Measurement of emotion
As the historical review showed, emotions are not an easy concept to study. The scientific measurement of this phenomenon is one of the most challenging aspects of any scientific endeavor. Through the years, the advertising discipline not only has utilized different methodologies of measurement but also has created its own.
Following Holbrook and O'Shaughnessy (1984), Kroeber-Riel (1986), and Wiles and Cornwell (1990) this review of advertising's measuring methods classifies them in two wide areas: verbal self reports and nonverbal (psychophysiological) techniques.
Verbal self reports
The methods that use written or oral descriptions given by a subject are the most common in the advertising field (Wiles and Cornwell 1990). The basic assumption required to use verbal reports is that emotional experiences can be described and communicated verbally. This assumption is an influence of psycholinguistics that asserts that any emotional experience is extensively language-dependent (Kroeber-Riel 1986).
The verbal measures of emotion suffer from unavoidable weaknesses. One possible flaw is that they tap the evaluative content of cognitive evaluation but they may do little to measure accompanying physiological changes (Holbrook and O'Shaughnessy 1984). This may be a serious flaw because much of the mental activity created by advertising may work at an unconscious level (Wiles and Cornwell 1990). Another argument against these techniques is that many emotional experiences that may affect a subject are sensible topics and it is likely that he will not accede to talk about them (Kroeber-Riel, 1986).
Wiles and Cornwell (1990) classify the basic verbal report techniques as: (a) Typology studies, (b) investigations which develop and use a set of items to measure response, and (c) studies which use magnitude scaling methods that differ from typical verbal rating scale techniques.
(a) Typology studies
Many of the researchers consider emotion a multidimensional phenomenon, therefore, the development of typologies is one of the most promising methods for understanding emotions (Wiles and Cornwell 1990).
An example of this methodology is the study done by Stout and Leckenby (1986). In their report, they present and test three levels of emotional response: descriptive, emphatic and experiential. The first level is the ability to recognize emotions expressed by others. The second level is related to an emphatic response from the audience, that is they feel what the character in a situation is feeling. Finally, the third level, experiential, is a valenced emotional response that occurs as a reaction to self-relevant experiences.
The major criticisms of this methodology are that the data collection method can not reflect all the possibilities of an emotional response. Also, the method is obtrusive per se, so it is common to have biased outcomes. In addition, some studies do not discriminate between cognitive responses and affective responses. Finally, the majority of the studies are executed in laboratory setting that do not replicate a natural environment (Wiles and Cornwell 1990).
(b) Development of measure response
The methodology uses some type of verbal rating scale, like a Likert-type scale, or semantic differential. An example of the former is the AIM (affect intensity measurement) questionnaire (Larsen and Diener 1987). This battery of questions measures the self-perceived strength of a subject's response to certain stimuli. A sample of questions of this questionnaire is presented:
Never = 1 Almost never = 2 Occasionally =3 Usually = 4 Almost always = 5 Always = 6
- 1. ___ When I accomplish something difficult I feel delighted or elated.
- 2. ___ When I feel happy it is a strong type of exuberance.
- 3. ___ I enjoy being with other people very much (Larsen and Diener 1987: 34).
- And so forth.
- 2. ___ When I feel happy it is a strong type of exuberance.
If you are interested in a thorough description of Larsen's theory and questionnaire read Larsen, Randy J. and Ed Diener (1987) "Affect Intensity as an Individual Difference Characteristic: A Review," Journal of Research in Personality, 21(1), 1-39. The other verbal rating scale, semantic differential, uses bipolar adjectives on a seven or five-point scale to measure beliefs, emotions or feelings (Assael 1995).
An example is the research done by Bagozzi and Moore (1994). In this study, an audience saw two PSAs (Public Service Announcements) with different appeals. After watching them, the members of the audience answered a questionnaire that measured the strength of the emotional reactions by a 5 point scale:
Very Angry ____ ____ ____ ____ ____ Not at all angry Very Sad ____ ____ ____ ____ ____ Not at all sad Very fearful ____ ____ ____ ____ ____ Not at all fearful Very tense ____ ____ ____ ____ ____ Not at all tense These two verbal rating scales are the most widely used because they are simple to construct, administer and code (Assael 1995). However, by offering a specific number of response categories, the respondent has a limit to the range of his emotional response. Another problem that suffers these techniques is the artificiality of the settings where the research is done (Wiles and Cornwell 1990).
(c) Magnitude scaling methods
In magnitude scaling studies, subjects are presented with sets of sensory stimuli and are asked to assign a number to the noticed magnitude of each stimulus in relationship to a verbal or numerical anchor (Wiles and Cornwell 1990).
One of the most relevant measure methodologies in advertising is the warmth monitor (Aaker, Stayman and Hagerty 1986). In this technique a respondent moves a pencil down paper while viewing an advertisement, moving it to the left or right to reflect his feelings at any given time.
Source: Aaker, David A., Douglas M. Stayman and Michael R. Hagerty (1986), "Warmth in Advertising: Measurement, Impact, and Sequence Effects", Journal of Consumer Research 12, (March), 365-381.
The limits of this methodology are that the magnitude of a feeling is measured by the perception of a subject, thus construct validity is questioned and in some cases the researcher uses obtrusive electronic devices such as potentiometers that may be distracting (Wiles and Cornwell 1990).
Nonverbal techniques
From the first application of psychophysiological measure like the Galvanic Skin Response (GSR) to the current use of a combination of techniques like heart rate, electromyography and verbal self reports, these methodologies had an impact in the advertising field.
Kroeber-Riel (1986) conceives verbal response as a limited methodology because conscious reporting of emotion is an undeveloped verbal ability and some emotions are linked to memory images that cannot be verbalized. Thus, it is necessary to apply nonverbal methods like brain wave analysis, facial EMG activity, pupillary responses, skin responses, voice analysis, heart rate and natural overt bodily responses.
While these techniques enjoy some success, they are difficult to administer and are not capable to establish specific emotions (Aaker, Stayman and Hagerty 1986). Also, these measures may prove undependable indicators of affective responses to ads, failing to reflect the valence of emotional reactions. Some authors have suggested that maybe the only valid use of nonverbal methods is to assess the intensity of emotional responses to commercials (Holbrook and O'Shaughnessy 1984).
Another big obstacle for psychophysiological studies is that this kind of reactions may represent a number of processes. That is, the data is analyzed as inputs and outputs without considering what are the processes that work inside the subject (Wiles and Cornwell 1990).
One of the most used measures is the electrodermal response, specially the skin resistance response or galvanic skin response. The mechanism of this method is simple, the skin's electrical resistance is affected by the activity of the sweat gland that are controlled by the sympathetic nervous system (Wiles and Cornwell 1990). The measure of that change can be seen as an expression of intensity of the emotional behavior (Kroeber-Riel 1986). However, it is needed to realize that the method is limited to measure the intensity of emotions.
As in other fields, the thoughtful and respectful combination of methods is the best alternative to develop an area like emotion in advertising. One researcher who is applying this approach is Hunt (1989). In his study, Hunt concluded that:
- (a) Self-reported emotion was more able to recognize believability
and comprehension differences.
- (b) Psychophysiological measures were more capable of finding differences in recognition.
- (c) Both types of techniques may be needed to measure recall.
- (b) Psychophysiological measures were more capable of finding differences in recognition.