Too Good To Be True? Effect of Consumers’ Brand Schemas on Apparel Brands’ Fair Labor Marketing Messages

Abstract

In this study, researchers provide a comprehensive model of how consumers process, remember, and evaluate positive fair labor-related brand messages that are congruent/incongruent to their existing brand expectations using both psychophysiological and self-reported measures. Data were collected across two different studies. Results indicated that consumers paid more attention to and better remembered messages for incongruity than congruity. Also attitude toward message was highest for congruent messages, followed by incongruity resolution and lowest for incongruity nonresolution. The combined study findings bridge the gap in literature between human attitude and cognition, helping both brands and consumer researchers understand consumers’ reaction to brands schema-message congruity/incongruity and guide decisions when brands hope to revitalize or reinvent a brand’s image.

Keywords

schema congruity schema incongruity brand expectations attention recognition attitude

Today’s consumers are interested in knowing about the social impact of their consumption choices (Nielsen, 2015). Therefore, labor practices, or brands’ fair labor (FL) initiatives, have been a focus of corporate social responsibility literature (Dickson, 2001). Typically, FL initiatives include ensuring a 40-hr workweek, adhering to the national minimum wage requirements, and providing proper pay for overtime work. The initiatives also prohibit employment of minors in jobs and under conditions detrimental to their health or well-being (U.S. Department of Labor, 2016). FL claims are especially important in the apparel industry, which employs more than 60 million people globally and, thus, has potential for considerable social impact (Fashion United, 2014). Especially, recent tragedies, such as the 2013 Bangladesh apparel factory collapse, have heightened awareness about FL practices (“Bangladesh Factory Collapse,” 2013), and consumers are paying attention to brands’ FL efforts, often preferring socially responsible apparel brands over others (Nielsen, 2015).

In response, apparel brands such as Patagonia and Nike are now creating marketing messages communicating their FL practices to allure consumers and persuade them about their FL initiatives. However, not all FL messages are equally effective. Given that messages are meant to inform, remind, and persuade consumers, researchers have often focused on message attributes or characteristics that can evoke the desired attitudinal/behavioral responses from consumers (Brown & Stayman, 1992). Particularly, researchers have studied the role of congruity/incongruity as a key message attribute in consumers’ message evaluations and the effectiveness of message communications (Hutter, Lawton, Pals, O’Connor, & McEachan, 2015; Sjödin & Törn, 2006). Most of these researchers have focused on the impact of incongruity within different message components or elements, such as that between pictures/words in messages, message/general viewer expectations, and brand extensions. This is because incongruity is known to influence attitudes toward messages and the product in general (Lee & Schumann, 2004).

However, little is known about incongruity between an apparel brand message and consumers’ expectations of the brand’s behavior. For instance, how do consumers process and evaluate marketing messages about an apparel brand’s FL initiatives when they expect the brand (based on their previous exposures) to not employ FL? To answer these questions, the authors investigated how consumers pay attention to, recognize, and form attitudes toward apparel brands’ FL messages that are congruent or incongruent with their expectations of brands’ FL initiatives. Specifically, the focus was on FL messages from apparel brands that may have different levels of reputation about their FL efforts. The findings are important for apparel brands that suffer from negative FL reputations but are willing to revitalize their images.

For this research, psychophysiological and self-reported responses were collected. Psychophysiology involves understanding the relations between human psychological and physiological processes (Potter & Bolls, 2012). Due to the strong connection between mind and body, humans’ dynamic psychological processes (e.g., cognitive resources allocated) can be observed in real time through corresponding changes in physiological activities (e.g., change in heart rate; Leshner, Bolls, & Thomas, 2009). Self-reported measures, by comparison, are designed to assess responses to extreme points in message content after the experience, rather than real-time responses to the message as a whole, and may be subject to social desirability bias (Potter & Bolls, 2012). Thus, between psychophysiological and self-reported measures, the results were thought to give a comprehensive understanding of consumers’ attention, recognition, and attitude formation toward apparel brands’ FL messages based on their brand expectations or schemas.

Theoretical Framework

Schema Congruity Theory, FL-Related Brand Schema, and Brand Message

Schema congruity theory explains how individuals process new information by comparing it to their existing expectations or schemas (Mandler, 1982). Schemas are defined as “representations of experience” stored in our memory that are developed from prior encounters with situations and guide our “actions, perceptions, and thoughts” (p. 3). When new information is congruent with schema, one feels a sense of familiarity and evaluates the event positively. On the contrary, incongruent information creates conflict and disturbance, and thus, evaluation is either strongly positive (if incongruity is resolved, as a reward for conflict resolution) or strongly negative (if incongruity is not resolved; Mandler, 1982).

Brand schemas are brand-related attributes, experiences, or beliefs that help consumers formulate the meaning of the brand and form brand-related opinions (Dahlén, Lange, Sjödin, & Törn, 2005). Incongruity within brand messages, such as incongruity between audio and image components of messages, has been found to result in both positive and negative message evaluations (Lee & Schumann, 2004). Congruity or slight incongruity is preferred as it can be effortlessly assimilated into an existing schema (Goodstein, 1993). In this case, the affective response may depend on the prior affect (positive/negative) attached to the activated schema into which the incongruent information is being assimilated (Lee & Schumann, 2004). Conversely, for moderate to severe incongruity, accommodation occurs and a new schema is generated. As a result, there is no prior affect attached to any existing schema, and the resulting affect is usually positive as a reward to reconcile the new information (Lee & Schumann, 2004).

However, limited literature exists on incongruity between brand message and consumers’ brand schema. In one of the few studies, Dahlén and Lange (2004) observed that incongruity between a brand message and brand schema did not result in increased positive evaluation. Rather, the authors suggested that schemas of familiar brands that are developed over repeated interactions with brands are personal, strong, and resistant to change. Therefore, brand messages congruent with consumers’ brand schemas are evaluated positively, while those incongruent might make consumers feel disturbed and evaluate negatively. However, researchers suggested that consumers’ message evaluations might depend on their ability to resolve incongruity rather than incongruity of the message itself. Therefore, the findings related to brand schema-message congruity/incongruity have been inconclusive. To fill this gap, two studies were conducted investigating how consumers process, recognize, and evaluate FL messages from apparel brands that may be congruent or incongruent with their existing expectations of such brands.

Study 1: Effect of Congruity/Incongruity on Attention, Recognition, and Attitude Toward Message

Hypothesis Development

Incongruity and physiological responses

In a cluttered advertisement context, consumers often indulge in schema-based processing because they can reuse the information already processed previously rather than reevaluating it (Lee & Schumann, 2004). Information congruent with existing schema is familiar and therefore encoded with minimum effort into that schema. However, incongruent information leads to higher attention, or greater allocation of cognitive resources and elaboration to encode the message to process unexpected message elements and resolve incongruity (Vanhamme & Snelders, 2001). For example, incongruent advertisements were viewed longer (Goodstein, 1993), elicited a greater number of thoughts (Lane, 2000), and led to more focused processing of product-related message arguments (Sujan, 1985) as well as increased time spent, or attention, on messages (Törn & Dahlén, 2008). These researchers measured attention using observations and/or self-reported time spent on an advertisement. However, this approach has been criticized, as time spent might not always indicate active attention devoted to message processing (Lang, 2000). At the same time, self-reported measures capture reaction after information has been processed, rather than measuring cognitive processes taking place during message processing (Lang, 2000).

The field of psychophysiology provides an opportunity to understand dynamic relationships between psychology and physiology while consumers process external stimuli (Potter & Bolls, 2012). An eminent theory in psychophysiology is the limited capacity model of motivated mediated media message processing (LC4MP; Lang, 2000). LC4MP is based on the premise that human beings have a limited amount of cognitive resources that can be allocated to mental tasks involved in perceiving, comprehending, and remembering information encountered. When exposed to a media message, mental tasks consist of three major processes performed simultaneously and continuously, namely encoding into working memory, storage into long-term memory, and retrieval from long-term memory. Particularly, Lang (2000) suggests that different amounts of cognitive resources will be allocated among the three tasks based on different message needs and individuals’ motivations to process information. Such differences in cognitive resource allocation can be assessed by observing the information recipients’ heart rates (Lang, 2000). Indeed, Bolls, Lang, and Potter (2001) empirically showed that the recipients’ heart rates decelerated (i.e., heart beat slower than their baseline data) when they had to allocate increased cognitive resources to encode novel, or unexpected, messages into working memory.

Within the consumer research context, given consumers’ increased interests in apparel brands’ FL efforts (Hyllegard, Yan, Ogle, & Lee, 2012; Nielsen, 2015), LC4MP could be well applicable. That is, when apparel brands’ FL messages are incongruent with their brand schemas, additional cognitive resources would be required to process and reconcile the difference between the schemas and the new information, leading to deceleration of consumers’ heart rates. However, this is yet to be known in the apparel consumer context. Therefore, it is hypothesized that:

Hypothesis 1: Attention, or cognitive resources allocated to encoding a message, will be higher, so that heart rate will decelerate more during exposure to incongruent compared to congruent FL messages from apparel brands.

Congruity/incongruity and message recognition

Incongruity in brand messages is found to increase message memorability (Lee & Schumann, 2004). Sujan (1985), using free recall as a measure of message memorability, found that incongruent information was better recalled than congruent information. Although free recall is a measure of memory, in very few instances, consumers are able to process marketing messages deeply enough to freely recall such information on later instances (Shapiro, 1994). For example, if consumers are asked to report what they remember about some advertisement, most consumers might recall only a couple of things. Yet this does not mean that they do not at all remember other aspects of the advertisement. If on the other hand, the consumers are given a list of statements, some from the advertisement shown and others not from the advertisement shown, they would probably correctly recognize many statements that they failed to recall (Shapiro, 1994). Consequently, in this study recognition is used as a measure of memory. In addition, recognition measures allow researchers to determine memory strength and judgment aspects of memory (Shapiro, 1994).

Message recognition is important since manufacturers of branded merchandise want consumers to recognize, then freely recall, brand names and marketing information in real shopping scenarios (Shapiro, 1994). In fact, if information contained in FL messages is not correctly recognized by consumers, it is unlikely that the FL messages have been processed deeply enough to potentially affect consumers’ attitudes and behaviors toward FL products (Leshner, Bolls, & Wise, 2011). Researchers found that message recognition is greater when higher attention is paid, or in other words, when greater cognitive resources are allocated to encode the message (Lang, Bradley, Park, Shin, & Chung, 2006). Given that incongruent information requires a greater amount of cognitive resources to process than congruent information (as proposed in Hypothesis 1), it is expected that message recognition or memorability will be higher for incongruent than congruent messages. Therefore, it is hypothesized that:

Hypothesis 2: Message content recognition will be higher for incongruent compared to congruent FL messages made by apparel brands.

Congruity/incongruity and attitude toward messages

Attitude toward message (AttM) is an indicator of message effectiveness and measures how consumers evaluate messages (Hyllegard et al., 2012). It is defined as a “predisposition to respond in a favorable/unfavorable manner” to a particular message stimulus “during a particular exposure occasion” (Mackenzie & Lutz, 1989, p. 49). Findings on the impact of brand schema-message incongruity/congruity on AttM have been inconclusive. Some researchers found that incongruity led to more favorable AttM for familiar brands but not for unfamiliar brands (Dahlén & Lange, 2004; Dahlén et al., 2005). However, according to Mandler’s (1982) schema congruity theory, consumers’ evaluations, or AttM, should depend on their congruity or incongruity resolution after message exposure and not the mere incongruity of a message. This proposition was also supported by Törn and Dahlén (2008) in a concept paper. In this light, Lee and Schumann (2004) argued that marketing messages can influence individuals’ existing schemas. Particularly, marketing messages can help individuals resolve incongruity by acting as a “resolution message or a hint” that can suggest a possible connection between the seemingly contradictory message and existing brand schema (Lee & Schumann, 2004, p. 75). Therefore, in this study, the authors investigate the effect of congruity, incongruity resolution and nonresolution on consumers’ AttM for FL messages.

Mandler’s schema congruity theory (1982) states that schema congruity leads to positive evaluation, incongruity resolution to evaluation more positive than congruity, and incongruity nonresolution to negative evaluation. However, for brand schema-message congruity/incongruity, AttM might be more positive for congruity than incongruity resolution, since consumers’ brand schemas tend to be strong and personal, and therefore resistant to change, thus favoring congruity (Dahlén & Lange, 2004). A similar relationship is expected between AttM and FL messages from apparel brands. Thus, it is hypothesized that:

Hypothesis 3: For FL messages from apparel brands, AttM will be most positive (favorable) for schema-message congruity (Hypothesis 3a), less positive (favorable) when schema-message incongruity is resolved (Hypothesis 3b), and negative (unfavorable) when incongruity is not resolved (Hypothesis 3c).

Method

Research Design

A 2 (schema: congruity/incongruity) × 4 (brand replications) × 4 (message replications) mixed model repeated measure experimental design for attention and recognition, and a 3 (schema: congruity/incongruity resolution/incongruity nonresolution) × 4 (brand replications) × 4 (message replications) mixed-model repeated measure experimental design for AttM were implemented for the study. Schema was measured between subjects, message replication was manipulated between subjects, and brand replication was manipulated within subjects. Although consumers’ schemas might be influenced by FL messages from multiple sources, in this study the authors only considered the effect of brands’ own FL messages on consumers’ attention, recognition of, and attitude toward FL marketing messages, to control for message source effect. Further, only positive FL messages were manipulated, as few brands would communicate negative aspects of their FL initiatives.

Because each media message is composed of an infinite number of attributes, there are risks of effects of confounding variables (Thorson, Wicks, & Leshner, 2012). In a single message design, any conclusion about the effect of manipulation must be restricted to the particular message rather than groups/categories of messages (Thorson et al., 2012). Since groups of messages, in this case FL messages, were a focus of this study, multiple but similar FL messages from different apparel brands in each treatment group helped generalize study results to a wider population of messages. Following Thorson, Wicks, and Leshner (2012), to reduce the confounding effect of text contents, four different message versions were developed, altering the text and overall look for each brand to represent the four levels of message replications. Additionally, four brand names (Nike, Adidas, Reebok, and New Balance) were used to represent the four levels of brand replications. This resulted in 16 different FL messages (static brand webpages). The aim was not to see differences across messages themselves but to generalize the findings across FL messages and brands by reducing between-message variance to random error (Thorson et al., 2012).

Stimuli and Manipulation Check

The study objective necessitated that brands in stimuli must be familiar to participants, so that their existing brand schemas can be measured (Dahlén & Lange, 2004). For that, six apparel professors and researchers identified six apparel brands that could be considered familiar to most U.S. consumers. Next, 100 participants recruited through Amazon Mechanical Turk (AMT) indicated if the brands used in the study were familiar (5-point Likert-type scale: 1 = not familiar at all to 5 = extremely familiar) as well as existing schemas about these brands’ FL efforts (7-point Likert-type scale: 1 = strongly disagree to 7 = strongly agree). The four most familiar brands were Nike ( $\bar{x}$ = 4.18, SD = 0.896), Adidas ( $\bar{x}$ = 3.83, SD = 0.947), Reebok ( $\bar{x}$ = 3.77, SD = 0.859), and New Balance ( $\bar{x}$ = 3.10, SD = 3.10). Meanwhile, participants’ mean schema scores about these brands’ FL efforts showed variations, ensuring that they had more favorable schemas toward some brands and less toward others. Thus, these four brands were deemed appropriate for the study (see Appendix for stimuli).

Measures

An independent variable, schema, was operationalized as consumers’ expectations about apparel brands’ FL initiatives. It was measured twice, before (preschema) and after (postschema) message exposure on a 7-point Likert-type scale (1 = being strongly disagree and 7 = being strongly agree), using 5 items adapted from Dickson (2001, p. 104): (brand) pays “fair wages to its workers,” (brand) “provides workers safe workplaces,” Child labor is “generally not used” by (brand), (brand) “requires that their workers do not work more than normal working hours without extra compensations,” and (brand) “cares about its workers.” Despite Dickson’s scale being developed in 2001, it is still used and found reliable and valid, capturing most of today’s apparel consumers’ labor concerns (Kozar & Hiller Connell, 2013; U.S. Department of Labor, 2016). The anchoring question was “based on your previous experience with the brand, please indicate your perception about (brand)” to capture brand schema as defined by Mandler (1982). The name of the brand whose message was shown was substituted in all items.

One mean score was obtained per participant for each preschema and postschema for conducting between-subjects mean comparison. Since attention (Hypothesis 1) and recognition (Hypothesis 2) depended on whether the message was congruent/incongruent with participants’ existing schema, two categories were created based on preschema scores: (a) messages congruent with participants’ schema (schema congruity) and (b) messages incongruent with participants’ schema (schema incongruity) stemming from brands’ FL messages. Schema incongruity occurred when participants believed that a brand did not practice FL but saw a FL message claiming that the brand does. Therefore, a preschema score less than or equal to 4 indicated that participants had incongruent/neutral preschema. On the contrary, schema congruity was when participants believed that a brand did practice FL and saw a message claiming it does. Therefore, a preschema score greater than 4 indicated participants’ congruent preschema.

In addition, because AttM (Hypothesis 3) depended on participants’ ability to resolve incongruity after message exposure, pre- and postschema scores were compared to form three groups: participants with (a) schema congruity before and after message exposure, (b) schema incongruity before but resolved after message exposure, and (c) incongruity before and not resolved after message exposure. The first was formed when participants’ preschema score was higher than 4 (indicating congruity before message exposure) and postschema score was also higher than 4 (indicating congruity after message exposure). The second was identified when participants had a preschema score less than or equal to 4 (indicating incongruity before message exposure) but a postschema score higher than 4 (indicating resolution of incongruity or congruity after message exposure). The final was created when preschema scores were less than or equal to 4 (indicating incongruity before message exposure) and a postschema score less than or equal to 4 (indicating incongruity after message exposure).

The dependent variable, attention, was conceptualized as cognitive resources allocated to encoding information to short-term memory and measured using heart rate deceleration pattern (Lang, Greenwald, Bradley, & Hamm, 1993). Heart rates were measured by placing electrodes (bipolar 8-mm Ag/AgCl and a ground) on participants’ forearms. Heart rate data were collected as milliseconds between R-spikes in the QRS complex of the cardiac cycle (Potter & Bolls, 2012) and converted to beats per minute averaged for every 1 s of data collection, resulting in 30 data points per claim exposure. A 5-s baseline signal was recorded prior to each message exposure. The recordings were synchronized with message exposure.

The dependent variable, message recognition, was conceptualized as how well information from stimuli messages was encoded into participants’ short-term memory (Lang et al., 2006). Recognition memory was operationalized as 16 image clips from each message (eight targets or true information, eight foils or false information, two targets and two foils from each message) and asking people (yes/no) to identify whether a clipping shown was from any of the messages they viewed during the experiment (Leshner et al., 2011). Recognition accuracy was measured as percentage of correct recognition of target clips (Shapiro, 1994). Two parameters of recognition performance were also calculated: sensitivity and criterion bias according to measures indicated by Shapiro (1994). Sensitivity indicates a person’s ability to correctly identify something as familiar and having been seen before. Criterion bias measures how liberal or conservative a person is in deciding if the information presented matches memory, and bias refers to the set limit that each person has before she or he can say that the item was seen before (Shapiro, 1994). Criterion bias varies on personal traits and on the task, or reward (or risk/punishment) for making a correct (or incorrect) decision. A higher number for criterion bias indicates a more conservative decision, meaning the item must be more familiar to be recognized.

Finally, AttM was measured using four items on a 7-point semantic differential scale (−3 to +3). Participants indicated whether the message was favorable/unfavorable, pleasant/unpleasant, good/bad, and like it very much/dislike it very much (Burton & Lichtenstein, 1988; Mackenzie & Lutz, 1989).

Sample Selection and Procedure

A sample of 75 adult participants was recruited in a large Midwestern state university through advertisements in the university news media. Participants needed to be adult consumers of apparel. First, participants completed demographic questions and indicated their preschemas about the four brands’ FL efforts on paper. Then, their skin was prepared for proper contact with electrodes for psychophysiological data collection. The computer program, MediaLab (Version 2014.1), was used to control presentation of all instructions, stimulus messages, and self-reported items (Jarvis, 2014). Participants were instructed to browse through static webpages as they would do while surfing the Internet. They were randomly exposed to four messages (one randomly selected from the four brands) for 30 s each, with their heart rate activities recorded during message exposure. After every message exposure, they completed postschema and attitude questions. Next, 16 recognition questions were randomly presented. Finally, participants were debriefed and compensated with a store gift card. The entire procedure took 45–60 min.

Data Analyses

For Hypothesis 1, heart rate data were analyzed as change scores from the baseline heart rates, using 2 (preschema congruity/incongruity) × 4 (message replication) × 30 (time) repeated measures analysis of variance (ANOVA). Change scores from the baseline were computed for each second of message viewing by subtracting each second of heart rate activity collected during message exposure from the average of 5 s of baseline data. This resulted in 30 data points per participant, since viewing time for each stimulus was 30 s. Due to equipment malfunction, responses from two participants could not be analyzed, leading to 73 usable responses for heart rate data (incongruity n = 35, 47.9%; congruity n = 38, 52.1%). For recognition, eight responses could not be analyzed due to missing data, leaving 67 usable responses. Independent samples two-group t test (schema congruity n = 35, 52.2%; incongruity n = 32, 47.8%) was performed for recognition (Hypothesis 2). For AttM (Hypothesis 3), one-way ANOVA (three missing data, n = 72) was conducted comparing three groups: participants with (a) schema congruity before and after message exposure (n = 35, 48.6%), (b) schema incongruity before but resolved after message exposure (n = 28, 38.9%), and (c) incongruity before but not resolved after message exposure (n = 9, 12.5%).

Results

Respondent Profile

The ages of participants ranged from 18 to 70 (14.5% male). Participants belonged to the following ethnic origins: 64 (84.2%) Caucasians, four (5.3%) African Americans/Blacks, one (1.3%) Hispanic/Latino, four (5.3%) Asians, and three (3.9%) others, including Middle Easterners. Participant characteristics are noted in detail in Table 1.

Table 1.

Demographic Characteristics of Study Respondents for Studies 1 and 2.

Variable	Levels	Study 1: Laboratory Experiment (n = 75)		Study 2: Online Experiment (n = 500)
Variable	Levels	Frequency	%	Frequency	%
Age	18–20	25	32.9	2	0.4
	21–24	22	28.9	8	1.6
	25–30	11	14.5	32	6.4
	31–35	4	5.3	33	6.6
	36–40	5	6.6	31	6.2
	41–45	2	2.6	52	10.4
	46–50	1	1.3	60	12.0
	51–55	0	0.0	79	15.8
	56–60	5	6.6	81	16.2
	61 and up	1	1.3	122	24.4
Gender	Male	11	14.5	250	50.0
	Female	65	85.5	250	50.0
Marital status	Single	28	36.8	113	22.6
Marital status	In a relationship	30	39.5	40	8.0
	Married	14	18.4	250	50.0
	Divorced/widower	4	5.3	97	19.4
Education level	Some high school education	0	0.0	7	1.4
Education level	High school degree	3	3.9	115	23.0
	Some college education	43	56.6	147	29.4
	College degree	9	11.8	148	29.6
	Some graduate education	8	10.5	21	4.2
	Graduate degree	13	17.1	61	12.2
	Other	0	0.0	1	0.2
Student status	Student	58	76.3	23	4.6
Student status	Nonstudent	18	23.7	477	95.4
Employment status	Not employed	15	19.7	120	24.0
Employment status	Employed part time (1–39 hr/week)	44	57.9	79	15.8
	Employed full time (40 or more hr/week)	17	22.4	183	36.6
	Retired	0	0.0	118	23.6
Ethnicity	Caucasian	64	84.2	419	83.8
	African American/Black	4	5.3	42	8.4
	Hispanic/Latino	1	1.3	19	3.8
	Asian	4	5.3	9	1.8
	Middle Eastern	3	3.9	2	0.4
	Pacific Islander	0	0.0	1	0.2
	Native American/Alaskan	0	0.0	7	1.4
	Other	0	0.0	1	0.2
Income	Less than US$10,000	28	36.8	24	4.8
	US$10,000–US$29,999	12	15.8	120	24.0
	US$30,000–US$39,999	8	10.5	70	14.0
	US$40,000–US$59,999	6	7.9	111	22.2
	US$60,000–US$89,999	12	15.8	95	19.0
	US$90,000–US$119,999	4	5.3	39	7.8
	US$120,000–US$199,999	3	3.9	34	6.8
	US$200,000 and above	3	3.9	7	1.4

Principal Component Analysis and Scale Reliability

Principal component analysis (PCA) with oblique rotation using eigenvalue > 1 revealed one principal component for each preschema (83.70% of total variance; factor loadings: 0.94–0.66), postschema (85.31% of total variance; factor loadings: 0.94–0.90), and AttM (91.91% of total variance; factor loadings: 0.94–0.97). In this study, oblique rotation was used because factor correlations were unknown and this method allowed for factor correlations (Abdi, 2003). The Cronbach α of preschema items was 0.95, postschema 0.96, and AttM 0.97, suggesting high reliability.

Hypotheses Tests

Attention (Hypothesis 1) was measured using heart rate deceleration patterns. For repeated measures ANOVA, the sphericity assumption was violated for the Heart Rate × Time interaction term, and therefore a Huynh-Feldt adjustment was used. The associated adjusted degree of freedom is presented as nonadjusted to aid interpretation. Analyses of the results indicated that Schema × Time interaction was statistically significant, F(29, 43) = 2.65, p = .037, partial η² = 0.04. Therefore, the heart rate deceleration pattern over time—attention—was higher for participants who experienced schema incongruity than for those with schema congruity, supporting Hypothesis 1. Findings for Hypothesis 1 and heart rate deceleration pattern are shown in Figure 1.

Figure 1.

Hypothesis 1 findings: Heart rate for preschema congruity/incongruity. Repeated measures ANOVA conducted for mean difference in heart rate (n = 73).

Recognition accuracy was higher, that is, participants better remembered incongruent messages ( $\bar{x}$ = 0.75, SD = 0.15) than congruent ( $\bar{x}$ = 0.5, SD = 0.16) ones (t = −6.46, p < .001), supporting Hypothesis 2. Memory sensitivity, or the ability to identify parts of the message as familiar, was also higher for incongruent ( $\bar{x}$ = 0.66, SD = 0.12) than congruent ( $\bar{x}$ = 0.56, SD = 0.07) FL messages (t = −4.83, p <.001). However, criterion bias did not show statistical differences between congruent ( $\bar{x}$ = 0.28, SD = 0.1) and incongruent ( $\bar{x}$ = 0.18, SD = 0.07) messages (t = 0.88, p = .38). Therefore, for both situations, participants were equally liberal or conservative in guessing if they had seen a message clipping before. Hypothesis 2 results are shown in Table 2.

Table 2.

Hypothesis 2 Findings: Recognition for Preschema Congruity/Incongruity.

Variable	Condition	Mean	SD	T test for Equality of Means
Variable	Condition	Mean	SD	t	df	Sig.
Accuracy	Incongruity	.75	.15	6.46	65	<.001
	Congruity	.50	.16
Sensitivity	Incongruity	.66	.12	–4.38	65	<.001
	Congruity	.56	.07
Criterion Bias	Incongruity	.18	.07	0.88	65	0.383
	Congruity	.28	.10

Note. Two-group independent samples t test conducted for fair labor message recognition (n = 67).

For Hypothesis 3, group differences were significant (F = 8.39, p <.001). AttM was most positive for incongruity resolution ( $\bar{x}$ = 1.16, SD = 0.87), followed by congruity ( $\bar{x}$ = 1.03, SD = 0.76), and negative for incongruity nonresolution ( $\bar{x}$ = −0.43, SD = 1.81). However, post hoc tests (Tukey) indicated a nonsignificant mean difference between incongruity resolution and congruity (p = .84). Therefore, Hypotheses 3a and 3b were not supported but Hypothesis 3c was supported (see Table 3).

Table 3.

Hypothesis 3 Findings: Attitude Toward Message for Schema Congruity/Incongruity Resolution/Incongruity Nonresolution.

Variable	Group	Mean	SD	F	Sig. (p)
Attitude toward message (Study 1)	Congruity	1.02	0.76	8.39	<.001
	Incongruity resolved	1.16	0.87
	Incongruity not resolved	–0.43	1.81
Attitude toward message (Study 2)	Congruity	1.85	0.97	112.78	<.001
	Incongruity resolved	1.19	1.09
	Incongruity not resolved	–0.23	1.05

Note. ANOVA conducted for group means for attitude toward message.

Study 2: Effect of Congruity/Incongruity on Attitude Toward FL Message

In Study 1, the sample size for the group incongruity nonresolution was small and lacked sufficient statistical power for the test. Hence, the purpose of Study 2 was to reexamine the lack of support for AttM using a larger national sample size.

In Study1, AttM for incongruity nonresolution was significantly less favorable than congruity and incongruity resolution, suggesting unresolved incongruity as a negative influencer of AttM. However, AttM for congruity and incongruity resolution was not significantly different. The lack of support for Hypotheses 3a and 3b in Study 1 led to additional empirical tests in Study 2. Therefore, Hypotheses 3a, 3b, and 3c were hypothesized and tested again in Study 2.

Method

Research Design, Stimuli Selection, and Measures

After the conclusion of Study 1, the same experimental design and stimuli as Study1 were used for Study 2. Independent variable schema and dependent variable AttM were measured the same as in Study 1.

Sample Selection and Procedure

Five hundred adult U.S. consumers (controlling for equal gender distribution) were recruited for an online experiment using a market-based research firm, Qualtrics. Participants completed demographic and preschema questions and were exposed to two messages (each timed for 30 s) from two randomly selected brands. Participants were only exposed to messages from two out of four brands to incorporate message variance yet avoid fatigue. After each message, participants answered postschema and attitude questions. Three attention filter questions were included to ensure participants’ attention toward the online study: Participants typed “survey” and selected “strongly disagree” on two occasions. Data for participants who inaccurately answered the questions were removed from analyses, leading to 466 usable responses. Participants took an average of 13 min to complete the study.

Data Analyses

As done in Study 1, participants were categorized into three groups: schema congruity (n = 265), incongruity resolution (n = 135), and incongruity nonresolution (n = 66). Although unequal, all cell sizes were sufficient (at least n = 20) for minimum statistical power (Hair, Anderson, Tatham, & Black, 1998). Multiple one-way ANOVAs were conducted to test study hypotheses.

Results

Respondent Profile

Participants’ ages ranged from 18 to 70 (24.4% aged above 60). Participants represented a diverse mix of ethnicities including 419 (83.8%) Caucasians, 42 (8.4%) African Americans/Blacks, 19 (3.8%) Hispanics/Latinos, nine (1.8%) Asians, and 11 (2.2%) of other ethnic origins. According to the 2012 estimate of the U.S. Census Bureau, the U.S. population comprised 63% Whites (not Hispanic), 16.9% Hispanic or Latino origin, 13.1% African Americans/Blacks, 5.1% Asians, and 2.2% others. Moreover, 13.7% of the population was 65 years or older. Detailed participant characteristics are shown in Table 1.

Hypotheses Tests

The homogeneity assumption was met for conducting ANOVA (Levene’s statistic [2, 463] = 2.26, p = .106). For AttM (Hypothesis 3), there were statistically significant group differences (F = 112.78, p < .001). AttM was most positive (favorable) for schema congruity ( $\bar{x}$ = 1.85, SD = 0.97), less positive (favorable) for incongruity resolution ( $\bar{x}$ = 1.19, SD = 1.09), and negative (unfavorable) for incongruity nonresolution ( $\bar{x}$ = −0.23, SD = 1.05) with significant mean differences (p < .001), supporting Hypothesis 3 (a, b, and c). Contrary to Study1, where means between congruity and incongruity resolution were not statistically different, Study 2 results indicated that congruity accounted for more positive AttM than incongruity resolution. Combined findings of Study 1 and Study 2 are presented in Figure 2.

Figure 2.

Findings from Studies 1 and 2 showing the effects of brand schema-message congruity/incongruity on attention, message recognition, and attitude toward message. The figure above summarizes the study findings. The ± signs indicate the strength of mean attitude toward message as hypothesized. *Represents supported study hypotheses. **Hypotheses 3a and 3b are not supported in Study1 but supported in Study 2.

Conclusions

In this study, the authors examined how consumers pay attention to, recognize, and form attitude toward apparel brands’ FL brand messages that are congruent/incongruent with their existing brand schemas. The study findings have several contributions and implications. The authors used two data collection methods (psychophysiological and self-reported) and three theoretical frameworks (schema congruity theory, LC4MP, and signal detection) across two experimental studies to arrive at a framework for understanding cognitive and attitudinal effects of congruent or incongruent apparel brands’ FL communications. Therefore, the combined study findings bridge the gap in the apparel brand literature between human attitude and cognition, helping both apparel brands and consumer researchers understand consumers’ reactions to schema-message congruity/incongruity and guide decisions when brands hope to revitalize or reinvent their image. Specific implications and contributions of the study are mentioned below.

First, the study has theoretical contributions. Mandler’s schema congruity theory (1982) focuses on the effects of congruity/incongruity on message evaluation. The authors extend Mandler’s schema congruity theory by investigating cognitive processes in the case of congruity and incongruity in addition to the effects of such congruity/incongruity on attitudinal responses. Contrary to Mandler’s (1982) proposal, congruity between brand schema and message resulted in more positive AttM. Rather, the study findings were similar to those of Dahlén and Lange (2004), who explained that brand schemas may be formed over time and could be extremely strong and personal. These strong personal schemas lead consumers to favor a congruent message over an incongruent one. This indicates a boundary condition, where application of the propositions of Mandler’s schema congruity theory may be limited and thereby call for a new theory to explain consumer schemas, messages, and attitude formation beyond Mandler’s theory.

Second, the authors offer insights into schema congruity theory. The majority of researchers have focused on novelty elements within the message itself that would challenge existing consumer schema to elicit favorable responses. Although a few researchers have looked at incongruity between brand message and brand image, no researcher has looked at incongruity between consumer expectations toward brands’ FL efforts and brand messages. While in such studies incongruity might have led to positive responses (Dahlén & Lange, 2004), according to this study, the effect of incongruity between apparel brands’ FL efforts and FL messages was rather negative or neutral, with greater message recognition. That is, not all novel incongruent messages would lead to positive consumer responses. This calls for a new theory on why consumers have different responses to unexpectedly good FL messages.

Third, heart rate deceleration pattern indicated that participants’ attention increased for incongruent FL messages more than for congruent ones. This implies that once consumers do not expect an apparel brand to employ FL practices, the brand’s FL messages might increase consumer scrutiny to reconcile the new information and to resolve incongruity. Therefore, it is important for apparel brands that are trying to change consumers’ existing schemas or recover from negative reputations to be extra cautious in designing their FL messages—even if they are positive. Also, participants better recognized content from incongruent FL messages compared to congruent ones. The findings are similar to existing research indicating that incongruity leads to better recall, higher number of thoughts, and increased time taken by participants (Lane, 2000; Sujan, 1985). Therefore, even though certain apparel brands might have a less than desirable schema related to their FL efforts, their FL marketing messages might be better encoded by consumers and hence better remembered. Thus, in evaluating subsequent FL marketing communications from the brand or in real-life purchase situations, consumers might be able to recognize apparel brands’ FL efforts mentioned in their previous marketing messages, ultimately helping the brand establish its revitalized FL-related schema in consumers’ minds.

Finally, according to the study results, attitude toward FL messages was contingent upon congruity with consumers’ schemas or their ability to resolve incongruity, instead of the FL message itself. AttM was positive for FL messages when incongruity was resolved and negative when not. Therefore, to improvise existing brand schemas, apparel brands have to design persuasive FL messages that are able to resolve consumers’ incongruity. In that case, even messages from apparel brands with negative FL-related schema might be evaluated positively by consumers. However, contrary to Mandler’s theory (1982), the authors did not find that consumers evaluate FL messages more positively when incongruity is revolved than congruity. Rather, if consumers believe an apparel brand to practice FL efforts, their existing schema can be expected to be positive, which can then be transferred to their subsequent evaluation of the brand (Lee & Schumann, 2004). Conversely, any conflict or incongruity, even if the message contains positive notes, can be regarded as disturbing, even when the incongruity is resolved, leading to less favorable evaluation. Many of today’s consumers are concerned about the social impact of their purchase choices and often prefer socially responsible brands (Nielsen, 2015). Thus, when these consumers believe that a brand should exercise FL practices but find out it does not, their schema toward that brand can be negative, subsequently negatively affecting their brand evaluation. These findings offer important implications for apparel brand managers who many establish or strengthen the brand’s FL-related reputation.

The study is not without limitations. First, the authors measured only positive/neutral (and not negative) updates of schema, reflecting the belief that many of today’s FL messages are often considered “too good to be true.” The stimuli comprised positive FL messages from brands considered to employ FL efforts. Thus, in the future researchers should consider including both directions of schema update to account for the effect of valence might be beneficial. Moreover, although consumers’ preschemas can be influenced by information from multiple sources, the authors only assessed how apparel brands’ FL messages influence consumers’ evaluation of FL messages. Therefore, in the future involving multiple sources of information might be beneficial.

Second, psychophysiological methods often suffer from a lack of ecological validity due to the data collection method of electrodes and specific laboratory conditions (Lang et al., 1993). Therefore, generalizations of study findings should be made with caution. Moreover, due to the presence of text in stimuli messages, heart rate deceleration patterns might have been influenced by both the encoding of information into memory and the retrieval of information from memory (Lang, 2000). However, the purpose of the study was to investigate differences in heart rate deceleration patterns not between memory and retrieval, but between congruity and incongruity conditions, which was apparent in the study results. Investigating specific effects of text and image will be beneficial in future studies. In addition, prescreening participants based on their schema scores to arrive at more comparable cell sizes might be beneficial.

Third, in this study, the authors measured the changes in respondents’ schemas immediately after message exposure. Some might argue that consumer schemas cannot be changed overnight by being exposed to a single brand message. However, marketing messages are usually targeted at making consumers form new schemas by informing them, reinforcing their existing schemas by reminding them, and reinforcing or altering their schemas by acting as “resolution messages or hints,” thus persuading them (Lee & Schumann, 2004, p. 75). Therefore, measuring change of schema immediately as well as after a considerable time delay might be beneficial in future research. Finally, schema was the only predictor of consumer attitude toward FL messages. Additional studies including other variables, such as consumers’ general expectations about apparel brands’ FL initiatives and general perceptions toward FL claims that can influence attitudes toward the claim, as well as including brand familiarity as a moderator, would be important. In addition, replication of the current study using additional measures of attention and recognition to further validate the current study findings would be essential.

Footnotes

Appendix

Acknowledgments

We would also like to thank Dr. Jana Hawley at the University of Arizona and Dr. Pamela Norum at the University of Missouri for their valuable insight. The study would not have been the same without their contributions.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by graduate student research grant awarded by the Center for Digital Globe at the University of Missouri.

References

Abdi

(2003). Factor rotations in factor analyses. In Lewis-Beck

Bryman

Futing

(Eds.), Encyclopedia for research methods for the social sciences (pp. 792–795). Thousand Oaks, CA: Sage.

Bangladesh Factory Collapse Toll Passes 1,000 (2013, 5 10). BBC News. Retrieved from http://www.bbc.com/news/world-asia-22476774

Bolls

P. D.

Lang

Potter

R. F.

(2001). The effects of message valence and listener arousal on attention, memory, and facial muscular responses to radio advertisements. Communication Research, 28, 627–651.

Brown

S. P.

Stayman

D. M.

(1992). Antecedents and consequences of attitude toward the ad: A meta-analysis. Journal of Consumer Research, 19, 34–51.

Burton

Lichtenstein

D. R.

(1988). The effect of ad claims and ad context on attitude toward the advertisement. Journal of Advertising, 17, 3–11.

Dahlén

Lange

(2004). To challenge or not to challenge: Ad-brand incongruency and brand familiarity. Journal of Marketing Theory and Practice, 12, 20–35.

Dahlén

Lange

Sjödin

Törn

(2005). Effects of ad-brand incongruence. Journal of Current Issues and Research in Advertising, 27, 1–12.

Dickson

M. A.

(2001). Utility of no-sweat labels for clothing consumers: Profiling label users and predicting their purchases. The Journal of Consumer Affairs, 35, 96–119.

Fashion United. (2014). Global fashion industry statistics-International apparel. Retrieved from https://www.fashionunited.com/global-fashion-industry-statistics-international-apparel

10.

Goodstein

R. C.

(1993). Category-based applications and extensions in advertising: Motivating more extensive ad processing. Journal of Consumer Research, 2, 87–99.

11.

Hair

J. F.

Jr Anderson

R. E.

Tatham

R. L.

Black

W. C.

(1998). Multivariate data analysis (5th ed.). Upper Saddle River, NJ: Prentice Hall.

12.

Hutter

R. R.

Lawton

Pals

O’Connor

D. B.

McEachan

R. R.

(2015). Tackling student binge drinking: Pairing incongruent messages and measures reduces alcohol consumption. British Journal of Health Psychology, 20, 498–513.

13.

Hyllegard

K. H.

Yan

R. N.

Ogle

J. P.

Lee

K. H.

(2012). Socially responsible labeling: The impact of hang tags on consumers’ attitudes and patronage intentions toward an apparel brand. Clothing and Textiles Research Journal, 30, 51–66.

14.

Jarvis

B. G.

(2014). MediaLab (Version 2014.1) [Computer Software]. New York, NY: Empirisoft Corporation.

15.

Kozar

J. M.

Hiller Connell

K. Y.

(2013). Socially and environmentally responsible apparel consumption: Knowledge, attitudes, and behaviors. Social Responsibility Journal, 9, 315–324.

16.

Lane

V. R.

(2000). The impact of ad repetition and ad content on consumer perceptions of incongruent extensions. The Journal of Marketing, 64, 80–91.

17.

Lang

(2000). The information processing of mediated messages: A framework for communication research. Journal of Communication, 50, 46–70.

18.

Lang

Bradley

S. D.

Park

Shin

Chung

(2006). Parsing the resource pie: Using STRTs to measure attention to mediated messages. Media Psychology, 8, 369–394.

19.

Lang

P. J.

Greenwald

M. K.

Bradley

M. M.

Hamm

A. O.

(1993). Looking at pictures: Affective, facial, visceral, and behavioral reactions. Psychophysiology, 30, 261–273.

20.

Lee

E. J.

Schumann

D. W.

(2004). Explaining the special case of incongruity in advertising: Combining classic theoretical approaches. Marketing Theory, 4, 59–90.

21.

Leshner

Bolls

Thomas

(2009). Scare’em or disgust’em: The effects of graphic health promotion messages. Health Communication, 24, 447–458.

22.

Leshner

Bolls

Wise

(2011). Motivated processing of fear appeal and disgust images in televised anti-tobacco ads. Journal of Media Psychology: Theories, Methods, and Applications, 23, 77–89.

23.

Mackenzie

S. B.

Lutz

R. J.

(1989). An empirical examination of the structural antecedents of attitude toward the ad in an advertising pretesting context. Journal of Marketing, 3, 48–65.

24.

Mandler

(1982). The structure of value: Accounting for taste. In Clark

M. S.

Fiske

S. T.

(Eds.), Affect and cognition: The 17th annual Carnegie Symposium (pp. 3–36). Hillsdale, NJ: Lawrence Erlbaum.

25.

Nielsen. (2015, 11 5). Green generation: Millennials say sustainability is a shopping priority. Retrieved from http://www.nielsen.com/us/en/insights/news/2015/green-generation-millennials-say-sustainability-is-a-shopping-priority.html

26.

Potter

R. F.

Bolls

(2012). Psychophysiological measurement and meaning: Cognitive and emotional processing of media. New York, NY: Routledge.

27.

Shapiro

M. A.

(1994). Signal detection measures of recognition memory. In Lang

(Ed.), Measuring psychological responses to media (pp. 133–148). Hillsdale, NJ: Lawrence Erlbaum.

28.

Sjödin

Törn

(2006). When communication challenges brand associations: A framework for understanding consumer responses to brand image incongruity. Journal of Consumer Behavior, 5, 32–42.

29.

Sujan

(1985). Consumer knowledge: Effects on evaluation strategies mediating consumer judgments. Journal of Consumer Research, 12, 16–31.

30.

Thorson

Wicks

Leshner

(2012). Experimental methodology in journalism and mass communication research. Journalism and Mass Communication Quarterly, 89, 112–124.

31.

Törn

Dahlén

(2008). Effects of brand incongruent advertising in competitive settings. European Advances in Consumer Research, 8, 234–239.

32.

U.S. Department of Labor. (2016). Compliance assistance-Wages and the Fair Labor Standards Act (FLSA). Wage and Hour Division. Retrieved from https://www.dol.gov/whd/flsa/

33.

Vanhamme

Snelders

(2001). The role of surprise in satisfaction judgments. Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, 14, 27–45.