Emotion questionnaires: A consumer-centric perspective

Highlights

• Discusses whether emotions are elicited by emotion questionnaires.

• Completion of product-elicited emotion questionnaires is “easy, but a bit weird.”

• Participants can struggle to connect food stimuli to questionnaire emotion words.

• When free listing, participants often elicit only 1–2 emotion words per food item.

Abstract

Interest in using questionnaires to measure emotions elicited by consumption of foods/beverages and use of personal/household care products is growing, and the investigation of methodological issues linked to the use of such questionnaires is therefore of relevance and interest to sensory and consumer scientists. Using a multi-method approach, Part 1 of this research presents a participant-centric perspective on “emotion” questionnaires. Qualitative data were obtained from ∼300 New Zealand consumers in response to: (i) a range of product categories; (ii) food stimuli presented as actual samples or as food names; and (iii) different “emotion” measurement scales, but predominantly the EsSense Profile®. Many participants found the EsSense Profile® task to be easy and intuitive. However, some participants also found that the task was a bit weird/strange/odd. Specific misgivings that were expressed included not feeling any of the “emotions” appearing on the list and not understanding why the test stimuli would be expected to lead to those “emotions”. Instructions aimed at lessening these misgivings presented prior to completing the EsSense Profile® were only partially effective at reducing participants’ difficulty in relating the food stimuli to some of the emotion terms. In Part 2 of this research, ∼40 participants were presented with food names and actual food samples, and asked to write down their emotional responses to these stimuli using a free listing method. These data were compared with EsSense Profile® responses collected previously for the same stimuli. It was found that participants elicited many fewer emotion words than the 39 included on the EsSense Profile®. Further, there was only a partial overlap between the emotion words generated by free listing and those contained in the EsSense Profile®. While the research raises as many questions as it answers, it contributes to a needed discourse on what are the responses obtained with “emotion” questionnaires such as the EsSense Profile®.

Introduction

The study of the emotional response to foods and beverages has become an important area of investigation in industry and academia. The growing interest in this area is due to both the potential for emotional responses to provide important information about products and differences between a set of products that go beyond traditional sensory and liking variables, and the importance of emotional responses to serve as critical tools by which to market specific categories and brands of foods/beverages.

In order to address the need for quantification of emotional responses, a number of techniques have been developed to capture the varied manifestations of human emotion. Among these approaches are physiological methods that aim to tap the underlying biological responses that accompany emotions (Lewis et al., 2008, Sequiera et al., 2009, Krebig, 2010), facial recognition methods that capture and categorize the facial muscle movements that accompany many emotions (Ekman and Friesen, 1978, Izard, 1979, Zeinstra et al., 2009) and questionnaire techniques that assess the experiential aspect of emotions (Laros & Steenkamp, 2005; Desmet and Schifferstein, 2008, Thomson, 2010, King and Meiselman, 2010, Thomson et al., 2010). These latter aspects of emotions are often referred to as “feelings” (Damasio, 2001). The present research focuses on questionnaire approaches, as it is the most common method for assessing emotional responding to foods/beverages in a wide range of product development applications.

If one examines the literature in which questionnaire approaches have been used to assess emotional responding to foods and products, there are a very large number of feelings and emotion-related terms that have been (or can be) used. Laros and Steenkamp (2005) compiled a list of 316 different English emotion words that could be used for this purpose. Similarly, Rousset et al. (2005) identified 237 possible French emotion words. Yet, both of these lists contain fewer words than other lists that have appeared in the general emotion literature, e.g. the 590 words examined by Johnson-Laird and Oatley (1989). Considering only English words, the terms compiled by Laros and Steenkamp (2005) range widely, from those that theorists have classically defined as basic emotions, e.g. “fear,” “shame,” “jealousy,” “sadness,” and “anger” to terms that are associated more with the physiological correlates of emotional responding, e.g. “nervousness,” “energetic,” “tense,” “alert” and “stimulated.” Other words, although not considered classic emotions, are more generally recognized as emotion-like experiences or “feelings”, e.g. “disappointment,” “dread,” “inspiration,” “pity,” and “respect.” However, one also finds a variety of words that seem less like classic emotions and more like personality traits or personal characteristics. Examples of such words include “self-conscious,” “unfulfilled,” “helpless,” “encouraged,” “sexy” and “warm-hearted” (Laros & Steenkamp, 2005). If one looks at other common emotion questionnaires used in the literature, one can also find such personality-based terms, e.g. “steady,” “good-natured,” “friendly,” and “warm” (King & Meiselman, 2010) or “arrogant,” “masculine,” and “sociable” (Thomson et al., 2010), as well as other terms that seem more like image-based characteristics of products, e.g. “mild,” “free,” and “whole,” (King & Meiselman, 2010) or “ordinary,” “comforting,” and “luxurious” (Thomson et al., 2010).

The wide array of terminology that has been used in emotion questionnaires for product testing raises the question of “what constitutes an emotion?” In 1884 the noted psychologist, William James, raised this question in an essay aptly entitled “What is an emotion?” (James, 1884). In his paper, James proposed that emotions were sensations or sets of sensations that resulted from a physiological response to a stimulus. James noted that “we feel sad because we cry, angry because we strike, and afraid because we tremble.” His work placed an important focus on two main elements of emotion – the first being a physiological component, the second being a perceptual or experiential component. Subsequent work on emotions by Schachter and Singer (1962) emphasized the fact that the physiological response to any stimulus is dependent upon a cognitive appraisal of both the stimulus and the context in which the physiological response to the stimulus occurred. For example, the physiological response to something thrown in one’s face could be shock or anger if the stimulus and context is mud thrown by a stranger in an alley. Alternatively, the same physiological response could be appraised as surprise or amusement if the stimulus and context is whipped cream thrown by a friend at your birthday party. This addition of a third, cognitive component to emotions led to further elaborations of the concept of cognitive appraisals and the elucidation of still another element of emotions – “intentionality” or “action tendency”. As reflected in the theories of Lazarus (1982), Roseman (1984) and Frijda (1986), action tendencies are considered to be states of readiness to execute a given action. Thus, fear readies the individual to flee, while happiness motivates the tendency to approach.

Today, most emotion theorists view emotions as complex response pattern that includes experiential, physiological, cognitive and intentional elements (Omdahl, 1995, Scherer, 2005). This gradual broadening of the components of emotion has resulted in a corresponding broadening of our conceptions about what constitutes an emotion, because the simple physiological component of an emotion can be experienced in a wide variety of ways, depending upon how the physiological experience is cognitively appraised within the overall situational context in which it occurs and the varied action tendencies that may be associated with these situations. This has led to a state in which there is now little agreement among researchers as to what constitutes a “basic emotion” and/or their number (Ortony and Turner, 1990). In fact, there are more likely to be large “families” of emotions that may be characterized in a more general sense under the broad rubrics commonly referred to as “basic emotions” (Ekman, 1999). Russell (2003), in an attempt to simplify this complex constellation of emotions, argued that these broad families of emotions are constructed from two fundamental elements of what he calls “core affect” – valence (positive, negative) and arousal (activation, deactivation).

The gradual broadening of the concept of emotions, away from that of a small set of “basic” emotions, has led to the wide variety of emotion words that are currently found in emotion questionnaires. In practice, while there are hundreds of emotion-related words that can be included on questionnaires, the actual number used by most investigators is limited to the number of words that can be reasonably listed in a checklist or scaled by the consumer. Thus, the number of words on most “emotion” questionnaires used with consumers ranges from 20 to 40. Desmet and Schifferstein (2008) used 22, Laros and Steenkamp (2005) used 33, Pineau et al. (2010) and Chrea, Grandjean, and Delplanque (2009) used 36, while King and Meiselman (2010) use 39 in their EsSense Profile®. The fact that the large number of possible emotion words is often reduced to a relatively smaller number raises the question of how this reduction occurs. In many cases, there may be a theoretical rationale underlying the choice (e.g., Desmet, 2002; Laros & Steenkamp, 2005). In other cases, the choice is more practical, related to how well the terms apply to foods or are used by consumers to describe food (King & Meiselman, 2010; Desmet and Schifferstein, 2008). In still other cases, the words are chosen for their general association with human feelings, attitudes, and associations to food and their likely relevance to positioning product brands within the context of today’s consumer interests. Thomson and colleagues’ work (Thomson, 2010, Thomson et al., 2010) makes this clear. In a very practical interpretation, Thomson characterizes these words not strictly as emotions, but as “associated emotional conceptualizations” that may have “either immediate or eventual emotional connotations” (Thomson et al., 2010). He separates these “emotional” associations from other product associations that may be “functional” and further posits that these associations may be based on “internal experiences” or may be “learned” (Thomson et al., 2010). Further, he argues that these conceptualizations can coalesce with the identity of the food item to “become as one in the mind of the consumer” (Thomson et al., 2010). In still more recent work, Thomson and Crocker (2013) refer to “feelings,” which they define as the subjective, affective experience of an emotion or mood.

It seems clear from the above that, while there is an evolving theoretical concept of what constitutes an emotional response, there is little consensus on what constitutes a basic emotion, how many different emotions there are, or what the number or nature of words should be to assess “emotional” responses to foods and beverages. It is also clear that, in the absence of any resolution to the above issues, a wide variety of “emotion” words are currently being used on a regular basis to extract data from consumers about products, without regard to either theoretical considerations of what constitutes an emotion or to the distinguishment among “emotions,” “feelings,” “associations,” etc. At the same time, all of this is transpiring with relatively little consideration of what may going on in the mind of the consumer when he/she is presented with a long list of quite varied words that they must apply to their feelings about foods, typically without any accompanying definition or meaningful context.

In light of the failure among researchers using questionnaire approaches to adequately distinguish among the words that they use as to which are “emotions”, which are “feelings” and which are “associations”, in the present paper we will simply use the term “emotion words” to refer to the wide range of words that have been used in the literature to describe the “emotions,” “feelings” or “emotional associations” that individuals may have in response to a food, food name, or other food-related stimulus (labels, images, etc.).

The above confusion about the terminology used in emotion questionnaires raises a variety of questions. What is the overall task like for the consumer? Is the task easy to complete? Is it as easy as making liking judgments? Is the scaling of emotions as easy as the scaling of hedonics? How are the words being interpreted by the consumer? Are they meaningful to them? Do the words relate to actual emotions that consumers experience at the time the food is tasted? Or are they cognitive associations to these foods? And lastly, how does the presentation of a long list of “emotion” words affect the obtained responses and would the same emotion words be elicited by a different method that did not cue for all of these words?

The above questions constitute the general focus of the present paper. In order to streamline our investigation, we have chosen a single questionnaire method to examine – the EsSense Profile® (King & Meiselman, 2010). This questionnaire was constructed from words contained in the MAACL-R (Zuckerman and Lubin, 1985), which is a clinically-oriented list of both “state” and “trait” affect words, and the POMS (McNair, Lorr, & Droppleman, 1971), another clinically-oriented list of transient mood states. Words from these lists were selected on the basis of their relevance for describing emotional or feeling responses to foods (see King & Meiselman, 2010).

We approach these questions using primarily qualitative methods that seek to uncover some of what is going on in the minds of consumers who are completing such a questionnaire, compare the findings to what occurs when consumers complete hedonic scaling or sensory characterization tasks, and contrast EsSense Profile® responding with responses obtained by a method that does not cue respondents with a long list of emotion words.

A series of experiments were conducted to further our understanding of the responses obtained with the EsSense Profile® (Table 1). Part 1 of the research consists of four experiments that collectively provide a participant-centric perspective on what it is like to complete the EsSense Profile®. To provide convergent validity, methodological triangulation (Denzin, 2006, Cohen and Manion, 2000) was accomplished using three different methods for data collection: think-aloud, thought-listing and open-ended questions. To provide comparative data with which to compare the emotion word task, think-aloud data for two additional common research tasks were obtained: hedonic scaling and sensory product characterization (CATA method). To ensure that results could be more readily generalized, variations in execution of the emotion word task (some extending beyond the EsSense Profile®) were implemented, and a range of food stimuli (some chosen for experimental convenience) were used.

Part 2 of the research assessed the number and nature of emotion words used by consumers when completing the EsSense Profile® compared to those used when consumers are asked to simply list all of the emotion words that come to mind when evaluating/tasting a food stimulus. Experiments 2A and 2B were motivated by our lack of understanding of the demand characteristic (see Orne, 1962, Rosenthal and Rosnow, 2009 for reviews) created by presenting a long list of emotion words to consumers and asking them to rate (or check) all those that apply to the stimulus. The demand characteristic created by presenting a long list of emotion words would be expected to result in more emotion words being rated/checked than with a free listing method (Bernard, 2011; Hough & Ferraris, 2010; Ares and Deliza, 2010, Libertine et al., 2011).