October 2011 // Volume 49 // Number 5 // Feature // 5FEA5
Intention to Consume Fruits and Vegetables Is Not a Proxy for Intake in Low-Income Women from Pennsylvania
Intention as an outcome measure for fruit and vegetable nutrition education interventions in low-income women was assessed through dietary assessment 3 weeks after a fruit and vegetable intervention in a federally funded program. Amount and variety of intake were compared to intentions expressed immediately following intervention. Findings suggested intentions did not approximate self-reported intake. For example, of 85 women indicating little baseline fruit variety, 47 intended to increase variety following the intervention; only two met the anticipated improvement. In all, only 25 participants met one or more intentions to improve intake. Implications for measuring intention and assessing outcomes are discussed.
The theory of planned behavior (TPB) as developed by Ajzen (1991) has driven intervention design for a variety of health-related topics and is widely used to address fruit and vegetable intake in adults (Guillaumie, Godin, & Vézina-Im, 2010). Contento (2007) described and modeled application of the TPB to nutrition education, suggesting that behaviors (e.g., intake) are more likely when intention to do so is evident. Behavioral intention is a function of attitude toward behavior, perceived social norms toward the behavior, and perceived self-efficacy to accomplish the behavior. Theory examination has revolved around measuring theoretical components, (e.g., attitude, perceived behavioral control) that influence intention to act.
However, the role of intention as a proxy for behavior has been questioned (Rothman, 2009). Snack choice intention did not predict snack selection among adult office employees in the Netherlands (Weijzen, de Graaf, & Dijksterhuis, 2008). Eating fruits and vegetables did not follow from intention to do so in a sample of young Norwegian men and women (Kaavik, Lien, Tell, & Klepp, 2005) or middle-aged Dutch women (Bogers, Brug, van Assema, & Dagnelie, 2004). In fact, despite an interest in healthy eating and fruit and vegetable intake, intention in the Dutch sample did not predict self-rated fruit consumption and was negatively related to self-rated vegetable consumption. A positive association was reported between fruit and vegetable consumption and intention to consume among college students; however, its value was tempered by the fact that all students were recruited from health and fitness classes and interested in fruit and vegetable intake (Blanchard et al., 2009).
The belief that intention equates with behavior is strongly held; intention has substituted for behavioral outcomes in studies informing nutrition intervention development for fruits and vegetables (Pawlak & Malinauskas, 2008), a plant-based diet (Wyker, Kirsten, & Davison, 2010), and a healthful diet (Pawlak, Malinauskas & Rivera, 2009).
Measuring nutrition intake intention is convenient, inexpensive, and encumbers little respondent burden. Thus, validation and promotion of the post/retrospective-pre assessment strategy (Rockwell & Kohn, 1989) marked the use of intention as a sound measure of behavior. Given recent concerns about intention as a substitute measure of behavior, we used a post/retrospective-pre-approach and a delayed measure of behavior to examine both intention and behavior in a federally-funded nutrition education intervention.
Association of dietary behavior with intention was assessed in a two-stage, post-education design. Assessment included survey administration immediately following a nutrition education session and three-telephone dietary recalls. As shown in Figure 1, each participant experienced one educational session focused on either 1) vegetables or 2) vegetables and fruit. Educational formats followed each agency's standard protocol. Common educational elements included a learner-centered, experiential approach, discussion, and focus on skill development. The educational experience served as a point of recruitment, with educators informing learners of the study before the educational session and inviting participation at its conclusion.
Program educators received training on the study protocol and were instructed to recruit persons 18 through 50 years of age from households in which food purchase and selection decisions were made by household members. Participants were recruited from 63 separate small group classes or one-on-one education sessions conducted by 18 educators as part of Supplemental Nutrition Assistance Program Education (28.6%) and Expanded Food and Nutrition Education Program (71.4%) programming. The Office of Research Protections at The Pennsylvania State University approved this study.
Data Collection Protocol
Data Collection and Measures
Figure 1 shows the data collection protocol for the overall study and highlights the individual participant data collection experience. Intention data were collected with a six-item survey, in a retrospective pre/post format (Rockwell & Kohn, 1989). Items addressed current (also termed retrospective-pre) and intended (also known as post) behaviors for fruits and vegetables regarding variety, snack intake, and overall daily amounts. Variety and snack item response sets included four choices: Almost always, usually, sometimes, and hardly ever. For example, participants were asked, "How often do you eat more than one kind of fruit each day?" and were instructed to answer for Before Today and to indicate their intentions After Today. Categorical responses for intake (in cups) were > 2.5, 1.5 to 2.5, .5 to 1.5, < .5, and "not sure" for fruits and > 3.0, 2.0 to 3.0, 1.0 to 2.0, < 1.0, and "not sure" for vegetables.
Experts confirmed content validity and cognitive interviews with a representative sample (n=20) revealed face validity and format fidelity. Field-testing with 815 SNAP-Ed participants showed that data collection and analysis was feasible. Cronbach's α, a statistical test to assess uniformity of responses, was acceptable for both the retrospective-pre (Cronbach's α=.82) and post (Cronbach's α=.81) survey components. Socio-demographic data included age, ethnicity, race, education, and Supplemental Nutrition Assistance Program participation status. Program educators bulk-mailed completed surveys to researchers in addressed, stamped envelopes; class, location, nutrition program, and instructor identifiers were included.
Dietary intake was assessed from three 24-hour dietary recalls collected during telephone interviews performed by trained staff at The Pennsylvania State University Diet Assessment Center. All interviewers completed 40 hours of intensive training (Smiciklas-Wright, JADA 1991). Dietary recalls occurred on random days over a 3-week period with the calls beginning 3 weeks after the educational experience. Average call length was 20 minutes. Recall data were collected and analyzed using the Nutrition Data System for Research (NDS-R) software version 2006, developed by the Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN. NDS-R provides structured prompts that standardize data entry using a multiple pass system of interview methodology.
Food security was measured with the six-item Short Form of the USDA Household Food Security Scale, administered at time of second dietary recall. The Short Form consists of statements about household food availability or affordability; households listing more than one item were scored as food insecure (Bickel, Nord, Price, Hamilton, & Cook, 2000).
Six dietary intake variables were constructed from 24-hour recall data: daily fruit, daily vegetables, fruit variety, vegetable variety, fruit snacks, and vegetable snacks. Daily fruit and vegetable variables were computed as mean cup equivalents, averaged from three 24-hour recalls. Average daily intake was recoded into a categorical variable with values that reflected a response range as denoted on the retrospective pre/post survey, (e.g., 1 to 2 cups was a category, and more than 2 to 3 cups was another category). Variety was assessed independently by two expert reviewers; a third expert reconciled differences. Daily variety was defined as consuming more than one kind of vegetable (fruit) in any amount as a meal, snack, or ingredient. For example, consuming potato salad that included onions would denote variety for that day. Number of days in which variety was observed could range from 0 to 3. A vegetable (fruit) snack was denoted by the presence of vegetables (fruit) in any amount consumed daily as a snack. Thus, variable values ranged from 0 (no vegetable/fruit snack consumption on any day) to 3 (vegetable/fruit snack consumption on all 3 recall days).
Intention to consume a specific amount of vegetables and fruits each day (as reported on the post/pre survey) was compared to diet recall responses using a schema shown in Table 1. Another schema interpreted diet recall variety and snack intake in the context of intention to consume vegetables and fruits (Table 1). Findings about intention and behavior were analyzed only for participants with 3 completed diet recalls (completers). Data were analyzed using descriptive statistics, Chi Square, and t-tests as appropriate. (SPSS, version 14.0, 2005, SPSS Inc, Chicago, IL). Statistical significance was indicated by p < .05.
|Amount of Vegetables (Fruit) Consumed|
|Daily mean3 intake in cup equivalents||Reported Dietary Intention on Post/Pre Survey|
| ≥3 cups|
|2 - 3 cups |
(1.5 - 2.5 cups)
|1 to 2 cups |
(0.5 - 1.5 cups)
|< 1 cup
(< 0.5 cup)
| ≥ 3 cups vegetables|
(≥ 2.5 cups)
|Between 2 and 3 cups vegetables|
(Between 1.5 and 2.5 cups)
|1 to 2 cups vegetables |
(0.5 to 1.5 cups)
|Less than 1 cup vegetables (Less than 0.5 cup)||Met|
|Dietary Intake Behaviors of (1) Variety and (2) Consumption as Snacks|
|Number of diet recalls indicating presence of vegetable/fruit||Reported Dietary Intention on Post/Pre Survey|
|Almost Always||Usually||Sometimes||Hardly Ever|
|All three recall days||Met||Met||Met|
|Two recall days||Met||Met|
|One recall day||Met||Met|
|Zero recall days||Met|
|1 "Met" indicates that
intention reported on Post/Pre survey was met by the corresponding
behavior reported through dietary recall.|
2Amounts for fruit shown in parentheses.
3 Mean of three days
The 133 participants who provided three diet recalls were more likely than the 56 participants who completed fewer than three diet recalls (non-completers) to have received education or training beyond high school (χ2 = 3.973, p = 0.046.). Completers and non-completers did not differ in age, race, gender, food assistance program participation during the past year, recruitment period, and receipt of a high school diploma or GED. Completers who provided demographic information were primarily non-Hispanic (96.1%), white (61.8%), female (93.2%), and high school or GED graduates (75.2%); 59.5% received education or training beyond high school. The 18-30 year old age group comprised 53.8% of the sample, and 46.2% were aged 31-50. Most were Supplemental Nutrition Assistance Program participants within the past year (68.5%) and food secure (60.9%).
Post/retrospective-pre survey responses indicated that 33.8% of completers were (almost always/usually) consuming a variety of fruit at baseline; 66.2% were not (hardly ever/sometimes). Snacks that included fruit were eaten by 27.8% and 30.2% consumed more than 1.5 cups of fruit each day. Baseline perceptions of participants included consuming a variety of vegetables (59.4%), vegetable-inclusive snacks (15.0%), and more than 2 cups of vegetables each day (23.3%).
Intentions to Consume as Predictors of Dietary Behavior
More participants met intentions related to vegetable variety, quantity, and use as a snack than intentions for these behaviors related to fruits. As shown in Table 2, no more than two-thirds of participants met any fruit or vegetable intake intention; in fact, with the exception of vegetable variety, intake intentions were met by fewer than half of the participants. Of note is the inclusion of static inaction in this count. Static inaction refers to not practicing recommended dietary behaviors and denoting no intention to change dietary behaviors. Static inaction was identified for 27 (20.3%) participants, i.e. change was not intended and indeed they didn't change intake behavior. This phenomenon of static inaction did not apply to any specific fruit or vegetable response.
|Dietary Intake Variables||Met Intention (%, n)|
|Vegetable Variety (N=131)||66.4, 87|
|Vegetable Daily Amount (N=124)||44.4, 55|
|Vegetable Snack (N=131)||29.8, 39|
|Fruit Variety (N=129)||20.2, 26|
|Fruit Daily Amount (N=117)||24.8, 29|
|Fruit Snack (N=130)||19.2, 25|
|1Sample sizes vary due to missing responses.|
Participant characteristics unrelated to meeting dietary intentions included education/ training beyond high school, and Hispanic ethnicity. Significant relationships among participant characteristics and achievement of dietary intentions did not follow a pattern (as shown in Table 3). Only age was associated with meeting intentions for more than one of the intake behaviors, i.e. fruit amount and fruit variety.
|Intended Fruit and Vegetable Intake Behaviors||Participant Characteristics|
|SNAP Participation1||Age2||High School Diploma or GED3||Program Type4||Race5||Food Security Status6|
|Vegetable Variety||χ2 = 4.159, p=.041||χ2 = 4.389, p=.036|
|Vegetable Daily Amount||χ2 = 3.914, p=.048|
|Fruit Variety||χ2 = 4.813, p=.028||χ2 = 6.595, p=.010|
|Fruit Daily Amount||χ2 = 5.260, p=.022||χ2 = 4.468, p=.035|
who had not
participated in the Supplemental Nutrition Assistance Program (SNAP)
over the past year were more likely to meet their vegetable |
2 Participants who had completed high school or received a GED were more likely to meet their vegetable variety intention.
3 Participants who were recruited from EFNEP classes were more likely to meet their daily amount of vegetables intention.
4 Participants in the "18-30" age category were more likely to meet their fruit variety intention and intention for fruit cups.
5 Participants who were food insecure were more likely to meet their fruit variety intention.
6 Non-white participants were more likely to meet their intention for fruit cups.
Of the eighty-five participants who reported not including a variety of fruit in their diet at baseline, 47 (55.3%) indicated intending to increase the variety of fruit following educational intervention. However, only two of these 47 actually met the intended behavior; i.e., rather than a success rate of 55.3%, the educational intervention was successful for only 2.4%. Comparing intervention effectiveness using intention versus behavior for the five remaining fruit and vegetable dietary behaviors revealed similar findings. Of 52 not consuming fruit as a snack, but intending to do so, only 9.6% met their intention. Two of the 36 participants (5.6%) who were not consuming greater than 1.5 cups of fruit daily but intended to do so met their stated intention. Of 25 who did not consume more than one kind of vegetable daily, but intended to do so, 16 (64%) met their goal. Only two of the 39 (5.1%) who didn't eat vegetables as snacks but reported an intention to begin the practice, actually did. Six of the 32 participants (18.8%) who were not consuming greater than 2 cups of vegetables daily but intended to do so met their stated intention.
In all, meeting at least one intention to improve dietary intake of fruits and/or vegetables represented only 25 participants. These 25 participants were mostly food secure (72%), white (64%), high school graduates (92%), with 60% having education beyond high school.
In this sample of 133 adults participating in an educational experience typical of a federally funded nutrition education program, few (25 or 18.8%) achieved an intended change in fruit or vegetable intake behavior, in amount, variety, or function. This finding is of interest because measurement of intention is a recommended nutrition education outcome measure, owing to use of the TPB in nutrition intervention development (Guillaumie, Godin & Vézina-Im, 2010) and reported advantages of the post/retrospective-pre assessment strategy (Rockwell & Kohn, 1989; Jayaratne, 2010).
Our findings are surprising because we were not measuring results of a specific intervention, but results from nutrition education following a field agency's typical practice of education delivery. Participants were recruited only from locations providing a lesson on fruits and vegetables or only vegetables. In addition, two separate recruitment periods were designed to account for seasonality of fruits and vegetables; the first recruitment period was held from July through August, and the second recruitment period was held from October through November. Intake of only two types (for fruits or vegetables) was liberally labeled as showing "variety." Fruit or vegetable snacking was defined by intake of "any amount" of fruit or vegetable.
Our snack and variety variables were constructed from our previous formative research with the target audience that examined what the response options almost always, usually, sometimes, and hardly ever represented to them. (Responses indicated that almost always meant approximately 6-7 times a week, usually meant 4-5 times a week, sometimes was 2-3 times a week, and hardly ever represented 0-1 times/week.) Finally, dietary data were collected via telephone by professionally trained, experienced, and unbiased assessment staff utilizing three 24-hour recalls.
Interpretation of our findings was limited because association to a specific lesson, instructor, or protocol was not possible. In addition, we knew that nutrition education interventions typical of these federally funded programs routinely incorporate knowledge, motivation, personalization, self-efficacy, and taste testing, but specific lesson content and process information were not available.
Guillaumie, Godin, & Vézina-Im (2010) note that intervention components or variables predictive of behavior may not be predictive of intention. Intention is a complex phenomenon; temporal stability, implementation intentions, and context mediate the impact of intention on behavior so that merely capturing the presence of intention (as is common in many nutrition education programs) may be an inadequate and misleading outcome measure (Webb & Sheeran, 2006). Implementation intentions detail when, where, and how one will perform the intended behavior and have been shown to be useful to increase fruit intake (de Nooijer, de Vet, Brug, & de Vries, 2006). Our intention measure assessed only behavioral intention; including implementation intentions in our assessment may have improved the behavioral predictions from intention.
Self-determination theory posits that self-determined motivation, which is a more sophisticated form of intention, is a better predictor of behavior, including dietary behavior (Pelletier, Dion, Slovinec-D'Angelo, & Reid, 2004) and that intrinsic goals motivate behavior (Deci & Ryan, 2008). Our measure of intention did not differentiate goals or motivation source.
Contextual impact of intention on behavior is an important consideration (Shaikh, Yaroch, Nebeling, Yeh, & Resinow, 2008). Relational culture is the emphasis on interdependence and interconnectedness among individuals and groups and prioritization of these connections (Pasick & Burke, et al., 2009). Pasick & Barker, et al. (2009) in their examination of mammography screening and health care utilization by Latina and Filipina women argued that behavioral intention is not a universal construct, but subject to relational culture. Stewart, Rakowski, & Pasik (2009) noted that intention was a predictor of behavior for white women to have mammography, but not for non-whites. Mediational analyses of contextual factors related to fruit and vegetable behavior (e.g., gender, age, socioeconomic status, social support) will better predict fruit and vegetable intake. Higher quality psychosocial and behavioral measures and use of affective measures (e.g. taste or pleasure) have been suggested to better predict fruit and vegetable intake (Guillaumie, Godin, & Vézina-Im, 2010).
Sutton (2008) diagrams a less direct route between intention and behavior than the theory of planned behavior. In one path, intention directly leads to planning, which in turn leads to behavior. Another path bypasses intention altogether, suggesting that conviction to re-adopt a behavior directly affects either planning or behavior. Thus, a measure of intention in an intervention that affects this recovery may be a poor indicator of behavior.
Implications for Practice
The study of low-income participants in federally funded nutrition education programs in Pennsylvania reported here demonstrated that a measure of behavioral intention alone was not a valid measure of nutrition behavior. As a result, to correctly identify outcomes, we urge that our limited evaluation resources support 1) development and application of direct behavior assessments and 2) the design of nutrition education programs more likely to effect behavior change by addressing motivation, goal setting, implementation intention, skill building, learner readiness, social support, and interconnectedness.
For example, standard nutrition education practice has been to develop and release a comprehensive nutrition education program for low-income mothers, tested only for face validity and usability and including an intention-based survey. Our findings support social, technical, and fiscal capital allocation to program development that accommodates experience, learner readiness, goals, implementation intention, and social support of low-income women as well as robust evaluation using valid and reliable behavioral assessments. Accurately aimed nutrition education programs with multi-modal, intensive behavioral assessments may better cultivate nutrition education efficacy than current broadly defined, population-based approaches that require assumptions about low-income women's interests, skills, needs, experience, and food practices.
The authors acknowledge the assistance of Diane Mitchell, Coordinator of The Pennsylvania State University Diet Assessment Center, and nutrition educators from Adagio Health, Montgomery County Community Action Development Commission, and Penn State Cooperative Extension Nutrition Links. Funding provided by the Pennsylvania Department of Public Welfare through the Pennsylvania Nutrition Education TRACKS as part of USDA's Supplemental Nutrition Assistance Program.
Ajzen, I. (1991). The theory of planned behavior. Organizational behavior and human decision processes, 50,179-211.
Bickel, G., Nord, M., Price, C., Hamilton, W., & Cook, J. (2000). Guide to measuring household food security. United States Department of Agriculture, Food and Nutrition Services. Retrieved from: http://www.fns.usda.gov/fsec/FILES/FSGuide.pdf
Blanchard, C. M., Fisher, J., Sparling, P. B., Shanks, T. H., Nehl, E., Rhodes, R. E., et al. (2009). Understanding adherence to 5 servings of fruits and vegetables per day: A theory of planned behavior perspective. Journal of Nutrition Education and Behavior, 41, 3-10.
Bogers, R. P., Brug, J., van Assema, P., & Dagnelie, P. C. (2004). Explaining fruit and vegetable consumption: The theory of planned behaviour and misconception of personal intake levels. Appetite, 42, 157-166.
Contento, I. R. (2007). Nutrition education, Linking research, theory, and practice. Sudbury, MA: Jones and Bartlett.
De Nooijer, J., de Vet, E., Brug, J., & de Vries, N. K. (2006). Do implementation intentions help to turn good intentions into higher fruit intakes? Journal of Nutrition Education and Behavior, 38, 25-29.
Deci, E. L., & Ryan, R. M. (2008). Self-determination theory: A macrotheory of human motivation, development, and health. Canadian Psychology, 49, 182-185.
Guillaumie, L., Godin, G., & Vézina-Im, L. (2010). Psychosocial determinants of fruit and vegetable intake in adult population: A systematic review. International Journal of Behavioral Nutrition and Physical Activity, 7, 1-12. Retrieved from: http://www.ijbnpa.org/content/pdf/1479-5868-7-12.pdf
Hill, L. G., & Betz, D. L. (2005). Revisiting the retrospective pretest. American Journal of Evaluation. 26, 501-517.
Jayaratne, K. S. U. (2010). Practical application of aspiration as an outcome indicator in extension evaluation. Journal of Extension [On-line], 48(2) Article 2TOT1. Available at: http://www.joe.org/joe/2010april/tt1.php
Kvaavik, E., Lien, N., Tell, G. S., & Klepp, K. I. (2005). Psychosocial predictors of eating habits among adults in their mid-30s: The Oslo youth study follow-up 1991-1999. International Journal of Behavioral Nutrition and Physical Activity, 2, 1-9. Retrieved from: http://www.ijbnpa.org/content/2/1/9
Pasik, R. J., Barker, J. C., Otero-Sabogal, R., Burke, N. J., Joseph, G., & Guerra, C. (2009). Intention, subjective norms, and cancer screening in the context of relational culture. Health Education and Behavior, 36, 91S- 110S.
Pasik, R. J., Burke, N. J., Barker, J. C., Joseph, G., Bird, J. A., Otero-Sabogal, R., et al. (2009). Behavioral theory in a diverse society: Like a compass on Mars. Health Education and Behavior, 36, 11S- 35S.
Pawlak, R., & Malinauskas, B. (2008). Predictors of intention to eat 2.5 cups of vegetables among ninth-grade students attending public high schools in eastern North Carolina. Journal of Nutrition Education and Behavior, 40, 392-398.
Pawlak, R., Malinauskas, B., & Rivera D. (2009). Predicting intention to eat a healthful diet by college baseball players: Applying the theory of planned behavior. Journal of Nutrition Education and Behavior, 41, 334-339.
Pelletier, L. G., Dion, S. C., Slovinec-D'Angelo, M., & Reid, R. (2004). Why do you regulate what you eat? Relationships between forms of regulation, eating behaviors, sustained dietary behavior change, and psychological adjustment. Motivation and Emotion, 28, 245-277.
Rockwell, S. K., & Kohn, H. (1989). Post-then-pre evaluation. Journal of Extension [On-line], 27(2) Article 2FEA5. Available at: http://www.joe.org/joe/1989summer/a5.php
Rothman, A. J. (2009). Capitalizing on opportunities to refine health behavior theories. Health Education and Behavior, 36, 150S-155S.
Shaikh, A. R., Yaroch, A., L., Nebeling, L., Yeh, M., & Resnicow, K. (2008). Psychosocial predictors of fruit and vegetable consumption in adults, a review of the literature. American Journal of Preventive Medicine, 34, 535-543.
Smiciklas-Wright, H., Mitchell, D. C., Norton, L. D., & Derr, J. A. (1991). Interviewer reliability of nutrient intake data from 24-hour recalls collected using the Minnesota Nutrition Data System. Journal of the American Dietetic Association, 91(suppl 9), 28.
Smiciklas-Wright, H., Mitchell, D., & Ledikwe, J.H. (2001). Dietary intake assessment: Methods for adults. In C. D. Berdanier (Ed.), CRC handbook of nutrition and food (pp. 447 - 493). Boca Raton, FL: CRC Press.
Stewart, S. L., Rakowski, W., & Pasick, R. (2009). Behavioral constructs and mammography in five ethnic groups. Health Education and Behavior, 36, 36S- 54S.
Sutton, S. (2008). How does the health action process approach (HAPA) bridge the intention-behavior gap? An examination of the model's causal structure. Applied Psychology, 57, 66-74.
Webb, T. L., & Sheeran, P. (2006). Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence. Psychological Bulletin, 132, 249-268.
Weijzen, P. L. G., de Graaf, C., & Dijksterhuis, G. B. (2008). Discrepancy between snack choice intentions and behavior. Journal of Nutrition Education and Behavior, 40, 311-316.
Wyker, B. A., & Davison, K. K. (2010). Behavioral change theories can inform the prediction of young adults' adoption of a plant-based diet. Journal of Nutrition Education and Behavior, 42, 168-177.