Gaining perspective: The effects of message frame on viewer attention to and recall of osteoporosis prevention print advertisements

Abstract

This study examined how framed messages affect viewer attention to and cognitive processing of osteoporosis prevention print ads. Attention was measured with eye tracking technology. Cognitive processing was assessed through masked recall. A total of 60 college-aged women viewed 12 gain-framed, 12 loss-framed, and 12 neutral-framed ads. Number of fixations, dwell time, and recall of gain-framed osteoporosis prevention ads were higher than loss-framed or neutral-framed ads, p < .01. Message recall was positively correlated with the number of fixations and dwell time for the gain-framed and neutral-framed messages, p < .01. These findings provide preliminary insight into potential mechanisms underlying message framing effects.

Keywords

attention and cognitive processing eye tracking technology message framing osteoporosis prevention prospect theory

Osteoporosis, a disease characterized by low bone mass and increased risk of fracture, affects over 10 million Americans—80 percent of whom are women (National Osteoporosis Foundation, 2011). Osteoporosis prevention is particularly important for younger women because peak bone mass is usually achieved by the age of 30 years (Recker et al., 1992). Osteoporosis prevention behaviors include weight-bearing physical activity and consumption of adequate calcium and vitamin D (Feskanich et al., 2003). Unfortunately, most young women do not perform sufficient weight-bearing exercises and do not meet the recommended daily calcium intake levels (Kasper et al., 2001). Yet, they do not perceive themselves of being at the risk of osteoporosis (Kasper et al., 2001). This disconnect between perceived risk and lack of engagement in osteoporosis prevention behaviors is a serious public health concern (Gerend et al., 2006). As such, research is sorely needed to identify effective strategies for communicating osteoporosis prevention messages to this at-risk population. Delivering appropriately framed messages emphasizing either the costs or benefits of adopting or failing to adopt osteoporosis prevention behaviors is one communication strategy that holds promise (Jung et al., 2011).

Osteoporosis messages can be framed to emphasize the consequences associated with failing to perform a prevention behavior (loss-framed message) or the benefits associated with performing a preventive behavior (gain-framed message). Osteoporosis messages can also have a neutral frame stating only the facts, focusing neither on the inherent losses nor on the gains associated with taking or avoiding action. Based on the postulates of prospect theory, Tversky and Kahneman (1981) and Rothman and Salovey (1997) proposed that gain-framed messages are optimal for influencing prevention behaviors, such as physical activity, that have certain and risk-averse outcomes (i.e. there is a high likelihood that the outcome will occur as a result of engaging in the target behavior (certain) and engaging in the behavior has low or no risk of resulting in a negative outcome (risk averse)). A recent meta-analysis of framing studies corroborates this hypothesis (Gallagher and Updegraff, 2011).

In terms of osteoporosis prevention specifically, a study by Jung et al. (2011) indicated that young women perceive osteoporosis prevention as a means of decreasing their risk of disease (i.e. a risk-averse prevention behavior). As such, they were more persuaded to engage in osteoporosis prevention behaviors when exposed to gain-framed messages compared to control messages that included loss-framed message content. Jung et al. also examined social cognitive variables (e.g. self-efficacy) as potential mediators of framing effects. However, the meditational models were not supported, thus highlighting the need for additional research that identifies the mechanisms underlying the effects of gain-framed messages that encourage disease prevention behaviors.

Results from research examining social cognitive variables as a mechanism of framing effects are mixed. Cognitive processing has been suggested as an alternate mechanism such that messages framed in a manner congruent with message recipients’ construal of a behavior may be processed more thoroughly than incongruent messages (Rothman and Updegraff, 2011). It may be that individuals who construe a disease prevention behavior as risk averse or as having certain outcomes process gain-framed messages (i.e. congruently framed messages) more thoroughly than loss-framed messages. In support of this possibility, a meta-analysis by O’Keefe and Jensen (2008) revealed that compared to equivalent loss-framed messages, gain-framed messages encouraging disease prevention behaviors were better recalled and elicited more relevant thoughts, with outcomes indicating thorough cognitive processing of message content.

Moreover, in a recent study using eye tracking technology, Bassett-Gunter et al. (in press) evaluated attention to framed physical activity messages that were preceded by health risk information. The study revealed that attention to gain-framed physical activity messages was significantly greater than comparable loss-framed messages. However, the study did not include a neutral-framed message leaving it to be determined whether a gain or loss frame draws additional attention to or detracts attention from messages relative to neutral-framed messages. Whether the framing effects reported by Bassett-Gunter et al. emerge when framed messages are not preceded by health risk information also has not been examined. Investigating attention to framed messages in the absence of health information is a practically relevant research question as motivational health messages presented in print format (e.g. magazine advertisements) often present short slogans focusing only on the benefits or consequences of engaging or failing to engage in the target behavior.

Extending the study by Bassett-Gunter et al. (in press), the primary purpose of this study was to examine how college-aged women cognitively process gain-framed osteoporosis prevention messages relative to loss-framed and neutral-framed messages. Messages were presented in the form of advertisements. From an empirical perspective, demonstrating that message frame affects attention is a step toward establishing cognitive processing as a mechanism underlying framing effects. From a practical perspective, determining which message frame best attracts attention is a critical advancement in message framing research. Fully engaged audiences are more likely to thoughtfully attend to and consider the message content and may ultimately act upon the information (Krugman et al., 1994; Skumanich and Kintsfather, 1996).

Given that young women construe physical activity and calcium and vitamin D consumption as risk-averse osteoporosis prevention behaviors (Jung et al., 2011), and that meta-analytic evidence demonstrates a processing advantage for gain-framed disease prevention behaviors (O’Keefe and Jensen, 2008), we hypothesized that gain-framed messages would enhance attention to osteoporosis ads above loss-framed and neutral-framed ads. Because neither a loss-frame nor neutral-frame are congruent women’s construal of osteoporosis prevention behavior, we hypothesized that neutral-framed and loss-framed ads would be attended to similarly. Finally, we explored the relationship between attention to and recall of the advertisements viewed. Because attention is a critical first step in information processing (Wedel and Pieters, 2000), we hypothesized that higher attention would be associated with higher ad recall.

Method

Participants

Participants were a convenience sample of 60 women, M_age = 21.25 ± 2.61 years, recruited broadly from across a midsized Canadian university campus using posters, announcements in classes, and direct e-mails to students. The majority of participants were undergraduate students (68%), wore glasses or contact lenses (62%), and were Caucasian (63%).

Eligible participants were English-speaking women aged 18–35 years of age with self-reported-normal to corrected-normal vision, who perceived physical activity and calcium and vitamin D consumption as having risk-averse outcomes. These inclusion criteria were used because (1) 80 percent of the people who develop osteoporosis are women (National Osteoporosis Foundation, 2011); (2) after the age of 35 years, peak bone mass is thought to deteriorate; thus, if risk factors can be identified before the age of 35 years, it is possible to reduce the impact of osteoporosis (Osteoporosis Canada, 2011); and (3) this cohort is often at a transition period in life and amenable to new ideas regarding healthy living (Andreasen, 1984).

Experimental design

The study used a within-participant experimental design. Both ad frame and ad content were manipulated within participants. Thus, participants served as their own control. All participants viewed gain-framed, loss-framed, and neutral-framed osteoporosis ads and fashion and beauty advertisements. The fashion and beauty ads were included to disguise the study objective. A secondary analysis of attention patterns to commercial fashion and beauty ads relative to osteoporosis ads is described in a supplemental report (O’Malley et al., 2011). A within-participant design was selected for this experiment to control for individual differences that may affect attention to the ads (e.g. reading speed). A within-participant study design also is more powerful than a between-participant comparison, thus reducing the sample size requirements to detect the small-to-medium-sized effects characteristic of framed messages. Given the extensive data processing required to manage eye tracking data, minimizing sample size requirements is desirable. The protocol was approved by the Institutional Ethics Review Board.

Screening measures

To determine participants’ construal of the target behaviors as having either a prevention versus detection function, participants indicated whether they perceived participation in physical activity and consuming adequate amounts of calcium and vitamin D as health enhancing or illness detecting. Exploratory research demonstrates that individuals who perceive a behavior as health enhancing have a cognitive mindset conducive to gain-framed messages (Rothman et al., 2008). Participants also reported whether they had normal to corrected-normal vision and whether they were 18–35 years of age.

Cognitive processing measures

The Tobii 1750 eye tracker was used to objectively measure participant attention. This device resembles a flat screen computer monitor and uses near infrared light-emitting diodes (NIR-LEDs) to generate reflection patterns from the cornea of the participants’ eyes. These reflection patterns are collected by an integrated camera and sent back to a computer for mathematical processing, in which custom software calculates the three-dimensional spatial position of each eyeball, to detect where participants look. Because the integrated camera tracks the position of the eyes, it allows for continuous position shifts and head movements. Measurement precision is better than .5° of visual angle, with a head movement range of approximately 55 cm. These factors differentiate the Tobii from other tracking devices, making this eye tracker completely non-invasive and unobtrusive. The eye tracker recorded participants’ attention via the number of eye fixations and dwell times.

Fixations

Number of eye fixations is defined as moments when the eye is relatively stable because it has paused to gather visual information about a specific contextual element (Pieters and Wedel, 2007). In this study, number of fixations was calculated by summing the total number of times that a participant’s eyes focused on a presented ad. The number of fixations is considered to provide a reliable indication of a person’s attention. When viewers carefully attend to an ad, higher fixation numbers occur (Wedel and Pieters, 2000).

Dwell time

Dwell time is the total duration that participants looked at an ad element (Pieters and Wedel, 2007). Dwell time was measured by summing up the total amount of time that a participant spent viewing an ad. Dwell time is considered to be a prominent, aggregated measure of eye movement data (Rosbergen et al., 1997) and is a valid indicator of visual attention (Christianson et al., 1991). It is also thought to be a direct measure of cognitive processing (Fox et al., 1998).

Recall

In the masked-recall component of the study, participants were shown a copy of previously seen ads with the textual component missing and were asked to recall the missing information. Participants were not informed of the recall test ahead of time because knowing about the test could have affected viewing and processing of the ads (Anderson and Pichert, 1978; Rayner, 1998).

The accuracy of participants’ recall was judged and assessed using a coding scale similar to the scale described by Krugman et al. (1994). Responses were categorized into four levels of recall. These four levels were as follows: Level 1, no recall or completely inaccurate/incorrect response; Level 2, incorrect message frame or response containing less than two key words; Level 3, correct message frame and recalled at least two key words; and Level 4, completely accurate response and recalled exact wording of the phrase (grammatical discrepancies allowed). The masked-recall coding scale used by Krugman et al. (1994) had an intercoder reliability rating of .89. The reliability ratings for this study achieved similar results, yielding an intercoder reliability rating of .90.

Visual stimuli

Of these 36 advertisements, 24 were related to osteoporosis prevention behaviors. Within the 24 osteoporosis ads, 12 featured content pertaining to physical activity and 12 were related to calcium and vitamin D consumption. To disguise the study objective, the remaining 12 ads were unrelated to osteoporosis prevention and served as control ads. All the control ads featured fashion and beauty products.

The text displayed in all 36 ads was digitally manipulated, resulting in three similar versions of the same ad, with an identical image: a gain-framed version, a loss-framed version, and a neutral-framed version. Thus, the message frame was manipulated by modifying the text and not the image.

The advertisements shown were obtained from recent, popular consumer lifestyle and fitness magazines aimed at women audiences. These magazines included Elle, Cosmopolitan, Fitness, Women’s Health, and Glamour. Print ads were selected from the magazines and scanned on a flatbed scanner at 600 dots per inch (DPI), to ensure a consistently high reproduction quality. Using the PhotoShop CS image editing software, the ads were all resized to 5 in. (width) × 6.5 in. (height) at 100 DPI. Each image was then digitally manipulated to create the appropriate message frame version and saved in JPEG format.

To ensure consistency and uniformity across all three ads, the text positioning, size, color, and fonts remained the same across each ad set. As well, the length of the text was matched, as best as possible, across all ads and message frames. The mean word count for gain-framed messages was 13.13 ± 2.94 words, for loss-framed messages, it was 15.08 ± 3.21, and for neutral-framed messages, it was 15.17 ± 3.21. Despite trying to control for word count, a repeated measures analysis of variance (RM ANOVA) revealed a significant difference in word count between message frames F(2,46) = 20.12, p < .001. The Bonferroni post hoc analysis tests indicated that gain-framed ad text was significantly shorter than both loss-framed and neutral-framed ad text, p < .001. There was no significant difference in word count between loss-framed and neutral-framed ads, p > .05. Given that word count varied between message frames, it was controlled for by dividing the number of words by each participant’s number of fixations and dwell time, per ad. This calculation enabled number of fixations and dwell time to be standardized for word count across each message frame. The word count variable was controlled for in this manner because previous experiments have found that the effects of framing disappear when uneven amounts of information are presented (Arora and Arora, 2004). Moreover, all text were contained within one to three lines, as research has shown that readers are more likely to read all the presented text when it is less than three lines in length (Rayner et al., 2001).

To ensure that the text fulfilled its purpose of anchoring the image and persuasively motivating viewers, a separate image neutrality pilot test was conducted, prior to the message frame study. A group of 100 undergraduate students was shown the images (without the text) via a PowerPoint slide show and asked to rate the images on a 7-point scale ranging from very negative (1) to very positive (7). Results revealed that all images were perceived as neutral, with a mean score of 4.76 ± .64; none of the images were viewed as extremely negative (lowest value, 3.03 out of 7) or positive (highest value, 5.84 out of 7).

Following the pilot test, the ads were randomized so that no order effect arose. This randomization resulted in the three ad frame versions (gain framed, loss framed, and neutral framed) being equally dispersed among the three types of ad content (physical activity, calcium/vitamin D consumption, beauty, and fashion), yielding four ads of each format and content. This randomization was put in place because previous eye tracking research has shown that serial position, or order, can affect viewer attention to ads (Wedel and Pieters, 2000).

In addition, so that no viewing bias occurred, participants were randomly assigned to one of three subgroups (A, B, and C), each containing 20 people. All three groups viewed the same 36 ads; however, the message frame was counter-balanced across groups. For example, the first gain-framed ad viewed by Group A was seen as a loss-framed message by Group B and a neutral-framed message by Group C. Attention to and recall of each type of message was compared within participants. An optimal design would have been for the ads to randomly rotate format for each individual participant, rather than by group; however, technological constraints precluded this approach. In the current design, all participants viewed 8 gain-framed, 8 loss-framed, and 8 neutral-framed osteoporosis prevention messages.

Procedure

Women were invited to participate in a study examining how people look at magazine ads. They were also notified that their eye movements would be tracked while viewing the ads. Participants provided informed consent and completed the screening and demographic questionnaire. Questionnaire responses were reviewed immediately; all participants were eligible. Participants were then seated at the eye tracking device, received instruction about how it worked, and then completed a short calibration exercise to ensure accuracy of data collection. Following successful calibration, participants viewed the experimental stimuli. They viewed each ad freely, for as long as desired, as if they were flipping through a magazine at home or while sitting in a waiting room. Participants advanced to the next ad using a button press. Previous eye tracking studies have shown that eye movement patterns and memory performance are influenced by the goal of the task in which participants are engaged (Anderson and Pichert, 1978; Rayner, 1998). As such, the free-viewing option encouraged participants to view, or process, the ads without a specific goal in mind. In addition, the free-viewing condition provided a more ecological, as opposed to a research-based approach, because outside of a laboratory setting, people generally take as much time as desired when attending to ads.

At the completion of the study, participants completed a masked-recall exercise. Participants were given printed black and white photocopies from a random selection of 12 of the ads just viewed. There was a blank box in lieu of text. Participants were asked to fill-in the framed message associated with the corresponding picture. Participants were not informed of the testing ahead of time because this information could have affected the way they viewed and processed the ads (Anderson and Pichert, 1978; Rayner, 1998). Participants received a US$5 gift certificate as compensation.

Analyses

The data, numerically describing participants’ number of eye fixations, dwell times, and masked-recall scores to the framed osteoporosis ads, were exported and analyzed in Statistical Package for the Social Sciences (SPSS) version 19. Sphericity was assumed for all analyses. Given that the objective of this study was to assess how message frame moderates cognitive processing of osteoporosis prevention print ads, only data for the physical activity and calcium/vitamin D ads were included in our analyses. The results of comparisons between osteoporosis ads and commercial health and beauty ads are reported elsewhere (O’Malley et al., 2011).

Data treatment

Fixation and dwell time outcomes were standardized for word length per ad. First, separate RM ANOVAs were conducted to examine whether fixation number, dwell time, and recall for the osteoporosis ads were affected by message content (physical activity, calcium, and vitamin D) × frame (gain, loss, and neutral). Next, separate multivariate analysis of variance (MANOVA) for fixation number, dwell time, and recall for each ad type (i.e. the gain, loss, and neutral versions of the ads) was run to examine the impact of group assignment (A, B, and C). Significant multivariate effects were followed up with the Bonferroni adjusted ANOVA, p < .005.

Hypothesis testing

Three separate RM ANOVA on eye fixations, dwell time, or masked-recall data were performed. Message frame (gain, loss, and neutral) was entered as within-participant factors. The Bonferroni post hoc analysis testing was carried out to determine where these significant differences occurred between message frames. In addition, correlations were run to determine whether there was a significant relationship between masked-recall results and number of fixations and dwell times.

Results

Data treatment

The content (physical activity, calcium, and vitamin D) × frame RM ANOVAs on fixation number, dwell time, and recall did not reveal any significant interactions (see Table 1). Thus, for all subsequent analyses, attention to and recall of physical activity and calcium and vitamin D ads were considered collectively as osteoporosis ads. In our test to determine whether attention to the ads varied across assigned groups (A, B, and C), the MANOVAs were significant, post hoc tests failed to reveal differences between conditions, p > .05. Thus, group assignment was not entered as a covariate in the subsequent analyses.

Table 1.

Results of data treatment analysis—mean number of fixations, mean dwell time, and masked-recall score for physical activity and calcium/vitamin D osteoporosis prevention ads.

	Fixations (No./word)		Dwell time (ms/word)		Recall
	M (SD)	d	M (SD)	d	M (SD)	d
Content	F(1,59) = 99.14, p < .001, h² = .63		F(1,59) = 57.28, p < .001, h² = .49		F(1,59) = 10.18, p < .002, h² = .17
PA	11.14 (4.02)	.70**	2583.51 (1098.51)	.52**	2.07 (.52)	.39**
Cal/Vit	8.68 (2.97)		2094.58 (762.24)		1.89 (.38)
Content × frame	F(2,118) = .11, p = .90, h² = .002		F(2,118) = .41, p = .67, h² = .007		F(2,118) = 2.43, p = .09, h² = .04
PA × frame
Gain	9.22 (3.46)	G vs L: .25	2208.31 (855.09)	G vs L: .17	2.11 (.64)	G vs L: .63
Loss	8.44 (2.90)	L vs N: .02	2065.35 (811.87)	L vs N: .07	1.76 (.46)	L vs N: .12
Neutral	8.37 (3.24)	G vs N: .25	2010.09 (818.45)	G vs N: .24	1.82 (.51)	G vs N: .50
Cal/Vit × frame
Gain	11.65 (4.63)	G vs L: .14	2762.67 (1430.12)	G vs L: .18	2.45 (.85)	G vs L: .72
Loss	11.03 (4.22)	L vs N: .07	2536.36 (1007.65)	L vs N: .08	1.92 (.59)	L vs N: .12
Neutral	10.75 (4.30)	G vs N: .20	2451.49 (1080.48)	G vs N: .23	1.85 (.71)	G vs N: .77

SD: standard deviation; PA: physical activity; Cal: calcium; Vit: vitamin; G: gain-framed ad; N: neutral-framed ad; L: loss-framed ad.

The main effect for message frame is reported in Table 2 describing the hypothesis tests.

p < .05, **p < .01.

Number of fixation

The RM ANOVA on number of fixations revealed that message frame significantly affected viewer attention to osteoporosis ads, F(2,118) = 8.18, p < .01, η² = .12. Gain-framed ads had a higher number of fixations than loss-framed ads, d = .20, p < .01, or neutral-framed osteoporosis ads, d = .24, p < .01. There, however, was no significant difference in the number of fixations loss-framed messages received in comparison to neutral-framed osteoporosis messages, d = .05, p > .05 (refer to Table 2).

Table 2.

Results of hypothesis testing—mean number of fixations, mean dwell time, and masked recall by message frame.

	F	Gain	Loss	Neutral	Effect Size (Cohen’s d)			Summary of results
					G versus L	G versus N	L versus N
Number of Fixations/word (M ± SD)	F(2,118) = 8.18, p < .01, h² = .12	10.44 (3.71)	9.73 (3.38)	9.56 (3.56)	.20**	.24**	.05	Participants had a higher numbers of fixations for, spent more time viewing, better recalled gain-framed ads compared to loss- and neutral-framed ads. There were no significant differences in number of fixations, dwell time or recall between loss-framed and neutral-framed ads
Dwell time/word (M ± SD ms)	F(2,118) = 9.84, p < .01, h² = .14	2485.49 (1071.14)	2300.86 (860.17)	2230.79 (904.26)	.19**	.26**	.08
Masked recall (M ± SD) out of 4	F(2,118) = 22.28, p < .01, h² = .27	2.28 (.61)	1.84 (.40)	1.84 (.50)	.85*	.79*	.001
Fixation—recall correlation (r)		.29*	−.01	.30*				Increased recall was associated with increased number of fixations and dwell time for gain-framed and neutral-framed ads but not loss-framed ads
Dwell—recall correlation (r)		.42**	−.02	.27*

SD: standard deviation; G: gain-framed ad; N: neutral-framed ad; L: loss-framed ad.

Note: *p < .05, **p < .01.

Dwell time

The RM ANOVA on dwell time revealed a main effect for message frame, F(2,118) = 9.84, p < .01, η² = .14. Gain-framed ads were dwelled on longer than loss-framed or neutral-framed ads, d > .19, p < .01. There was, however, no significant difference in the amount of time participants dwelled on loss-framed ads in comparison to neutral-framed ads, d = .08, p > .05.

Masked recall

The RM ANOVA on masked recall revealed a main effect for message frame, F(2,118) = 22.28, p < .01, η² = .27. The gain-framed ads were recalled more accurately than loss-framed or neutral-framed osteoporosis ads, d > .79, p < .01. There was, however, no significant difference in the ability to correctly recall loss-framed versus neutral-framed ads, d = .001, p > .05.

Correlations

Finally, correlations were run to determine whether there was a significant relationship between masked-recall results to both number of fixations and dwell times. Results revealed that the ability to correctly recall gain-framed message content was significantly positively correlated with viewers’ number of eye fixations, r = .29, p < .02 and dwell time, r = .42, p < .01 The longer viewers attended to the ads, the better they were able to recall the information. A similar finding appeared for neutral-framed message content, showing that as number of fixations, r = .30, p < .02, and dwell time, r = .27, p < .04, increased, masked-recall improved. However, loss-framed messages did not follow this same pattern. There was no significant correlation between viewers’ ability to accurately recall loss-framed information and their fixation numbers or dwell time, ps > .05.

Discussion

As indicated by fixation and dwell time results, gain-framed osteoporosis prevention ads received significantly more attention than either loss-framed or neutral-framed ads. Gain-framed ads were also better recalled. There were no significant differences in attention or recall of loss-framed and neutral-framed ads. Moreover, as hypothesized, greater attention correlated with improved ability to accurately recall gain-framed and neutral-framed information. From a practical perspective, our data suggest that a gain-framed ad should be used when aiming to create attention-grabbing, memorable osteoporosis prevention messages and advertisements.

Accordingly, message frame appears to have influenced the extent to which osteoporosis prevention messages were attended to and cognitively processed. It may be that because gain-framed content was congruent with the risk-averse behavior of osteoporosis prevention, message congruency elicited greater attention, leading to more cognitive processing than incongruent loss-framed and neutral-framed messages. Because gain-framed osteoporosis prevention messages were more highly attended to, and better recalled than other framed messages, these messages may be able to more effectively motivate behavior change. Due to the within-participant design of the current study and the lack of a behavioral outcome, this possibility could not be investigated in the current study. However, this postulate seems plausible given that previous research has found a gain-framed advantage for motivating increased calcium intake (Jung et al., 2011).

The neutral-framed messages were likely less effective at capturing attention than the gain-framed messages because in the neutral-framed messages, the stated benefits of taking preventative action likely did not outweigh the suggested consequences of failing to take action, or vice versa. Viewers were not likely prompted to attend to neutral-framed messages because there were no obvious benefits associated with doing so; again, the benefits did not outweigh the consequences as such, the message frame was not congruent with participants’ behavioral construal.

Loss-framed messages were likely no more effective at capturing attention than neutral-framed messages because the loss-framed messages were also incongruent with the participants’ perception of the health behavior; there was a mismatch between behavioral construal and the information they received. This incongruency potentially impeded information processing. Furthermore, because loss-framed messages likely generated unfavorable feelings of vulnerability, viewers may have been prompted to defensively deny or reject the message content (Jepson and Chaiken, 1990; Petty and Cacioppo, 1981; Shiv et al., 1997; Siero et al., 1984). To avoid these negative feelings, participants may have spent little time attending to the message.

Given that participants did not thoroughly attend to the loss-framed ads, it was not surprising that they were unable to accurately recall message content. It is interesting, however, that attention to the loss-framed ads, although limited, did not correlate with message recall. This finding is contrary to the findings for the gain-framed and neutral-framed ads, where a positive association was observed. The lack of association may indicate that when participants did attend to the loss-framed ads, they were not attending to the ad text—the components of the ad that they had to recall. It may be that message frame not only affects how long people dwell and fixate upon a message but it may also influence how they view a message. Participants may allocate their attention to image versus text content of ads differently depending on the ad content. In the future, researchers might consider assessing recall of both the text and the image elements of ads. It may also be worthwhile to examine viewing patterns of framed messages.

Indeed, the current study points to several areas for future research. Additional research is needed, to establish cognitive processing as a mediator in the relationship between framed health messages and behavioral outcomes. This research must use a between-participant pre/posttest study design to demonstrate that increased attention directly results in behavior change. Using a between-participant design will also necessitate consideration of individual differences that may affect viewing patterns (e.g. reading speed and issue relevance). Several studies have identified factors that moderate the impact of framed messages (e.g. Gainforth and Latimer, 2011; Hevey et al., 2010; Hwang et al., 2011). Whether these factors moderate attention to ads is yet to be determined. Viewing ads on an eye tracking system may alter natural viewing tendencies; research conducted in a laboratory setting using objective measures of attention (i.e. eye tracking) should be replicated in a naturalistic setting using alternate measures of cognitive processing (e.g. recall). Moreover, these studies should make use of larger demographic data, so that study findings can be generalized to the wider population, not just university-aged women. Ideally, such studies also should be conducted within other countries and cultures, as it would be valuable to determine if differing populations attend to and recall framed messages differently.

Finally, alternate methods to assess recall should be used. The masked-recall task used in the current study uses contextually driven processing, which is a form of top-down processing where individuals’ higher level knowledge and expectations influence their processing and thus can affect their recall (Blaxton, 1989). Thus, if participants were more familiar with gain-framed versus loss-framed osteoporosis messages, they may have been apt to have better recall of the gain-framed message simply because of this increased familiarity. A prompted recall task should be considered to minimize bias.

As one of the first studies to apply eye tracking technology to examine attention to framed health messages, the current study makes an important contribution in laying the preliminary groundwork for examining cognitive processing as a mechanism underlying framing effects. Our study offers direction for future research in the area and provides some indication that gain-framed messages may be useful for capturing women’s attention to osteoporosis prevention ads.

Footnotes

Funding

This research was funded by the Social Sciences and Humanities Research Council of Canada.

References

Anderson

Pichert

(1978) Recall of previously unrecallable information following a shift in perspective. Journal of Verbal Learning and Verbal Behavior 17(1): 1–12.

Andreasen

(1984) Life status changes and changes in consumer preferences and satisfaction. Journal of Consumer Research 11(3): 784–794.

Arora

(2004) The impact of message framing and credibility: Findings for nutritional guidelines. Services Marketing Quarterly 26(1): 35–53.

Basset-Gunter

Latimer-Cheung

Martin Ginis

. (in press) I spy with my little eye—A persuasive message? Journal of Health Communication.

Blaxton

(1989) Investigating dissociations among memory measures: Support for a transfer appropriate hypothesis. Journal of Experimental Psychology-Learning Memory and Cognition 15(4): 657–668.

Christianson

Loftus

Hoffman

. (1991) Eye fixations and memory for emotional events. Journal of Experimental Psychology-Learning Memory and Cognition 17(4): 693–701.

Feskanich

Willett

Colditz

(2003) Calcium, vitamin D, milk consumption, and hip fractures: A prospective study among postmenopausal women. American Journal of Clinical Nutrition 77(2): 504–511.

Fox

Krugman

Fletcher

. (1998) Adolescents’ attention to beer and cigarette print ads and associated product warnings (Special Issue on Advertising to Children). Journal of Advertising 3(27): 57–68.

Gainforth

Latimer

(2011) Risky business: Risk information and the moderating effect of message frame and past behaviour on women’s perceptions of the human papillomavirus vaccine. Journal of Health Psychology. 17(6) 896–906.

10.

Gallagher

Updegraff

(2011) Health message framing effects on attitudes, intentions, and behavior: A meta-analytic review. Annals of Behavioral Medicine. Available at: http://www.ncbi.nlm.nih.gov/pubmed/21993844

11.

Gerend

Erchull

Aiken

. (2006) Reasons and risk: Factors underlying women’s perceptions of susceptibility to osteoporosis. Maturitas 55(3): 227–237.

12.

Hevey

Pertl

Thomas

. (2010) Body consciousness moderates the effect of message framing on intentions to use sunscreen. Journal of Health Psychology 15: 553–559.

13.

Hwang

Cho

Sands

. (2011) Effects of gain- and loss-framed messages on the sun safety behavior of adolescents: The moderating role of risk perceptions. Journal of Health Psychology.

14.

Jepson

Chaiken

(1990) Chronic issue-specific fear inhibits systematic processing of persuasive communications. Journal of Social Behavior and Personality 5(2): 61–84.

15.

Jung

Ginis

KAM

Phillips

. (2011) Increasing calcium intake in young women through gain-framed, targeted messages. Psychology & Health 26(5): 531–547.

16.

Kasper

Peterson

Allegrante

(2001) The need for comprehensive educational osteoporosis prevention programs for young women: Results from a second osteoporosis prevention survey. Arthritis and Rheumatism 45(1): 28–34.

17.

Krugman

Fox

Fletcher

. (1994) Do adolescents attend to warnings in cigarette advertising? An eye-tracking approach. Journal of Advertising Research 34(6): 39–52.

18.

National Osteoporosis Foundation (2011) Fast facts on osteoporosis. Available at: http://www.nof.org/startaconversation/getthefacts

19.

O’Keefe

Jensen

(2008) Do loss-framed persuasive messages engender greater message processing than do gain-framed messages? A meta-analytic review. Communication Studies 59(1): 51–67.

20.

O’Malley

Berenbaum

Latimer

(2011) Using eye tracking technology to determine the most effective format for osteoporosis prevention print advertisements. Journal of Applied Biobehavioral Research 16: 167–186.

21.

Osteoporosis Canada (2011) What is osteoporosis? Available at: http://www.osteoporosis.ca/index.php/ci_id/5526/la_id/1.htm

22.

Petty

Cacioppo

(1981) Attitudes and Persuasion: Classic and Contemporary Approaches. Dubuque, IA: Wm. C. Brown.

23.

Pieters

Wedel

(2007) Goal control of attention to advertising: The Yarbus implication. Journal of Consumer Research 34(2): 224–233.

24.

Rayner

(1998) Eye movements in reading and information processing: 20 years of research. Psychological Bulletin 124(3): 372–422.

25.

Rayner

Rotello

Stewart

. (2001) Integrating text and pictorial information: Eye movements when looking at print advertisements. Journal of Experimental Psychology-Applied 7(3): 219–226.

26.

Recker

Davies

Hinders

. (1992) Bone gain in young adult women. JAMA-Journal of the American Medical Association 268(17): 2403–2408.

27.

Rosbergen

Pieters

Wedel

(1997) Visual attention to advertising: A segment-level analysis. Journal of Consumer Research 24(3): 305–314.

28.

Rothman

Salovey

(1997) Shaping perceptions to motivate healthy behavior: The role of message framing. Psychological Bulletin 121(1): 3–19.

29.

Rothman

Updegraff

(2011) Specifying when and how gain- and loss-framed messages motivate health behavior: An integrated approach. In: Keren

(ed.) Perspectives on Framing. New York: Psychology Press, pp. 257–278.

30.

Rothman

Wlaschin

Bartels

. (2008) How persons and situations regulate message framing effects: The study of health behavior. In: Elliott

(ed.) The Handbook of Approach and Avoidance Motivation. New York: Psychology Press, pp. 475–486.

31.

Shiv

Edell

Payne

(1997) Factors affecting the impact of negatively and positively framed ad messages. Journal of Consumer Research 24(3): 285–294.

32.

Siero

Kok

Pruyn

(1984) Effects of public education about breast cancer and breast self-examination. Social Science & Medicine 18(10): 881–888.

33.

Skumanich

Kintsfather

(1996) Promoting the organ donor card: A causal model of persuasion effects. Social Science & Medicine 43(3): 401–408.

34.

Tversky

Kahneman

(1981) The framing of decisions and the psychology of choice. Science 211(4481): 453–458.

35.

Wedel

Pieters

(2000) Eye fixations on advertisements and memory for brands: A model and findings. Marketing Science 19(4): 297–312.