Master’s Thesis
Monique Alblas (10419721)
Research Master Communication Science Graduate School of Communication Supervised by dr. A. J. Linn
19 January 2015
TAILORING THE MODE OF DELIVERY
:
A MATTER OF PREFERENCE
?
A study into the determinants of the delivery mode preference for receiving health instructions and the effects of tailoring to delivery mode preference
2
Abstract
Background: Although theories such as the multimedia theory of cognitive learning suggest that combining written text with illustrations or spoken text with visuals will expand cognitive capacity, research has not yet reached agreement on which mode of delivery is most effective for health instructions in terms of recall, self-efficacy, and intention. As individuals differ in their preferences for a particular mode of delivery, the effectiveness of health instructions might depend on one’s delivery mode preference. The aims of the current study are to examine (1) the determinants of delivery mode preference, and (2) whether tailoring to delivery mode preference is more effective than non-tailoring in terms of recall, self-efficacy, and intention.
Methods: A tailored versus non-tailored factorial design was used. Health instructions (breast self-examination instructions (BSE) in Study 1, skin self-examination instructions (SSE) in Study 2) were either tailored to the participant’s delivery mode preference (text, text with illustrations, or video) or non-tailored by randomly providing one of these three modes of delivery.
Results: In Study 1, 205 women completed the online questionnaire. Need for cognition (NFC) was a determinant of delivery mode preference: participants with a low NFC preferred the video mode more often than participants with a high NFC. In addition, the intention to perform BSE was higher when the instructions were tailored to delivery mode preference than when not tailored. There were no effects of tailoring on recall and self-efficacy to perform BSE. In Study 2, 200 men and women participated. Introvert participants, participants that preferred to watch television, and men preferred the video mode more often than extravert participants, participants that did not prefer television, and women. There were no effects of tailoring on recall, self-efficacy, or the intention to perform SSE.
3
could be taken into account when choosing the mode of delivery for health interventions. Furthermore, tailoring to delivery mode preference is effective in improving the intention to perform BSE. Although more research is needed, tailoring to delivery mode preference represents a potentially fruitful gateway for advancing tailored health communication.
4
Introduction
Health instructions can be provided in various modes of delivery, such as in written text, written text with supporting illustrations, or video. Based on the cognitive theory of multimedia learning (CTML), previous research has tried to explain why some modes of delivery are more effective than others (Mayer, 2002). The CTML states that people understand instructional material better when presented in multiple forms, including verbal and visual representations, as compared to singular forms. This is because processing
information that is presented in multiple forms is cognitively less demanding than processing information in singular forms, which may result in deeper information processing. Previous research on the effectiveness of presenting health instructions in various modes of delivery has, however, not reached agreement on which mode of delivery is most effective in terms of recall of the information, self-efficacy to perform the behaviour advocated in the instructions, and the intention to perform this behaviour (e.g., Dowse & Ehlers, 2005; Labranche, Helweg-Larsen, Byrd, & Choquette, 1997; Morrell, Park, & Poon, 1990). As individuals differ in their preference for receiving health instructions (Van Hooijdonk, 2008), it is possible that the inconsistent results from previous research can be explained by individual delivery mode preference. However, if the effectiveness of different modes of delivery depends on the individual’s delivery mode preference, it is important to know what determines one’s preference. Based on the individual’s characteristics it can then be predicted which mode of delivery is most effective for him or her.
Research suggests that delivery mode preference can potentially be determined by personal characteristics (Van Hooijdonk, 2008; Wright et al., 2008). For example, people with a high need for cognition (NFC) have an intrinsic motivation to put effort into deep
information processing (Cacioppo & Petty, 1982). As processing written information requires more cognitive effort than processing audio-visual information (Mayer, 2002), people with a
5
high NFC may prefer written information more as compared to audio-visual information. In addition to personal characteristics, media preferences might determine the individual’s delivery mode preference as a medium is often related to a certain mode of delivery (television generally consists of audio-visual material, while newspapers and magazines combine written text with illustrations) (Heo & Cho, 2009).
A number of researchers have reported that individual differences and delivery mode preferences should be taken into account when examining the effects of different modes of delivery (Mayer, 2001; Van Hooijdonk, 2008). Specifically, a message that is tailored to the needs and preferences of the recipient is considered as more personally relevant by the recipient (Jensen, King, Carciopplo, & Davis, 2012) and will therefore be processed more deeply (Kreuter & Wray, 2003). This deeper message processing will lead to better recall of the information (Bandura, 1977; Klein, Noe, & Wang, 2006), a higher self-efficacy to perform the advocated behaviour (Noar, Pierce, Larson, & Black, 2010), and a higher intention to perform this behaviour (Brug, Glanz, Van Assema, Kok, & Van Breukelen, 1998). Tailoring health instructions to the individual’s delivery mode preference might thus improve the effectiveness of health instructions.
Until now, however, surprisingly little research has been conducted to investigate the effectiveness of interventions tailored to individuals’ delivery mode preferences. Research on the determinants of delivery mode preference is scarce as well. In addition, the results can be used to improve the effectiveness of existing tailored health interventions. The aims of this research are to examine (1) the determinants of delivery mode preference and (2) whether tailoring to delivery mode preference is more effective than non-tailoring in terms of recall of the information, self-efficacy, and the intention to perform the health behaviour. This will be examined in two studies: the first concerns instructions for breast self-examination (BSE), whereas the second focuses on skin self-examination (SSE) instructions.
6
Theoretical Background
The Effects of Health Instructions Presented in Various Modes of Delivery
Health instructions can be presented in various modes of delivery such as in a written text with or without illustrations or in a video. Combining visual and verbal information together in one single instruction environment is called multimedia learning (Soto Mas, Plass, Kane, & Papenfuss, 2003). Multimedia learning does thus not refer to the presentation of instructional messages in different media (such as newspapers and television), but to different modes of delivery (such as written information and illustrations). The CTML proposes that instructional messages are better understood when they are presented in multiple forms as compared to singular forms. Specifically, the multimedia principle of the CTML states that written information that is supported by illustrations is better processed as compared to written information alone (Mayer, 2002). Additionally, the modality principle of the CTML states that instructional messages are better processed when they are presented audio-visually as compared to written information. These assumptions are derived from the dual coding approach, which assumes that two separate information processing systems are stimulated when people are exposed to a multimedia message: a visual system for processing visual knowledge and a verbal system for processing verbal knowledge (Mayer, 1997; Paivio, 1971). When instructions consist of both visual and verbal information, processing this information is cognitively less demanding because both systems are used. This may result in deeper processing of the information.
Several studies empirically tested the multimedia principle and the modality principle of the CTML. These studies examined the effectiveness of various modes of delivery on different outcomes such as recall of information, self-efficacy to perform the behaviour that is advocated in the health instructions, and the intention to perform this behaviour (e.g., Dowse & Ehlers, 2005; Labranche et al., 1997; Morrell et al., 1990). However, results from these
7
studies are inconsistent. A number of studies found support for the multimedia principle (Delp & Jones, 1996; Dowse & Ehlers, 2005; Sojourner & Wogalter, 1998). These studies
demonstrated that written health instructions provided with supporting illustrations were more effective than written instructions alone for improving recall of the information (Delp & Jones, 1996; Sojourner & Wogalter, 1998) and adherence to the advocated behaviour (Dowse & Ehlers, 2005). On the contrary, research also showed that written information alone led to better recall than written information supported by illustrations (Morrell et al., 1990). In addition, some studies found no difference between written information alone and written information combined with illustrations on improving recall (Kools, Van de Wiel, Ruiter, & Kok, 2006), self-efficacy, and intention (Labranche et al., 1997). Results of studies on the modality principle are inconsistent as well. Some studies found support for the modality principle (Linn & Bol, submitted; Van Hooijdonk, 2008) and showed that health instructions presented in a video were better recalled (Linn & Bol, submitted) and that they resulted in better exercise performance (Van Hooijdonk, 2008) as compared to instructions presented in written text. On the contrary, Mayer, Hegarty, Mayer, and Campbell (2005) concluded that written information was more effective than audio-visual information in terms of recall and comprehension. Some other studies, however, did not find differences between audio-visual and written information on recall of information (Carroll & Wiebe, 2004) and on the self-efficacy (Michas & Berry, 2000).
Thus, to date, there has been little agreement on which mode of delivery is more effective than the other modes. As was pointed out in the introduction to this paper, several researchers emphasize the potential influence of individual differences on the effectiveness of these different modes of delivery (Mayer, 2001; Van Hooijdonk, 2008). It is likely that these inconsistent results are a result of different preferences for receiving health instructions in a particular mode of delivery and there is a possibility that the effectiveness of the mode of
8
delivery depends on the individual’s preferences (Van Hooijdonk, 2008). For this reason, it is important to first explore the determinants of delivery mode preference.
Personal Characteristics, Media Preferences, and Delivery Mode Preference Personal Characteristics and Delivery Mode Preference
As discussed above, the effectiveness of the mode of delivery might depend on the individual’s delivery mode preference. There are some factors that can possibly determine delivery mode preference, namely several personal characteristics (Van Hooijdonk, 2008; Wright et al., 2008) and media preferences (Gaglio, Glasgow, & Bull, 2012; Heo & Cho, 2009). A personal characteristic that might predict delivery mode preference is need for cognition (NFC), which is the extent to which someone enjoys to participate in effortful cognitive activities (Cacioppo & Petty, 1982). Individuals with a high NFC have an intrinsic motivation to put effort into deep information processing, while individuals with a low NFC have less or no motivation to process information deeply. Because processing written information requires more cognitive effort as compared to audio-visual information (Mayer, 2002), it is likely that individuals with a high NFC prefer to receive written health
instructions. Individuals with a low NFC do not (or less) enjoy cognitive effort and are
therefore expected to prefer audio-visual health instructions. These assumptions are supported by research that showed that individuals with a high NFC preferred to read complex
information more than individuals with a low NFC (Stokmans, 2000). Furthermore, Henning and Vorderer (2001) demonstrated that individuals with a low NFC preferred television more than individuals with a high NFC. It is yet unknown whether individuals with different levels of NFC also prefer different modes of delivery. However, based on previous research, it can be hypothesized that individuals with a high NFC prefer written health instructions more as compared to individuals with a low NFC (H1a). In addition, it is hypothesized that individuals with a low NFC prefer audio-visual health instructions more as compared to individuals with
9
a high NFC (H1b).
In addition to NFC, personality traits might predict someone’s delivery mode preference as well. The HEXACO model of personality structure includes the main six personality traits: honesty-humility, emotionality, extraversion, agreeableness,
conscientiousness, and openness to experience (Ashton et al., 2004). Each trait consists of a dimension on which individuals can score high or low. Especially extraversion might determine delivery mode preference as extraversion is related to sensory preferences (Digman, 1990; Eysenck, 1967). Extravert individuals are outwardly focused, enjoy social activities, and require much external sensory stimulation and excitement (Digman, 1990). Introvert individuals, on the other hand, have little need for external stimulation (Eysenck, 1967). This preference for external stimulation can be related to delivery mode preference. Modes of delivery differ in pacing, which refers to the perceived control over the moment and speed of receiving information (Bronner & Neijens, 2006). In modes of delivery with external pacing, such as video, the moment and speed of receiving the information cannot be
controlled by the recipient. This causes much sensory stimulation and excitement (Bolls, Mueling, & Yoon, 2003) which is preferred by extravert individuals. Research indeed showed that extravert individuals were more positive about commercials when they were presented at high speed, while introvert individuals preferred commercials that were presented at low speed (Van Rootselaar, 2008). It can therefore be expected that introvert individuals prefer written health instructions more than extravert individuals because when processing written information the recipient has more control over the external sensory stimulation than when processing audio-visual health instructions (H2a). On the contrary, extravert individuals are expected to prefer audio-visual health instructions such as video more than introvert
individuals (H2b).
10
is health literacy. Health literacy refers to the cognitive and social skills which determine the motivation and ability of individuals to gain access to, understand and use health information (Nutbeam, 1998). Low-literate individuals are less able to process written information than high-literate individuals (Davis, Williams, Marin, Parker, & Glass, 2002). A possible explanation for this is that low-literate individuals are generally auditively oriented while high-literate individuals are visually oriented (Mohnen, 2002). This assumption is supported in various empirical studies (e.g., Gaglio et al., 2012; Houts, Doak, Doak, & Loscalzo, 2006; Murphy, Chesson, Walker, Arnold, & Chesson, 2000; Soroka et al., 2006). Houts et al. (2006) and Soroka et al. (2006) showed that low-literate individuals preferred spoken health
information over written health information. On the contrary, high-literate individuals preferred written health information (Gaglio et al., 2012). Moreover, Murphy et al. (2000) showed that low-literate individuals were less motivated than high-literate individuals to read written health information. However, when the same information was presented
audio-visually there was no difference between low-literates and high-literates in their motivation to process the information. Based on these results, it can be expected that high-literate
individuals prefer written health instructions more than low-literate individuals (H3a). On the contrary, it is expected that low-literate individuals prefer audio-visual health instructions more than high-literate individuals (H3b).
Media Preferences and Delivery Mode Preference
In addition to personal characteristics, media preferences might determine individuals’ delivery mode preference. Some individuals prefer to watch television, others enjoy reading books or magazines, or listen to the radio (Heo & Cho, 2009). There seems to be a relation between media preferences and delivery mode preference, as individuals often prefer various media that are presented in the same mode of delivery. For instance, individuals who read newspapers also read magazines and articles on the internet (Heo & Cho, 2009). These
11
individuals are called the print-oriented group. With regard to health information, individuals who prefer to receive health information via pamphlets also prefer to read hand-outs or written information on the internet (Gaglio et al., 2012). Another category of media users is the video-audio-combined group (Heo & Cho, 2009). Individuals within this group enjoy watching television and playing television games. With regard to health information, individuals who prefer face-to-face health communication generally also prefer to receive health information via the telephone (Gaglio et al., 2012). Based on these results, it is
hypothesized that individuals who prefer printed media prefer written health instructions more than individuals who do not prefer printed media (H4a). In addition, it is hypothesized that individuals who prefer the medium television prefer audio-visual health instructions more than individuals who do not prefer television (H4b).
The Effects of Tailoring to Delivery Mode Preference
As individual differences exist in the delivery mode preference for receiving health instructions (Van Hooijdonk, 2008), it is reasonable to assume that health instructions are not equally effective for every individual. These differences suggest that individual mode
preference should be taken into account by providing instructions in one’s preferred mode of delivery. Adjusting a message to the needs of the individual, such as the individual’s
background characteristics or preferences, is called tailoring (Kreuter, Farrell, & Olevitch, 2000). Research suggests that tailoring is more effective than non-tailoring because an adjusted message is considered as more personally relevant to the recipient of the message (Jensen et al., 2012). This increased personal relevance might lead to deeper processing of the message (Kreuter & Wray, 2003), which in turn might result in better recall of information (Bandura, 1977; Klein et al., 2006), a higher self-efficacy to perform the advocated behaviour (Noar et al., 2010), and a higher intention to perform this behaviour (Brug et al., 1998; Brug, Steenhuis, Van Assema, & De Vries, 1996). These outcomes (recall, self-efficacy, and
12
intention) are important predictors of many health behaviours (Ajzen, 1991; Bandura, 1986; Linn, Van Dijk, Smit, Jansen, & Van Weert, 2013).
To our knowledge, only two studies examined health messages tailored specifically to delivery mode preference (Lewis, Napolitano, Whiteley, & Marcus, 2006; Vandelanotte, Duncan, Plotnikoff, & Mummery, 2012). These studies both investigated whether a mode tailored intervention to improve physical activity was more effective in terms of performing the advocated behaviour (physical activity) as compared to a non-tailored intervention. Both studies did not find differences between participants that received their preferred mode of delivery and participants that received a random mode of delivery. A possible explanation for this is that individuals in these studies did not have a strong preference for one of the modes of delivery. This is supported in the study by Vandelanotte et al. (2012) in which the mode of delivery that participants chose beforehand was not always their preferred mode of delivery afterwards. This might be because it was unclear to them beforehand what the intervention with different kinds of motivational information and feedback would look like in the different modes of delivery. In the current research, individuals only receive one short set of
instructions, so it might be more easy for them to imagine the health instructions in different modes of delivery. As a result, individuals in this study might have a stronger preference for a specific mode of delivery which may lead to stronger effects of tailoring as compared to the previous studies. To examine the effects of tailoring to delivery mode preference the
following research question will be explored: To what extent is tailoring to delivery mode preference more effective than non-tailoring in terms of recall of the information (RQ1), self-efficacy to perform the advocated behaviour (RQ2), and the intention to perform this
13
Study 1 Method Design
This first study investigated whether personal characteristics and media preferences determined delivery mode preference. To subsequently examine the effects of tailoring on recall, self-efficacy, and intention a factorial design with two conditions (tailoring vs non-tailoring) was used. Participants (N = 205) were randomly assigned to one of these conditions. Participants in the tailoring condition (N = 106) were asked which of the three modes of delivery (text, text with illustrations, or video) they preferred for receiving the instructions. Subsequently, participants received the instructions in their preferred mode of delivery. The second condition was non-tailoring (N = 99). In this condition, participants randomly received the instructions in the text mode (N = 34), the text with illustrations mode (N = 34), or the video mode (N = 31).
Stimulus Material
Initially, three different versions of breast self-examination (BSE) instructions were developed: written instructions without illustrations (the text mode), written instructions with supporting illustrations (the text with illustrations mode), and audio-visual instructions (the video mode). These instructions were based on the existing video instructions as developed by Pink Ribbon, a Dutch breast cancer foundation (Ziekenhuis Amstelland, 2011). To create the video mode, a part of the original video was used. The text and text with illustration modes were created by transcribing the spoken text in the video mode into written text.
Subsequently, this text was structured into six steps to perform BSE. An illustration (a print screen from the original video instructions) was added to each step in the text with
illustrations mode. The three modes of delivery are thus identical regarding their content and only differed in mode of delivery (see Appendix A).
14
Participants and Procedure
In November and December 2013 participants (N = 205) were recruited via email, social media, and several online forums. All participants gave online informed consent. Participants were informed that they would receive BSE instructions and that they had to answer some questions about the instructions afterwards. After the informed consent,
participants were randomly assigned to one of the two conditions (tailoring or non-tailoring). In the tailoring condition, participants had to indicate their delivery mode preference and subsequently received the instructions in their preferred mode of delivery. In the non-tailoring condition, participants were randomly assigned to one of the three modes of delivery.
Participants could take as long as they wanted to look at the instructions in order to simulate a natural setting. Participants were not able to return to the BSE instructions while answering the questions. The questionnaire included questions about the outcome variables (recall, self-efficacy, and intention), the control variables (frequency of performing BSE and familiarity with the BSE instructions), and the manipulation check. In addition, the participant’s personal characteristics (NFC, extraversion, and health literacy), media preferences, and demographic variables were measured. Completing the questionnaire acquired approximately 15 to 20 minutes. This online experiment was approved by the institutional review board of the research institute.
Measures
NFC. NFC was measured with 7 items on a 7-point Likert scale (α = 0.74, M = 4.80, SD = 0.93) (Pieters, Verplanken, & Modde, 1987). The scale included items such as
‘Thinking is not my idea of fun’ and ‘I genuinely enjoy a task in which one has to come up with new solutions for problems’ (1 = strongly disagree, 7 = strongly agree).
Extraversion. Extraversion was measured with 10 items on a 5-point Likert scale (α = 0.79, M = 3.53, SD = 0.61) (De Vries, Ashton, & Lee, 2009). The scale included items such as
15
‘In social situations, I’m usually the one who makes the first move’ and ‘I prefer jobs that involve active social interaction to those that involve working alone’ (1 = strongly disagree, 5 = strongly agree).
Health literacy. Health literacy was measured with 14 items on a 4-point scale (α = 0.87, M = 2.06, SD = 0.68) (Van der Vaart et al., 2012). The scale included items such as ‘In reading instructions or leaflets from hospitals/pharmacies, how often do you need a long time to read and understand them?’ (1 = never, 4 = always), and ‘When you obtain health
information yourself, how difficult do you find it to extract the information that you want?’ (1 = easy, 4 = difficult).
Media preferences. Media preferences were measured with one question: ‘What is your favourite medium?’ Participants could choose between internet, television, newspapers, magazines, radio, and else (by choosing this last option, participants could write down another medium). Participants were allowed to choose one or more options. Newspapers and
magazines were merged into ‘printed media’. Most participants preferred the internet (N = 191) and/or television (N = 143). Printed media (N = 92) and radio (N = 53) were less preferred.
Delivery mode preference. Participants’ delivery mode preference was measured with the following question: ‘If you could choose how you would receive BSE instructions, in which way would you prefer these instructions?’ Participants preferred the text with
illustrations mode (N = 59) or the video mode (N = 47). None of the participants chose the text mode.
Recall. Recall of the information was measured with an adapted version of the Netherlands Patient Information Recall Questionnaire (Jansen et al., 2008). In total, 5 open questions were generated from the information in the BSE instructions, such as ‘Could you briefly describe what BSE is?’, and ‘Could you briefly describe how you could examine your
16
armpit?’. All questions were provided with the options ‘This is not discussed’, ‘This is discussed, but I can’t remember’, or ‘This is discussed, namely...’. When participants chose the third option, they could write down the answer. A before-handmade codebook (which can be found in Appendix C) was used for allocating scores to each of the recall questions. Scores on each question could range from 0 (not recalled), 1 (recalled partially) or 2 (recalled correctly). The scores on the five questions were computed into a summed recall score with possible scores ranging from 0 to 10. Additionally, the summed recall scores were calculated into a percentage of the highest possible recall score (M = 28.73, SD = 20.11).
Self-efficacy. Self-efficacy was measured with 6 items on a 7-point Likert scale (α = 0.80, M = 6.00, SD = 0.87) (Norman & Hoyle, 2004). The scale included items such as ‘Whether or not I perform BSE in the next month is entirely up to me’ and ‘How confident are you that you will be able to perform BSE in the next month?’ (1 = completely disagree, 7 = completely agree).
Intention. Intention was measured with 4 items on a 7-point Likert scale (α = 0.97, M = 4.48, SD = 1.92) (Ajzen, 2006). The scale included items such as ‘I intend to perform BSE at least once a month in the next three months’ and ‘I intend to examine my nipple and its surrounding region at least once a month in the next three months’ (1 = very unlikely, 7 = very likely).
Control variables. Two control variables were included: frequency of performing BSE and familiarity with the BSE instructions. Frequency of performing BSE was measured with the following question: ‘To what extent do you perform BSE?’ (1 = never, 7 = always) (M = 2.69, SD = 1.54). Familiarity with the instructions was measured with the following question: ‘Have you seen BSE instructions before?’ (yes / no / don’t know). Most participants had never seen BSE instructions (N = 112), a number of participants had (N = 88), and some participants did not know (N = 5).
17
Manipulation check. To check whether the manipulation of tailoring to preference had been successful, the following statement was posed: ‘The way in which I have received the instructions in this study (via text, text with illustrations, or video) corresponds to my preferred way of receiving BSE instructions’ (1 = not at all, 7 = very much).
Demographic variables. Demographic measures included participants’ age, gender, educational level, work (yes/no), and study (yes/no). Educational level was divided into low level of education (no education, primary education, lower vocational education, preparatory secondary vocational education, and intermediate secondary vocational education), middle level of education (senior secondary vocational education and university preparatory vocational education), and high level of education (higher vocational education and university).
Statistical Analysis
First, a manipulation check was conducted to check whether the manipulation of tailoring had been successful. Next, independent t-tests and chi-square tests were performed to check for unequal distribution of the variables over the two conditions. To examine which personal characteristics and media preferences were related to delivery mode preference, a binary logistic regression analysis was conducted with the personal characteristics and media preferences as independent variables and delivery mode preference as dependent variable. To examine to what extent tailoring to delivery mode preference was more effective than non-tailoring, one-way analyses of variance (ANOVAs) were conducted with tailoring as independent variable and recall, self-efficacy, and intention as dependent variables.
Results Sample Characteristics
Two-hundred and five female participants filled out the online questionnaire (N = 205). The age of these participants ranged from 19 to 72 (M = 34.36, SD = 14.09). Most
18
participants were highly educated (56.1%), 38.5% of the participants were student and 87.3% had work (see Table 1 for an overview).
[Include Table 1 around here]
Manipulation and Randomisation Checks
Manipulation check. A manipulation check was conducted to examine whether the two conditions (tailoring and non-tailoring) differed in the extent to which the received mode of delivery corresponded with the participant’s preferred mode of delivery. In the tailoring condition, the received instructions corresponded more with the participants’ delivery mode preference (M = 5.08, SD = 0.96) than in the non-tailoring condition (M = 4.23, SD = 1.33), t(177.62) = 5.17, p < .001, 95% CI [0.52, 1.17], although it must be taken into account that the assumption of equal variances in the population was violated, Levene’s F(203) = 9.72, p = .002. This violation means that the variances of the scores were greater in the non-tailoring condition than in the tailoring condition. Nevertheless, it can be concluded that the
manipulation of tailoring had been successful.
Randomisation check. To check for successful randomization (i.e., whether conditions differed on age, educational level, the extent to which participants already performed BSE, and the number of participants that were familiar with BSE instructions), independent t-tests and chi-statistics were conducted. The participants’ age was equally divided over the tailoring and non-tailoring conditions, t(203) = -1.01, p = .316, 95% CI [-5.86, 1.90], as was their educational level, χ2(2, N = 205) = 2.59, p = .273, the extent to which participants already performed BSE, t(203) = -0.81, p = .419, 95% CI [-0.60, 0.25], and the number of participants that were familiar with BSE instructions, χ2(2, N = 205) = 3.39, p = .184. As controlling for these variables did not alter the effects, these variables were not included as covariates in further analyses.
19
The Determinants of Delivery Mode Preference
A binary logistic regression analysis was executed to examine whether personal characteristics and media preferences determined delivery mode preference. It was expected that individuals with a high NFC would prefer written instructions and that individuals with a low NFC would prefer audio-visual instructions (H1). None of the participants chose the text mode. With regard to the other modes of delivery, NFC predicted delivery mode preference, Exp(B) = 2.12, p = .022. Participants that preferred the text with illustrations mode had a higher level of NFC as compared to participants that preferred the video mode. H1 is therefore supported. Next, it was expected that introvert individuals would prefer written instructions while extravert individuals would prefer audio-visual instructions (H2). The level or extraversion did, however, not predict delivery mode preference, Exp(B) = 1.33, p = .448. H2 is therefore not supported. Furthermore, it was expected that high-literate individuals would prefer written instructions and that low-literate individuals would prefer audio-visual instructions (H3). However, health literacy did not predict delivery mode preference, Exp(B) = 0.61, p = .333. H3 is therefore not supported. Finally, regarding media preferences it was expected that a preference for printed media would predict a preference for written
instructions and that a preference for television would predict a preference for audio-visual instructions (H4). However, a preference for printed media did not predict delivery mode preference, Exp(B) = 1.77, p = .283, and neither did a preference for television, Exp(B) = 2.74, p = .137. H4 is therefore not supported.
The Effects of Tailoring to Delivery Mode Preference
To address the second aim, it was examined whether tailoring to preference was more effective than non-tailoring in terms of improving recall, self-efficacy, and intention. First, a one-way analysis of variance (ANOVA) was conducted to examine the effect of tailoring to delivery mode preference on recall. Results showed no effect of tailoring on recall of the
20
information, F(1, 203) = 0.14, p = .706, 95% CI [25.94, 31.49]. Participants recalled the information better when the BSE instructions were tailored to their delivery mode preference (M = 29.25, SD = 20.32) as compared to when they received the instructions in a random mode of delivery (M = 28.18, SD = 19.97), but this difference was not significant (see Table 2).
[Include Table 2 around here]
Second, a one-way analysis of variance (ANOVA) was conducted to examine the effect of tailoring to delivery mode preference on self-efficacy. Results showed no effect of tailoring on self-efficacy, F(1, 203) = 0.92, p = .339, 95% CI [5.88, 6.12]. Participants had a higher self-efficacy to perform BSE when the instructions were tailored to their delivery mode preference (M = 6.06, SD = 0.81) as compared to when they received the instructions in a random mode of delivery (M = 5.94, SD = 0.93). However, this difference was not significant (see Table 3).
[Include Table 3 around here]
Finally, a one-way analysis of variance (ANOVA) was conducted to examine the effect of tailoring to delivery mode preference on intention. There was a significant effect of tailoring on the intention to perform BSE, F(1, 203) = 5.69, p = .018, 95% CI [4.20, 4.73]. Participants that received the instructions in their preferred mode of delivery reported a higher intention to perform BSE (M = 4.78, SD = 1.97) as compared to participants that received the instructions in a random mode of delivery (M = 4.15, SD = 1.81) (see Table 4).
[Include Table 4 around here] Post-hoc Analysis
None of the participants chose the text mode. Therefore, differences between the tailoring and non-tailoring condition might be found because the text mode was least effective
21
and only participants in the non-tailoring condition received this mode of delivery. To exclude this possibility, the effects of tailoring were examined again without the text mode. Excluding the text mode from the analyses did, however, not alter the effects of tailoring on recall, self-efficacy, and intention. Recall of the information and the self-efficacy to perform BSE did still not significantly differ between the tailoring and non-tailoring condition and the intention to perform BSE was still significantly higher in the tailoring condition than in the
non-tailoring condition. Thus, the alternative explanation for the differences between the non-tailoring and the non-tailoring condition should be rejected.
Discussion
Study 1 investigated whether personal characteristics (NFC, extraversion, and health literacy) and media preferences predicted delivery mode preference. In addition, it was examined whether it was more effective to provide these instructions in the individual’s preferred mode of delivery as compared to a random mode of delivery.
The results showed that delivery mode preference can be determined by NFC. As expected, participants that preferred written instructions had a higher level of NFC as
compared to participants that preferred audio-visual instructions. This supports the hypothesis that individuals with a low NFC less enjoy to engage in cognitive effortful activities than individuals with a high NFC, and prefer audio-visual instructions because these are cognitively less demanding to process than written instructions (Mayer, 2002). The other personal characteristics and media preferences did not predict delivery mode preference. One possibility is that situational factors predict delivery mode preference better than stable
personal characteristics and preferences. For instance, delivery mode preference could depend on the expected difficulty of the content of the instructions (Wright et al., 2008). When
someone expects that the instructions will be complicated (such as instructions that consist of many detailed steps that have to be performed in a specific order), he or she probably prefers
22
another mode of delivery than when he or she expects that the instructions will be easy. This possibility is supported in studies by Soroka et al. (2006) and Wright et al. (2008) which showed that delivery mode preference differed within different types of instructions.
With regard to the effects of tailoring, Study 1 examined whether tailoring to delivery mode preference would lead to better recall of the information, a higher self-efficacy, and a higher intention to perform BSE. The results showed that tailoring to delivery mode
preference was more effective than non-tailoring in terms of the intention to perform the advocated behaviour. Participants that received the instructions in their preferred mode of delivery showed a higher intention to perform BSE in the next three months as compared to participants that received the instructions in a random mode of delivery. Tailoring to delivery mode preference did, however, not lead to a better recall of the information, and neither to higher self-efficacy to perform BSE. This is interesting, because tailoring is presumed to increase personal relevance (Jensen et al., 2012), which leads to deeper information
processing (Kreuter & Wray, 2003) and subsequently to positive effects on outcome variables such as recall, self-efficacy, and intention. In addition, previous empirical studies on tailoring did not only find an effect on the intention (Brug et al., 1998; Brug et al., 1996) but also on recall (Bandura, 1977; Klein et al., 2006) and self-efficacy (Noar et al., 2010). However, these previous studies did not focus on delivery mode tailoring but on content tailoring (i.e.,
providing relevant information that is tailored to the recipient’s needs). As only an effect on the intention were found in this study, delivery mode tailoring could work in a different way than content tailoring.
Surprisingly, none of the participants preferred the text mode in this study. This is an interesting finding because many written health instructions are not supported by illustrations (Katz, Kripalani, & Weiss, 2006). It is valuable to investigate whether these preferences only hold for BSE instructions or also for other health instructions, and whether these preferences
23
only hold for women, or also for men. In addition, including male participants in future research could be valuable for another reason because previous research concluded that men and women differed in their preferences for learning visually and auditively (Wehrwein, Lujan, and Dicarlo, 2007). Gender might thus predict delivery mode preference.
A limitation of Study 1 is that participants in the tailoring condition were obliged to choose between one of the modes of delivery, while not every participant might have a preference for a specific mode of delivery. For instance, a study by Rogers, Courneya, Verhulst, Markwell, and McAuley (2008) showed that 24% of all participants did not have a specific delivery mode preference for receiving exercise instructions (concerning women who have been cured of breast cancer). The effects of tailoring could possibly be stronger if all participants in the tailoring condition would have a strong preference for one of the modes of delivery.
Study 2
Study 2 was conducted to unravel and expand the results of Study 1. First, other health instructions were used as stimulus material and both male and female participants were invited to participate in order to generalise the results and to investigate whether gender predicted delivery mode preference. Second, several variables were added: personal relevance was investigated as possible mediator of the effects of tailoring, and the expected difficulty of the instructions was examined as possible determinant of delivery mode preference. Third, participants could choose the answer option ‘no preference’ when asked about their preferred mode of delivery in the tailoring condition to find out how many people did not have a specific delivery mode preference.
24
Method
Design
The design of this second study was roughly the same as the design of the first study. Similar to Study 1, a factorial design with two conditions (tailoring vs non tailoring) was used to which participants (N = 200) were randomly assigned. Participants in the tailoring
condition were asked which mode of delivery they preferred (text, text with illustrations, or video) or could choose the answer option ‘no preference’. Participants in the tailoring condition (N = 97) received the instructions in their preferred mode and participants in the non-tailoring condition (N = 103), including the participants with no preference, were
randomly assigned to the text mode (N = 29), the text with illustrations mode (N = 37), or the video mode (N = 37).
Stimulus Material
The stimulus material consisted of skin self-examination (SSE) instructions in three different versions: the text mode, the text with illustrations mode, and the video mode (see Appendix A). These instructions were based on existing video instructions as developed by KWF Kankerbestrijding, a Dutch cancer foundation (KWF Kankerbestrijding, 2014). Similar to Study 1, a part of the original video was used to create the video mode and the spoken text from the video was transcribed to create the text and the text with illustrations modes (for a detailed description, see Study 1).
Participants and Procedure
Between September and November 2014 participants (N = 200) were recruited via email, social media, and several online forums. The procedure of the online experiment was the same as in Study 1 (for a detailed description, see Study 1), but some extra variables were added to investigate additional questions and possible explanations. After participants read the
25
introduction and agreed with the informed consent, they indicated the expected difficulty of the SSE instructions. Subsequently, participants in the tailoring condition chose their
preferred delivery mode and received the instructions in this mode. Participants that chose the answer option ‘no preference’ were randomly assigned to one of the three modes of delivery, as were the participants in the non-tailoring condition. After exposure to the SSE instructions, participants answered questions on the outcome variables (recall, self-efficacy, and intention) and subsequently on personal relevance, the control variables, and the manipulation check. In addition, the participant’s personal characteristics (NFC, extraversion, and health literacy), media preferences, and demographic variables were measured. Similar to Study 1,
participants completed the questionnaire in approximately 15 to 20 minutes.
Measures
Personal characteristics (NFC: α = 0.68, M = 3.17, SD = 0.87; extraversion: α = 0.72, M = 4.50, SD = 0.51; health literacy: α = 0.88, M = 1.91, SD = 0.51), media preferences (internet (N = 193); television (N = 126); print (N = 75); radio (N = 43)), self-efficacy (α = 0.84, M = 5.66, SD = 1.09), frequency of performing SSE (M = 2.26, SD = 1.41), familiarity with the SSE instructions (yes (N = 181); no (N = 15); do not know (N = 4)), the manipulation check, and the demographic variables were measured in the same way as in Study 1 (see Study 1 and Appendix B).
Delivery mode preference. Participants’ delivery mode preference was measured in the same way as in Study 1, however, participants could also choose the answer option ‘no preference’. Participants preferred the text mode (N = 1), the text with illustrations mode (N = 71), or the video mode (N = 25). In addition, some participants did not have a preference for one of the three modes of delivery (N = 7).
Recall. Recall of the information was measured with 5 open questions which were generated from the information in the SSE instructions, such as ‘Could you briefly describe
26
what SSE is?’ and ‘Could you briefly describe how you can examine your thighs?’ The procedure to code the answers on the recall questions was the same as in Study 1. The
codebook that was used to assign scores to the answers can be found in Appendix C. For each participant, the scores on all questions were computed into a summed recall score. In addition, a percentage of the total score was calculated (M = 36.60, SD = 18.42).
Intention. The questions in Study 1 were specifically designed for the intention to perform BSE and were not applicable to SSE. Therefore, participants’ intention to perform SSE was measured with different items, namely with 5 items on a 7-point Likert scale (α = 0.97, M = 4.81, SD = 1.73) (Van Haasteren, 2014). The scale included items such as ‘I plan to perform SSE in the future’ and ‘I will try to perform SSE in the future’ (1 = very unlikely, 7 = very likely).
Expected difficulty. To measure the expected difficulty of the SSE instructions, the following question was asked: ‘How difficult do you think the instructions for performing skin self-examination will be?’ (1 = not at all difficult, 7 = very difficult) (M = 2.76, SD = 1.44).
Personal relevance. Personal relevance was measured with 4 items on a 5-point Likert scale (α = 0.85, M = 2.26, SD = 0.92) (De Vet, De Nooijer, De Vries, & Brug, 2007). The scale included items such as ‘The information was specifically directed to me’ and ‘I felt personally addressed’ (1 = completely disagree, 5 = completely agree).
Statistical Analysis
After examining the manipulation of tailoring and randomization, a binary logistic regression analysis was conducted to examine the predictors of delivery mode preference. Subsequently, one-way analyses of variance (ANOVAs) were conducted to examine the effects of tailoring on recall, self-efficacy, and intention. In addition, Process (a macro for SPSS) was used to investigate whether personal relevance mediated the effects of tailoring.
27
Results
Sample Characteristics
Two-hundred participants filled out the online questionnaire (N = 200). The age of these participants ranged from 18 to 74 (M = 33.19, SD = 13.38). Most participants were women (60.5%) and 54.5% of the participants were highly educated. Almost half of the participants were student (42.5%) and 85.5% had work (see Table 5 for an overview).
[Include Table 5 around here] Manipulation and Randomisation Checks
Manipulation check. A manipulation check was conducted to examine whether the two conditions (tailoring and non-tailoring) differed in the extent to which the received mode of delivery corresponded with the participant’s preferred mode of delivery. In the tailoring condition, the received instructions corresponded more with the participants’ delivery mode preference (M = 4.56, SD = 1.37) than in the non-tailoring condition (M = 3.82, SD = 1.45), t(198) = 3.71, p < .001, 95% CI [0.35, 1.14]. It can therefore be concluded that the
manipulation of tailoring had been successful.
Randomisation check. To verify whether the tailoring and non-tailoring conditions differed on gender, age, educational level, the extent to which participants already performed SSE, and the number of participants that were familiar with SSE instructions, independent t-tests and chi-statistics were conducted. Gender was equally divided over the tailoring and non-tailoring conditions, χ2(2, N = 200) = 0.04, p = .843, as was age, t(198) = -0.53, p = .599, 95% CI [-4.74, 2.74], educational level, χ2(2, N = 200) = 0.46, p = .796, the extent to which participants already performed SSE, t(198) = -0.02, p = .982, 95% CI [-0.40, 0.39], and the number of participants that were familiar with SSE instructions, χ2(2, N = 200) = 1.69, p =
28
.429. As controlling for these variables did not alter the effects, these variables were not included as covariates in further analyses.
The Determinants of Delivery Mode Preference
A binary logistic regression analysis was conducted to examine whether personal characteristics and media preferences predicted delivery mode preference. It was expected that individuals with a high NFC would prefer written instructions and that individuals with a low NFC would prefer audio-visual instructions (H1). However, as only one participant chose the text mode, the text mode was not examined with regard to its determinants. NFC did not predict a preference for text with illustrations or video, Exp(B) = 1.03, p = .954. H1 is therefore not supported. Next, it was expected that introvert individuals would prefer written instructions while extravert individuals would prefer audio-visual instructions (H2). The level or extraversion predicted delivery mode preference, Exp(B) = 0.14, p = .026, however, in the direction opposite to the hypothesis. Participants with a preference for the text with
illustrations mode were more extravert than participants with a preference for the video mode. H2 is therefore not supported. Furthermore, it was expected that high-literate individuals would prefer written instructions and that low-literate individuals would prefer audio-visual instructions (H3). However, health literacy did not predict delivery mode preference, Exp(B) = 0.66, p = .620. H3 is therefore not supported. Regarding media preferences it was expected that a preference for printed media would predict a preference for written instructions and that a preference for television would predict a preference for audio-visual instructions (H4). A preference for printed media did not predict delivery mode preference, Exp(B) = 0.40, p = .384. A preference for television, however, did predict delivery mode preference, Exp(B) = 0.01, p = .001. Participants that preferred television more often chose the video mode as compared to participants that did not prefer television. H4 is therefore partially supported.
29
Finally, gender predicted delivery mode preference, Exp(B) = 23.10, p = .015. Male participants preferred the video mode more often as compared to female participants. The Effects of Tailoring to Delivery Mode Preference
First, a one-way analysis of variance (ANOVA) was conducted to examine the effect of tailoring to delivery mode preference on recall. Results showed no effect of tailoring on recall of the information, F(1, 198) = 1.00, p = .318, 95% CI [33.99, 39.13]. Participants recalled the information worse when the SSE instructions were tailored to their delivery mode preference (M = 35.26, SD = 19.10) as compared to when they received the instructions in a random mode of delivery (M = 37.86, SD = 17.75), but this difference was not significant (see Table 6).
[Include Table 6 around here]
Second, a one-way analysis of variance (ANOVA) was conducted to examine the effect of tailoring to delivery mode preference on self-efficacy. Results showed no effect of tailoring on self-efficacy, F(1, 198) = 0.05, p = .826, 95% CI [5.50, 5.81]. Participants had a lower self-efficacy to perform SSE when the instructions were tailored to their delivery mode preference (M = 5.64, SD = 1.15) as compared to when they received the instructions in a random mode of delivery (M = 5.67, SD = 1.03). However, this difference was not significant (see Table 7).
[Include Table 7 around here]
Finally, a one-way analysis of variance (ANOVA) was conducted to examine the effect of tailoring to delivery mode preference on intention. There was no significant effect of tailoring on the intention to perform SSE, F(1, 198) = 0.56, p = .457, 95% CI [4.57, 5.05]. Participants that received the instructions in their preferred mode of delivery had a higher intention to perform SSE (M = 4.90, SD = 1.60) as compared to participants that received the
30
instructions in a random mode of delivery (M = 4.72, SD = 1.84), but this difference was not significant (see Table 8).
[Include Table 8 around here]
Although there were no direct effects of tailoring to delivery mode preference on recall, self-efficacy, and intention, there could be a full mediation via personal relevance. Therefore, mediation analyses were performed with the macro Process by Preacher and Hayes (2004). In this macro, model 4 was selected with tailoring as independent variable, recall as dependent variable and personal relevance as mediator. Personal relevance did, however, not mediate the effect of tailoring to delivery mode preference on recall of the information, B = 0.14, SE = 0.31, BC 95% CI [-0.16, 1.28]. Likewise, personal relevance did not mediate the effect of tailoring on self-efficacy, B = 0.01, SE = 0.02, BC 95% CI [-0.01, 0.08], or on intention, B = 0.08, SE = 0.11, BC 95% CI [-0.15, 0.30].
Post-hoc Analysis
Some participants (N = 7) in the tailoring condition did not have a preference for one of the three modes of delivery. Because these participants were randomly assigned to one of the three modes of delivery, they were analysed together with the participants in the non-tailoring condition. It could be argued, however, that these participants do not really belong to the non-tailoring condition because they had the opportunity to choose between the modes of delivery. Therefore, the effects of tailoring were examined again, first with these seven participants assigned to the tailoring condition and then with these participants excluded from the analyses. However, this did not change the results, as there were still no significant effects of tailoring on recall, self-efficacy, or intention.
31
Discussion
The aim of Study 2 was to unravel and expand the results of Study 1. First, this second study investigated whether delivery mode preference could be determined by the expected difficulty of the health instruction and by gender in addition to the personal characteristics and media preferences. Furthermore, this study examined personal relevance as possible
underlying mechanism of the effects of tailoring to delivery mode preference, and used different health instructions in order to generalise the results of Study 1.
This second study showed that delivery mode preference can be determined by several personal factors, namely extraversion, a preference for television, and gender. First,
participants that preferred written instructions were more extravert than participants that preferred audio-visual instructions. This result was in contrast with the expectations, because processing audio-visual material such as video would lead to much sensory stimulation which is preferred by extravert individuals (Digman, 1990). A possible explanation for the
contradicting results can be found in previous research. Fralick (2011) suggested that
extravert individuals are easily distracted and therefore learn best in an environment in which they actively have to process the information. Because written information has to be
processed more actively than audio-visual information, extravert individuals might prefer written information. Second, a preference for television predicted a preference for audio-visual information. This result is in line with the expectations, as previous studies found that individuals preferred various media that use the same mode of delivery (Gaglio et al., 2012; Heo & Cho, 2009). Clearly, people that prefer an audio-visual medium also prefer to receive health instructions via an audio-visual mode of delivery. Third, gender predicted delivery mode preference: male participants preferred audio-visual instructions more often than female participants. This corresponds with previous research which showed that women more often preferred to learn visually or via printed words, while men more often preferred to learn by
32
listening to information (Wehrwein et al., 2007). The other personal characteristics and media preferences did not predict delivery mode preference. Moreover, the expected difficulty of the instructions did not predict delivery mode preference although this was suggested in previous research (Wright et al., 2008). Clearly, people do not choose different modes of delivery for easy or difficult health instructions but instead, personal characteristics determine delivery mode preference. This suggests that individual variables are more important than situational variables in predicting delivery mode preference.
Furthermore, it was examined whether tailoring to delivery mode preference would lead to better recall of the information, a higher self-efficacy, and a higher intention to
perform SSE. No effects of tailoring were found. Participants that received the instructions in their preferred mode of delivery did not recall the instructions better, did not report a higher self-efficacy, or a higher intention to perform SSE as compared to participants that received the instructions in a random mode of delivery. Furthermore, personal relevance did not influence the relation between tailoring and recall, self-efficacy, and intention. The results of this study are partially in line with the results of Study 1. In Study 1 as well as in Study 2, no effects on recall and self-efficacy were found. In Study 1, however, there was a significant effect of tailoring on intention while this was not found in Study 2. A possible explanation for the discrepancy between the results of the two studies is that delivery mode tailoring is
effective for some health behaviours, but not for all. This possibility will be further discussed in the next section.
In this study, only one participant preferred the text mode. The finding from Study 1 that the text mode was not preferred for BSE instructions is thus also applicable to SSE instructions and male participants. In addition, a few participants reported that they had no preference for a specific mode of delivery. Although this is less than the 24% found by Rogers et al. (2008), it is interesting that some participants do not mind in which mode of
33
delivery they receive health instructions. More importantly, however, is that most people do have a specific delivery mode preference for receiving health instructions. The implications for these results will be discussed in the next section.
General Discussion and Conclusion
The aim of this research was twofold: (1) to explore the determinants of delivery mode preference for receiving health instructions and (2) to examine the effects of tailoring to delivery mode preference on recall of the information, self-efficacy, and the intention to perform the health behaviour that was advocated in the instructions. In Study 1, BSE
instructions were provided to female participants. In Study 2, SSE instructions were used, and both men and women could participate.
The results showed that several personal characteristics, media preferences, and demographic variables predicted delivery mode preference: NFC in Study 1 and extraversion, a preference for watching television, and gender in Study 2. From these studies, it can be concluded that several relatively stable, personal variables can determine delivery mode preference. In particular, individuals with a low NFC, who are introvert, prefer to watch television, and are men, prefer audio-visual instructions more often than individuals with a high NFC, who are extravert, do not prefer to watch television, and are women. Although the results of both studies showed that personal variables can predict the preference for a specific mode of delivery, there was no clear pattern of determinants across the studies. These
differences between Study 1 and Study 2 might have been found because different factors predict delivery mode preference in men and in women. Although there are too few
participants to be able to investigate this possibility with the current data, previous research showed an interaction between gender and personal characteristics in predicting media preferences (Chausson, 2010). Specifically, some predictors of movie genre preference were more important for men and other predictors were more important for women. Future research
34
should be conducted to find out whether this interaction is also applicable to delivery mode preference.
Regarding the effects of tailoring, Study 1 showed an effect of tailoring on the
intention to perform BSE, while Study 2 did not show an effect of tailoring on the intention to perform SSE. Although no previous studies investigated the effects of delivery mode tailoring on intention, some studies on content tailoring found an effect of tailoring on the intention to perform the advocated behaviour (Brug et al., 1998; Brug et al., 1996), while other research did not find such an effect (Drossaert, Boer, & Seydel, 1996). A possible explanation for the different effects on intention between Study 1 and 2 is that tailoring to delivery mode
preference is effective for some health behaviours (such as BSE), but not for other behaviours (such as SSE). Because BSE is a more personal, sensitive topic than SSE, people might be more personally involved with the instructions. As a result, they might find their own preferences more important when they receive BSE instructions as compared to SSE
instructions. A study by Shepherd, Tattersall, and Butow (2007) supports this possibility as it showed that cancer patients found it more important that their preferences were taken into account in breast cancer care as compared to other types of cancer care. Future research could explore this possibility.
Next to the differences between Study 1 and 2, it is interesting that no effects of
tailoring on recall and self-efficacy were found. As discussed, content tailoring is presumed to be more effective than non-tailoring because it increases personal relevance (Jensen et al., 2012), and subsequently information processing (Kreuter & Wray, 2003), which might in turn result in improved recall (Bandura, 1977; Klein et al., 2006), self-efficacy (Noar et al., 2010), and intention (Brug et al., 1998; Brug et al., 1996). If delivery mode tailoring would work in the same way as content tailoring, there should be effects on recall and self-efficacy as well. Delivery mode tailoring might therefore work in a different way than content tailoring, which
35
is supported by the absence of effects of personal relevance in Study 2. A possible alternative mechanism is that people are more satisfied with the instructions when their personal
preferences are taken into account. This is supported by Floyd and Moyer (2010), who concluded that participants were more satisfied with an intervention when they were able to choose a mode of delivery as compared to when they could not choose. Satisfaction could subsequently lead to a higher intention to perform the advocated behaviour without an improved recall or self-efficacy. Previous research also showed that satisfaction with
communication could directly improve health behaviour (Ley, 1989). A second possibility is that tailoring to preference only affects the intention because of empowerment. Empowerment is the belief that one has control or power over something (Zimmerman, 1995). Several
studies found that empowerment improved the intention to perform health behaviour (Cooper, Goldenberg, & Arndt, 2011; Luszczynska, Durawa, Scholz, & Knoll, 2012). Because
participants in the tailoring condition were able choose a mode of delivery themselves, they could have experienced empowerment which might have improved their intention to perform BSE without a better recall or a higher self-efficacy. Future research could investigate
whether satisfaction and/or empowerment can explain the effects of delivery mode tailoring.
Limitations
This research has some limitations. First, more than half of the participants in both studies were highly educated. The determinants of delivery mode preferences and the effects of mode tailoring might differ between higher and lower educated people and therefore, the results of these studies cannot be generalised to the Dutch population. To improve external validity, future research could investigate whether the results hold true in a more
representative sample regarding educational level.
Second, as the studies were conducted online, participants read and watched the instructions via a website. However, people process online information differently (with less
36
attention and less linearly) than offline information (Liu, 2005). In future research, the written instructions could be presented on paper and the audio-visual instructions could be shown on a television screen to see whether the effects are the same.
Third, the ultimate goal of studies on the effectiveness of health instructions is to increase the number of people that perform health examinations. The current studies, however, did not examine the effects of tailoring on actual behaviour (performing BSE and SSE). Based on these studies, only conclusions about recall, self-efficacy, and the intention to perform these behaviours can therefore be drawn. Future research should investigate whether tailoring to preference has an effect on behaviour as well.
Conclusions and Implications
This innovative study is a valuable contribution to theory. First, personal variables can determine delivery mode preference. Future research can build on this research by further investigating these and other predictors. Second, the inconsistent results from previous research on the CTML can be partially explained by tailoring. The intention to perform BSE does not depend on a specific mode of delivery, but on providing each individual with the instructions in the preferred mode of delivery.
The results of the current studies offer implications for practice as well. When designing health interventions, it may not always be possible to let each individual choose between different modes of delivery. Personal characteristics, media preferences, and demographic variables can then be used to provide each individual with a mode that is tailored to his or her preference. This may be especially fruitful in an online environment where a computer can automatically provide the health intervention in a specific mode of delivery based on an assessment of the individual’s personal characteristics, media
preferences, and demographic variables. Second, it is recommended to add illustrations to written health instructions as people prefer written health information that is supported by
37
illustrations. Providing the instructions in the preferred mode of delivery showed to be important for the intention to perform the behaviour in Study 1.
Although more research on the determinants of delivery mode preference and the effects of tailoring to delivery mode preference is needed, the results of these studies show that taking into account the individual’s delivery mode preference can increase the
effectiveness of health instructions. This suggests that tailored health communication might be advanced by moving beyond content tailoring and by further exploring delivery mode tailoring.
References
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179-211. doi:10.1016/0749-5978(91)90020-T
Ajzen, I. (2006). Theory of planned behaviour. Retrieved from http://www.people.mass.edu/aizen/tpb.html
Ashton, M. C., Lee, K., Perugini, M., Szarota, P., De Vries, R. E., Di Blas, L.,… De Raad, B. (2004). A six-factor structure of personality-descriptive adjectives: Solutions from psycholexical studies in seven languages. Journal of Personality and Social Psychology, 86, 356-366. doi:10.1037/0022-3514.86.2.356
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191-215. doi:10.1037/0033-295X.84.2.191
Bandura A. (1986). Social foundations of thought & action: A social cognitive theory. Englewood Cliffs, NJ: Prentice Hall.
