hygienic bread handling across the sign content conditions.

Figure 1. Frequencies of shoppers compliant and noncompliant behaviors with the norm of hygienic bread handling across the sign content conditions.

Discussion

Our hypothesis, which stated that signs containing both a justification for and a consequence of the request would achieve the highest compliance (be most persuasive), was confirmed. The second study showed that regardless of the persuasive content, signs are an effective form of changing customer behavior. Consequently, this finding seems to strengthen the prediction from goal-framing theory which states that signs serve as cues in activating normative goal (Lindenberg & Steg, 2007). Interestingly, signs containing both a justification and a consequence affected the subjects’ behavior more frequently (in comparison to other signs), whereas signs with only justification or consequence revealed as equally effective. On the other hand, the presence of signs with neutral content occurred as least effective intervention.

The results of the natural experiment do not allow us to answer the question of which form of request (prescriptive or proscriptive) evokes higher compliance in the context of the norm of hygienic bread handling. However, results showed statistical tendency towards higher persuasion of proscriptive form.

General discussion

Both studies revealed that, as predicted, signs with a written justification and consequence appear to be more persuasive than the use of either justification or consequence alone. This regularity emerged from both descriptive and observational measures. However, it is worth noticing that the use of content containing only a justification or consequence is also sufficient; this confirmed findings from previous studies. The results showed that messages including neutral content (except for the request) have limited impact. These findings could be possibly explained within the goal-framing theory. Adding consequence to a regulation could elicit gain goal in the recipient, whereas additional content of justification lowered the possible negative reactions induced by consequence – therefore eliciting hedonic goal in the recipient. However, presented interpretation should be treated cautiously since we did not measure the induction of these goals and statistically significant results were associated with small effect sizes.

As for the research question (Which request form is more persuasive?) – the first study did not reveal significant differences between prescriptive and proscriptive form of sign. Furthermore, natural observation analysis showed differences between request form but on the level of statistical tendency. Hence, any further interpretation should be treated cautiously. Our research question should be further examined in the context of sign persuasion. Nevertheless, our findings show the importance of using both subjective and objective measurements in calculating sign persuasiveness. Including only one of these methods could leave researchers with false predictions.

Limitations and future research

We did not control the longitudinal effect of signs on attitude and behavior change. Therefore, we could not assess the durability of sign persuasion. Furthermore, our observational data lacked information about the sex and age of respondents (which could lead to more possible explanations of sign effectiveness). We suggest the use of more precise measurements for observation to gather these types of information and exclude possible errors.

Regarding sign design, we concluded that our consequence content (‘Bread touched without the use of plastic bag is considered bought.’) could be replaced with a monetary sanction. Unfortunately, in this specific situational context, we were not able to use this kind of information because of consumer law. As for the justification content, it could be taken into consideration if a false argument on a sign could be as persuasive as a real one.

Future research should address the limitations of the two presented studies and take into consideration different situational contexts in which signs could be evaluated. The possibilities for written sign design are as numerous as their limitations. Further research can provide important information about effective message design, which can lead to improving people’s engagement in the public environment and the upholding of social norms.

Disclosure statement

Authors have no relevant financial or nonfinancial relationships to disclose.