- Email: [email protected]
Perceived Benefits and Barriers of Increased Fruit and Vegetable Consumption: Validation of a Decisional Balance Scale
R E S E A R C H A RT I C L E ...............................................................................................
Perceived Benefits and Barriers of Increased Fruit and Vegetable Consumption: Validation of a Decisional Balance Scale A N N I E M E I C H UA N L I N G 1 A N D C A RO L I N E H O RWAT H 2 1
2
Department of Nutrition, Level 5, Institute of Health, Singapore 168937; Department of Human Nutrition, University of Otago, Dunedin, New Zealand
ABSTRACT Objective: To develop and validate scales to assess perceived benefits and barriers (decisional balance) for increasing fruit and vegetable consumption. Design:
A cross-sectional mail and telephone survey was conducted.
Subjects/Settings: A total of 1200 Chinese households were randomly selected from the Singapore residential telephone listings, and 71% responded to the mail survey; 390 males and 406 females participated (mean age = 39.3). Main Outcome Measures:
Decisional balance, stage of change, and fruit and vegetable consumption were measured.
Statistical Analyses Performed: Using a split-half sample approach, developmental sample responses were analyzed by principal-components analysis and validation sample responses by confirmatory factor analysis.Analyses of variance were used to examine stage differences in decisional balance. Results: Principal-components analysis indicated two components representing benefits (or pros) (Cronbach’s α = 0.86) and barriers (or cons) (α 0.79) of change. Confirmatory factor analysis strongly supported the two-component structure (Goodness of Fit Index = 0.97).There was a shift from cons to pros being more important across the stages.The increase in pros across the stages of change (p < .0001) corresponded to a medium effect size, and the decrease in cons (p < .01) corresponded to a small effect size. Implications:
Decisional balance scales may be used to guide interventions to influence fruit and vegetable consumption.
KEY WORDS:
benefits, barriers, stage of change, fruit, vegetables (JNE 33:257–265, 2001)
INTRODUCTION
Relatively little, however, is known about what might motivate people to eat more fruits and vegetables. Understanding how motivational considerations can influence the process of changing fruit and vegetable consumption may be enhanced by applying the decisional balance construct of Janis and Mann’s Decision-Making Model.9 Decision making was regarded as a comparative process involving balancing four categories of gains (instrumental gains for self and others and approval for self and others) against four categories of costs (instrumental costs for self and others and disapproval from self and others). Transtheoretical Model (TTM) researchers have used the decisional balance construct to examine how motivational considerations are related to stage of readiness to change.10 According to the model, behavior change progresses through
The dietary guidelines of many countries1 encourage fruit and vegetable consumption since there is compelling evidence that higher intakes are associated with reduced cancer risk.2–4 In Singapore, as in the United States5 and Britain,6 fruit and vegetable consumption falls well below recommended levels.7,8 ................................................... This work was sponsored by the National Medical Research Council, Singapore and the International Life Sciences Institute, South East Asia. Address for correspondence: Annie Ling, B.Sc., M.Sc., Ph.D., Department of Nutrition, Level 5, Institute of Health, 3 Second Hospital Avenue, Singapore 168937; Fax: (65) 438-3605; E-mail: [email protected]. ©2001 SOCIETY FOR NUTRITION EDUCATION
257
258
Ling and Horwath/PERCEIVED BENEFITS AND BARRIERS OF INCREASED FRUIT AND VEGETABLE CONSUMPTION
five stages. During the early stages of precontemplation and contemplation, change is conceptualized as cognitive (moving from no intention of changing to considering it).Those in the preparation stage plan to change in the near future, those in the action stage have recently adopted the new behavior, and those in the maintenance stage are working at sustaining the change. Research across diverse health behaviors has demonstrated two orthogonal factors representing the perceived benefits (pros) and barriers (cons) of change11–16 rather than the eight-factor structure hypothesized by Janis and Mann.9 These and other studies17,18 have revealed cons to outweigh pros for precontemplators, whereas pros outweigh cons for those in the action and maintenance stages. Longitudinal studies of smoking cessation have demonstrated decisional balance to be a strong predictor of movement, especially through the early stages of change.19 Dietary applications of the decisional balance construct within a TTM framework have focused on dietary fat reduction.14,15,18,20,21 This is one of the first studies to examine decisional balance in relation to stage of readiness to increase fruit and vegetable consumption. In a study of African Americans, Campbell et al.22 reported that barriers to change were more prevalent among precontemplators than those contemplating or planning change. Perceived benefit, assessed using one item, was generally high across all stages. Qualitative research using focus groups has suggested the main perceived benefits of a higher fruit and vegetable intake to be nutritional value, a sense of well-being (e.g., digestive regularity, feeling light/refreshed, and doing some good for oneself), health, and weight loss.23–27 Common barriers to greater intake appeared to be cost in time and money, effort to prepare, difficulty in changing habits, preference for other foods, lack of availability, concern for pesticides, and lack of awareness of what constitutes sufficient amounts to be consumed. Quantitative research examining perceived benefits of fruit and vegetable consumption has involved measures limited in coverage of the construct domain mainly to items concerning nutrient value and health benefits.22,28,29 Scales have often involved general belief statements (e.g.,“fruits and vegetables are good sources of fiber”),28 rather than personalized items with clear statement of the goal behavior (e.g.,“eating more fruits and vegetables would help me avoid constipation”), and were more likely to tap knowledge than value appraisal of outcomes.Although a few studies have reported low (0.48) to moderate (0.70) alpha coefficients for the benefit and barrier scales,28,30 none has established the factor structure of decisional considerations for fruit and vegetable consumption. Nevertheless, in one study, barriers accounted for the greatest variability in consumption28; another found those in the highest intake tertile to have more positive attitudes,30 and yet another failed to find an association between beliefs about the healthfulness of fruits and vegetables and intake levels.29 This study aims to fill a gap in research by developing a scale to measure the pros and cons of decision making for eating more fruits and vegetables. Using a split-sample technique,
developmental sample responses were analyzed by principalcomponents analysis to determine an initial dimensional structure, and validation sample responses were analyzed by structural modeling procedures to confirm the structure.The scales were also evaluated for their ability to differentiate individuals at varying stages of readiness to change. As nutrition beliefs and attitudes have been reported to vary among men and women,31,32 gender differences were also examined.
METHODS Sample and study design. The Singapore residential telephone listings were used to randomly select 1200 Chinese households. Each household was mailed both Chinese and English versions of a self-administered questionnaire with a cover letter and postage-paid return envelope. Up to three reminders were sent to nonrespondents at 10 days, 3 weeks, and 5 weeks after the initial mailing,33 including replacement questionnaires with the second reminder. Cover letters were randomly assigned to requesting participation by either a male or female adult aged 21 to 60 years. If no one of the specified age and gender was present in the household, a person of the opposite gender was invited to participate.To encourage participation, $10 telephone cards were promised on return of the completed questionnaire. For those who responded, a short telephone interview followed to assess stage of change. Each household was called up to seven times. Measures. As part of a larger study investigating factors influencing fruit, vegetable, and grain consumption,34 the questionnaire included a validated fruit and vegetable food frequency questionnaire (FFQ),8 a decisional balance scale, and questions on demographics, supplement use, general beliefs, meal venues, and health conditions. Stage of change algorithm. Based on the combined fruit and vegetable estimates from the FFQ, subjects were asked the length of time they had been at their current level of intake (for those having adequate numbers of servings) or whether they intended to increase intake to meet the Singapore dietary guideline of four servings per day (excluding juices).35 Subjects consuming adequate servings but for less than 6 months were classified as being in the action stage, distinct from those in the maintenance stage, who had been at that level of intake for more than 6 months. Those falling short of recommended intake were classified as being in the precontemplation stage if they had no intention of increasing intake over the next 6 months, as being in the contemplation stage if they intended to increase intake in the next 6 months, and as being in the preparation stage if they planned to increase intake over the next 30 days and were consuming at least two servings daily. Two servings per day correspond to the median fruit and vegetable consumption.7,8 Intentional criteria and timeframes used in stage definitions were the same as those used by Prochaska et al.10
Journal of Nutrition Education Volume 33 Number 5
In a substudy of 101 subjects undertaken 3 months after the original survey, the performance of the algorithm relative to three 24-hour dietary recalls was examined. From preaction to post-action, there was a significant increase in the mean number of servings of fruits and vegetables, with those classified as being in the action and maintenance stages having mean intakes that met the serving recommendation.36 Decisional balance scale. Drawing on previous research and focus group interviews, items were written to represent the eight categories of Janis and Mann.9 Three focus group sessions were conducted to elicit perceived benefits and barriers to eating more fruits and vegetables. Discussions were tape recorded and transcribed.Thirty-five items were generated with approximately equal numbers of pros and cons. These items were pretested for clarity, relevance, and understanding in focus groups involving another 30 Singaporean subjects. Several potentially ambiguous items were excluded and reworded. A comparison of two different 5-point response formats suggested that “important-not important” ratings were more likely than “agree-disagree” ratings to reflect value evaluations rather than knowledge. Two nutrition researchers at the University of Guelph (Ontario, Canada) familiar with the decisional balance construct sorted the final 29 items according to the eight decisional balance categories. Only items for which sorting was congruent were retained.The final 28 items specified clear outcomes associated with performing the behavior change. Subjects were asked to rate on a 5-point scale from extremely important (1) to not important (5) the importance of each item in their decision to (or not to) eat more fruits or vegetables. Statistical analyses. Exploratory analysis. The data were randomly split into halves.An initial principal-components analysis with varimax rotation was performed on one sample.This was considered appropriate because the components were assumed to be minimally related as suggested by theory9 and previous research.11,12,16 Both the scree plot and eigenvalues were employed to determine the number of components to retain. Items were considered to represent a component if they achieved loadings of 0.4 or greater and did not load on another component.37 Individual components were assessed for internal consistency using item-scale correlations and alpha coefficients. Confirmatory analysis. The component structure was crossvalidated in the second sample using principal-components analysis followed by structural modeling procedures. As item scores were ordinal, the Weighted Least Squares method of the Lisrel 8.0 program38 was used to assess data fit for four nested models. Given that there is no one best index of fit for evaluating structural equation models,39 five different indices were examined.The chi-square statistic, root mean square residual (RMR), and Goodness of Fit Index (GFI) are absolute measures of fit.The RMR is a measure of the average residuals that
September • October 2001
259
result from fitting the model-predicted correlations to the observed correlations. Hence, small values (less than 0.1) are desirable.The chi-square difference tests measure the change in chi square relative to degrees of freedom and provide a useful means of comparison between the models. The NonNormed Fit Index (NNFI) and the Comparative Fit Index (CFI) are relative fit indices that measure fit relative to a null model.The GFI, NNFI, and CFI can range from 0 to 1.0, with a minimum value of 0.90 indicating acceptable model fit. Relationships with demographics/dietary practices/general beliefs. Summated pro and con scores were calculated such that higher scores indicated greater importance. Using Mann-Whitney U and Kruskal Wallis tests, the skewed pro and con scores were examined for their associations with demographic, dietary, and general belief variables. Assessing construct validity. Pro and con scores were converted into T-scores (mean = 50, SD = 10). Due to the small numbers in action (n = 4, all women), these were combined with those in maintenance to form one post-action (A/M) group.Thus, the following analyses were based on four stage groupings. A multivariate analysis of covariance (MANCOVA) was performed using a combination of pro and con scores as the dependent variable and stage as the independent variable. MANCOVA allows testing for correlations among the composite dependent variables that may go undetected by examining each dependent variable separately in univariate tests.When cell sizes vary across stage groupings, as in the present study, an assumption of particular importance is the equivalence of variances of the dependent measures between groups. Results of the Box’s M test suggested that the variance-covariance matrices of the collective dependent variables were not different (p = .38) across cells. Thus, use of MANCOVA is appropriate. Following a significant combined difference across stage groupings, separate ANCOVAs were performed for each dependent variable. Like the Box test, the univariate Levene tests indicated nonsignificant heterogeneity of variances (p = .06 for pros, p = .10 for cons). Significant univariate effects were further evaluated using the Least Significant Difference post hoc test. The covariates used in the MANCOVA and ANCOVA were identified from the bivariate analyses in the previous section, followed by multiple regression analyses to allow assessment of the joint and individual effects of the covariates. To determine the proportion of variance in the pro and con ratings that were attributable to stage groupings, effect size was calculated separately for men and women as d2 η2 = d2 + 4 where d is the difference between two means, expressed in standard deviation units.40 All analyses were performed using the SPSS 8.0 program.41
260
Ling and Horwath/PERCEIVED BENEFITS AND BARRIERS OF INCREASED FRUIT AND VEGETABLE CONSUMPTION
RESULTS Subjects. The mail survey response rate was 70.8% (21 nondeliveries due to change or incorrect address; 835 questionnaires returned).Thirty questionnaires had more than a page of questions unanswered and were discarded. Questions with missing or ambiguous responses were clarified in telephone interviews. Nine respondents were excluded as they exceeded the specified age limit. Of the final sample of 796 (49% men, 51% women; mean age 39.3 years, range 21–60; SD = 9.5), the majority had secondary, primary, or no education (60.9%) and lived in four to five-room governmentsubsidized flats (51.0%). Of those who responded, interviewers were able to assess by telephone 716 (49% men; 51% women) for stage of readiness to meet fruit and vegetable recommendations. There were no demographic differences between those interviewed and those who could not be contacted by telephone. Scale development and validation. Factor structure. Exploratory principal-components analysis revealed that a two-component solution best represented the data: gains/approval relating to self and losses/disapproval relating to self and others. After excluding six items, the remaining items were subjected to another round of principal-components analysis. The first component (13 items; mean loading = 0.66) represented the benefits of eating more fruits and vegetables (labeled “pros”).The second component (labeled “cons”) consisted of nine items describing barriers to eating more fruits and vegetables (mean loading = 0.64) (Table 1). Both components had good internal consistency of 0.89 and 0.83, respectively, and accounted for 45% of total item variance. Cross-validation and scale refinement. Repeating the principal-components analysis on the 22 items in the validation sample replicated the results observed with the developmental sample, with item loadings ranging from 0.49 to 0.77 (mean = 0.62) (see Table 1). To achieve parsimony and balance, the scales were reduced to seven items per component by maximizing breadth of construct coverage and minimizing item content overlap. The final pro and con scales demonstrated good internal consistency (0.86 and 0.79, respectively) (Table 2). Factor loadings ranged from 0.62 to 0.79 (mean = 0.69), with pros accounting for 31% and cons 19% of variance in the solution.The two scales were weakly correlated (r = .22, p < .001). Subscale relationships. Comparison of fit indices across the four nested models suggested that the two-factor correlated model best described the relationships in the data (p < .0001) (Table 3). Standardized loadings of scale items ranged from 0.62 to 0.81 (mean = 0.74), with all parameter estimates significant (p < .01).The square multiple cor-
relations for the items ranged from 0.38 to 0.69 (mean = 0.55), indicating that a good proportion of variance in the items was accounted for by the two-factor correlated model. Decisional balance and demographics, dietary practices, beliefs. Pro scores positively correlated (Spearman correlation = 0.24, p < .01) and con scores negatively correlated (Spearman correlation = –0.10, p < .01) with daily servings of fruits and vegetables. Furthermore, although cons did not differ significantly, pros for eating more fruits and vegetables were higher among women (p < .05), older adults (p < .0001) and fiberproduct users (i.e., those who took wheat bran, fiber products, or whole grains) (p < .0001). Similarly, for certain belief statements, those who expressed agreement (e.g., would rather drink juices than eat more fruit [p < .01], can do without fruit or vegetables on some days [p < .0001]) or disagreement (e.g., have made it a habit to eat plenty of fruits and vegetables since childhood [p < .0001]), rated the pros as less important but showed no difference in con ratings.These observations were empirically meaningful, providing evidence of construct validity for the scales.42 Decisional balance and stages of change. Of all of the variables considered (stage classification, demographic, and dietary variables), only stage emerged in multiple regression analyses as significantly associated with con ratings. Several variables, however, emerged as significant predictors of pro ratings, including age, fiber-product use, opinion of whether fruits and vegetables are daily essentials, and whether the habit of eating plenty of fruits and vegetables has been cultivated since childhood.Thus, these variables were included as covariates in MANCOVA. The proportions of explained variance in the pro and con ratings attributable to the stage variable were 18% (p < .001) and 9% (p < .05), respectively. A significant main effect was found for the composite pro and con scores across stage groupings (Wilk’s Λ = 0.96, F = 4.80, df = 6, 1414, p < .0001). Follow-up ANCOVAs revealed significant differences across stage groupings for both pro (F = 5.64, df = 3, 708, p < .001) and con scores (F = 2.73, df = 3, 708, p < .05). Subjects in preparation reported significantly higher pros than those in the precontemplation (p < .0001) and contemplation (p < .01) stages (Fig. 1). For cons, mean ratings were significantly lower among subjects in post-action than among those in the precontemplation (p < .01), contemplation (p < .05), and preparation (p < .05) stages. Decisional balance and gender. When performed separately for each gender, principal-components analysis and confirmatory factor analysis by structural modeling led to replication of the two-component structure. All 14 items achieved loadings greater than 0.55 on the appropriate component. Pro and con scores for men and women varied in a similar pattern across stage groupings (data not shown).
Journal of Nutrition Education Volume 33 Number 5 Table 1.
September • October 2001
Decisional considerations for eating more fruits and vegetables: subscales, scale items, and factor loadings.
Subscale and Items
Loading
Loading
Loading
Content
(Exploratory
(Confirmatory
(After Scale
Category
Sample)
Sample)
Refinement)
Pros 1 Eating more fruits or vegetables would give me more vitamins and minerals
Gain, self
0.69
0.75
0.75
Gain, self
0.68
0.72a
—
Approval, others
0.62
0.60a
—
Gain, self
0.56
0.57a
—
Approval, self
0.73
0.72
0.77
Approval, others
0.65
0.63
0.64
Gain, self
0.59
0.60a
—
Gain, others
0.58
0.58a
—
Approval, others
0.61
0.66
0.68
Gain, self
0.74
0.75
0.78
Approval, self
0.75
0.77
0.79
Gain, self
0.61
0.67a
—
Gain, self
0.72
0.68
0.70
Loss, self
0.60
0.63
0.65
Loss, self
0.54
0.49a
—
Disapproval, self
0.68
0.56
0.57
Loss, self
0.56
0.59
0.62
Disapproval, others
0.68
0.68a
—
Loss, self
0.69
0.64
0.66
Loss, others
0.69
0.75
0.77
Loss, self
0.59
0.64
0.66
Disapproval, others
0.73
0.68
0.67
2 Eating more fruits or vegetables would help to “cleanse” my body 4 I would be following the advice of my doctor or nurse if I ate more fruits or vegetables 7 Eating more fruits or vegetables would help me cut down calories 9 I would feel good about looking after my health by eating more fruits or vegetables 10 My family would be pleased if I ate more fruits or vegetables 11 I enjoy the taste of fruit 12 Other family members would worry less if I looked after my health by eating more fruits or vegetables 13 I would be following the advice of the government’s National Healthy Lifestyle Campaign if I ate more fruits or vegetables 15 Eating more fruits or vegetables every day would help to keep me regular (avoid constipation) 16 I would feel good about eating more fruits as they are fresh 19 Eating more fruits or vegetables would mean that I’m less likely to get cancer 24 Eating more fruits or vegetables would help me look better Cons 17 Eating more fruits would be expensive 18 Ordering more fruits or vegetables when I eat out wouldn’t give me value for money 20 I would feel I was overeating if I ate more fruits 21 I would worry about pesticides if I ate more fruits or vegetables 22 My family would think I was fussy if I ate more vegetables instead of other foods such as meat or seafood 25 Preparing and cooking vegetables would be time consuming 26 It would make meal planning difficult for my family if I asked for more vegetable dishes 27 I would get a bad reaction (e.g., cough, wind, cramps, etc.) if I ate more of certain fruits or vegetables 28 Others would think I was fussy if I kept worrying about having more fruits or vegetables a
261
Items that were excluded in the process of shortening the pro and con scales.
262
Ling and Horwath/PERCEIVED BENEFITS AND BARRIERS OF INCREASED FRUIT AND VEGETABLE CONSUMPTION
Table 2.
Summary statistics for the pro and con decisional balance scales.
Scale
No. of Items
Mean Score a
SD
Coefficient Alpha
Skewness
Spearman Correlation b
7
0.22**
Pros 24.89
6.16
0.86
–0.43
Men (n = 390)
Total sample
24.40
6.46
0.87
–0.40
Women (n = 406)
25.37
5.82
0.83
–0.42
1.18
Cons Total sample
a
12.35
4.88
0.79
Men (n = 390)
7
12.59
5.03
0.79
1.20
Women (n = 406)
12.13
4.72
0.79
1.15
The decisional balance scale consisted of seven pro and seven con items scored from 1 (not important) to 5 (extremely important). For the pro scale,
scores ranged from 7 to 35 and for the con scale from 7 to 33. b
Spearman correlation was computed as the distributions were skewed.
**p < .01.
DISCUSSION
efits of eating more fruits and vegetables than on reducing the barriers.10 Results from a number of studies15–18 suggest that intervention should target increasing the pros to get people to think about change, followed by decreasing the cons to allow the behavior to change. In this study, the cons remained relatively constant, but pro ratings became increasingly positive from precontemplation to preparation. Among those meeting action criteria, both the pros and the cons appeared to recede in importance. Due to combination of the action and maintenance stages, it is not possible to determine where the reduced saliency occurs. Decisional factors may be less relevant for individuals in the later stages as they have already decided that there are more pros than cons to performing the change.44 Less positive attitudes have also been reported among those in maintenance for eating more fruits and vegetables45 and for reducing dietary fat.21 It is possible that people in maintenance already have a relatively high intake and are well aware of the benefits of fruits and vegetables but
The 14-item scale developed in this study consisted of two orthogonal scales representing the pros and cons for increasing fruit and vegetable consumption.This is consistent with application of the decisional balance construct to other health-11,12,13,17 and diet-related behaviors.15,16 The two-factor structure reliably emerged across two split-half samples and gender and demonstrated good internal consistency. Evidence of construct validity was provided by replication of the hypothesized crossover pattern from cons to pros being more important across the stages. The larger change in pros across the stages (a medium effect size)40 relative to the change in cons (a small effect size)40 supports Prochaska’s43 contention of the strong and weak principles of behavior change. Although confirmation is required in prospective studies, this suggests that in order to stimulate movement from the early stages, a greater emphasis may be needed on raising perceptions of the ben-
Table 3.
Comparison of model fita among four nested models (n = 398). χ2
df
6847
91
398
latent variables) Two factor correlated (allows for correlated latent variables)
Model Null (assumes all manifest variables independent)
RMR
GFI
NNFI
CFI
77
0.23
0.94
0.94
0.95
276
77
0.12
0.96
0.97
0.97
236
76
0.09
0.97
0.97
0.98
One factor (assumes all manifest variables loaded on one latent variable) Two factor uncorrelated (assumes no correlation among two
a
Fit indices are based on the Weighted Least Squares method. The relatively high values may be a result of the estimation method used. Compari-
son of indices, however, should be relative across models rather than absolute. The chi-square difference test between the two-factor correlated model and the next alternative model, the uncorrelated model, gives ∆χ2 = 40.38, ∆df = 1 (p < .001). The other indices (RMR, GFI, NNFI, and CFI) consistently perform best for the two-factor correlated model when compared with the two-factor uncorrelated and one-factor models. RMR = root mean square residual, GFI = Goodness of Fit Index, NNFI = Non-Normed Fit Index, CFI = Comparative Fit Index.
Journal of Nutrition Education Volume 33 Number 5
Figure 1. Pro and con scores by stages of change. Scores were presented as standardized T-scores (mean = 50, SD = 10). Pros: P > PC (p < .0001); P > C (p < .01). Cons: PC > A/M (p < .01); C, P > A/M (p < .05). Plotted for four stage groupings, with action and maintenance stages combined into the post-action stage. PC (precontemplation, n = 210), C (contemplation, n = 234), P (preparation, n = 207), A/M (post-action, n = 65).
might not see any added benefit from increasing their intake any further.45 The pros and cons crossed over between contemplation and preparation. Depending on the health behaviors examined, the crossing over from the pros to the cons being more important generally occurred between the contemplation and preparation stages.15–17 Some authors have reported the crossover to occur between the preparation and action stages.18,46 These findings suggest that people would need to perceive higher pros than cons prior to taking action.17 Previous work12,15,16 has also demonstrated that the eight decision-making considerations were not equally salient and that the final item pool reflected the nature of the behavior studied. In this study, the pro items represented self-approval, instrumental gains for self, and approval from others. None of the items relating to instrumental gains for others were included by the selection process. Loadings for the pro items were also weighted toward the “self ” items, suggesting that social influence may be less salient in determining perceived benefits to eat more fruits and vegetables. On the other hand, items representing perceived social barriers loaded high on the con scale. In Singapore society, where eating out and sharing of communal dishes are common, it is likely that fussing over having more fruits and vegetables or requesting extra vegetable dishes would inconvenience others. It is of interest to note that considerations of taste, price, and inconvenience did not emerge as particularly salient in driving consumption decisions. In the 5 A Day baseline survey, most respondents reported that they liked the taste of fruits (81.7%) and vegetables (71.4%).5 In a nationwide U.K. survey,47 taste, healthfulness, nutrition, family liking, and fit-
September • October 2001
263
ting into eating habits emerged as the five most important beliefs for eating more fruits and vegetables. However, in a recent review of environmental and personal influences on dietary behaviors for chronic disease prevention,48 it was noted that widely held perceptions such as taste or time constraints are not necessarily predictive of dietary behaviors. Consumer research conducted for the 5 A Day for Better Health campaign27 also found that taste, price, and convenience are both benefits and barriers. Although focus group participants mentioned (good) prices, taste, and convenience as benefits derived from buying and eating fruits and vegetables, complaints about texture, flavor, availability, and preparation time were also common reasons given for not eating more of these foods. Similarly, a recent study examining cancer-related nutrition beliefs and attitudes49 revealed that Americans may believe that healthful foods taste good, but these were not necessarily the best-tasting foods. Although necessary, taste may not be sufficient to motivate dietary change. Importantly, decisional balance items represented immediate gains or losses rather than long-term outcomes.This is consistent with the findings of Balch et al.,27 whose focus group participants considered cancer a remote concern beyond their control and rarely taken into account when making food choices. The present study achieved a fairly good response rate of 71%, which is close to the reported average of 73% for general public surveys that used similar protocols.33 Although practically every household in Singapore has a telephone, the sampling approach excludes the 14.5% of the residential telephone numbers not listed.50 However, housing type (indicative of socioeconomic status in Singapore) and educational level characteristics of the study sample were comparable with national averages data.51 The administration of study instruments was also scheduled to reduce the likelihood of response bias.The 3-month delay between the mail survey and the 24-hour dietary interviews employed to validate the staging algorithms reduced the likelihood that prior completion of the food frequency and decisional balance questions influenced subjects’ dietary recalls. Moreover, subjects were unaware that they would be asked to recall the specific day’s diet, and the interviews assessed all food and drinks, not just fruits and vegetables.
IMPLICATIONS FOR RESEARCH AND PRACTICE To effect change, we will probably need to apply both individual change processes and public health policies.43 The results of this study suggest that to raise fruit and vegetable consumption, it may be most effective to target those in the early stages of change with messages about more immediate benefits directly relevant to the self. For those in the later stages, public health policies (e.g., promoting availability of fruits and vegetables at subsidized prices) may help to reduce
264
Ling and Horwath/PERCEIVED BENEFITS AND BARRIERS OF INCREASED FRUIT AND VEGETABLE CONSUMPTION
barriers and allow these individuals to take action and maintain the behavior change. The robustness of the decisional balance and stage constructs in a non-white population, rarely investigated elsewhere,22 is encouraging. Further research is needed to determine the longitudinal shift in pros and cons as people progress through the stages and the influence of intervention trials on decisional balance. Further testing of the scale is also needed to examine its generalizability to an ethnically diverse population.
ACKNOWLEDGMENTS This work was undertaken as part of the Ph.D. dissertation of the first author. Ethical issues related to the study were considered as part of the grant-approving process by the National Medical Research Council, Singapore.The authors are grateful to the Department of Nutrition, Singapore, for support and assistance with survey implementation; Professor Rob Lawson and Mrs. Sheila Williams at the University of Otago for statistical advice; Dr. Pauline Gulliver for assistance with data analysis; and Drs. Stephanie Ounpuu and Judy Sheeshka for feedback on the scales.
REFERENCES 1. World Health Organization. Diet, nutrition and chronic diseases: report of a WHO study group. Technical Report Series 797. Geneva: WHO, 1990. 2. Steinmetz KA, Potter JD.Vegetables, fruit and cancer. I. Epidemiology. Cancer Causes Control 1991;2:825–57. 3. Steinmetz KA, Potter JD.Vegetables, fruit and cancer. II. Mechanisms. Cancer Causes Control. 1991;2:427–41. 4. Block G, Patterson B, Subar A. Fruit, vegetables and cancer prevention: a review of the epidemiological evidence. Nutr Cancer 1992;18:1–29. 5. Krebs-Smith SM, Cook A, Subar AF, Cleveland L, Friday J. US adults’ fruit and vegetable intakes, 1989 to 1991: a revised baseline for the Healthy People 2000 Objective. Am J Public Health 1995;85:1623–9. 6. Ministry of Agriculture, Fisheries and Food. National food survey, 1995. London: HMSO, 1996. 7. Ministry of Health, Singapore, Food and Nutrition Department. Fruit and vegetable consumption—survey of shoppers. Singapore: Ministry of Health, 1990. 8. Ling AMC, Horwath C, Parnell W. Validation of a short food frequency questionnaire to assess consumption of cereal foods, fruit and vegetables in Chinese Singaporeans. Eur J Clin Nutr 1998;52:557–64. 9. Janis IL, Mann L. Decision-making: a psychological analysis of conflict, choice and commitment. New York: Collier MacMillan, 1977. 10. Prochaska JO, Redding C, Evers K. The Transtheoretical Model of behavior change. Health behaviour and health education: theory, research and practice. 2nd Ed. San Francisco: Jossey-Bass, 1997:60–84. 11. Velicer WF, DiClemente CC, Prochaska JO, Brandenburg N. Decisional balance measure for assessing and predicting smoking status. J Pers Soc Psychol 1985;48:1279–89.
12. O’Connell D,Velicer WF. A decisional balance measure and the stages of change model for weight loss. Int J Addict 1988;23:729–50. 13. Marcus BH, Rakowski W, Rossi JS.Assessing motivational readiness and decision making for exercise. Health Psychol 1992;11:257–61. 14. Steptoe A,Wijetunge S, Doherty S,Wardle J. Stages of change for dietary fat reduction: associations with food intake, decisional balance and motives for food choice. Health Educ J 1996;55:108–22. 15. Ounpuu S.The pros and cons of reducing fat as intermediate outcome variables in the Transtheoretical Model: validation of a measurement model. Unpublished Ph.D. dissertation, University of Guelph, Guelph, Ontario, Canada, 1997. 16. Gulliver P, Horwath CC. Decisional balance and stage of change for dairy consumption.Abstract presented at the 20th Annual Convention of the Society for Behavioral Medicine, San Diego, CA, March 1999. 17. Prochaska JO,Velicer WF, Rossi JS, et al. Stages of change and decisional balance for 12 problem behaviors. Health Psychol 1994;13:39–46. 18. Herrick AB, Stone WJ, Mettler MM. Stages of change, decisional balance and self-efficacy across four health behaviors in a worksite environment. Am J Health Promotion 1997;12:49–56. 19. Prochaska JO, DiClemente CC,Velicer WF, Ginpil S, Norcross JC. Predicting change in smoking status for self-changers. Addict Behav 1985;10:395–406. 20. Rossi SR, Greene GW, Reed G, Rossi JS, Prochaska JO, Velicer WF. Cross-validation of a decisional balance measure for dietary fat reduction. Abstract presented at the 14th Annual Convention of the Society for Behavioral Medicine, Boston MA, April 1994. 21. Brug J, Hospers HJ, Kok G. Differences in psychosocial factors and fat consumption between stages of change for fat reduction. Psychol Health 1997;12:719–27. 22. Campbell MK, Symons M, Demark-Wahnefried W. Stages of change and psychosocial correlates of fruit and vegetable consumption among rural African-American church members. Am J Health Promotion 1998;12:185–91. 23. Reicks M, Randall JL, Haynes BJ. Factors affecting consumption of fruits and vegetables by low-income families. J Am Diet Assoc 1994;94:1309–11. 24. Brug J, Debie S, van Assema P,Weijts W. Psychosocial determinants of fruit and vegetable consumption among adults: results of focus group interviews. Food Qual Pref 1995;6:99–107. 25. Treiman K, Fremuth V, Darmon D, et al.Attitudes and behaviors related to fruits and vegetables among low-income women in the WIC programme. J Nutr Educ 1996;28:149–56. 26. Loughrey KA, Balch GI, Lefebvre RC, et al. Bringing 5-a-Day consumers into focus: qualitative use of consumer research to guide strategic decision making. J Nutr Educ 1997;29:172–7. 27. Balch GI, Loughrey KA,Weinberg L, Lurie D, Eisner E. Probing consumer benefits and barriers for the national 5-a-Day campaign focus group findings. J Nutr Educ 1997;29:178–83. 28. Dittus KL, Hillers VN, Beerman KA. Benefits and barriers to fruit and vegetable intake: relationship between attitudes and consumption. J Nutr Educ 1995;27:120–6. 29. Krebs-Smith AM, Heimendinger J, Patterson BH, Subar AF, Kessler R, Pivonka E. Psychosocial factors associated with fruit and vegetable consumption. Am J Health Promotion 1995;10:98–104. 30. Brug J, Lechner L, deVries H. Psychosocial determinants of fruit and vegetable consumption. Appetite 1995;25:285–96.
Journal of Nutrition Education Volume 33 Number 5 31. Lappalainen R, Saba A, Holm A, Mykkanen H, Gibney MJ. Difficulties in trying to eat healthier: descriptive analysis of perceived barriers for healthy eating. Eur J Clin Nutr 1997;51(Suppl 2):S36–40. 32. Zunft HJF, Friebe D, Seppelt B, et al. Perceived benefits of healthy eating among a nationally-representative sample of adults in the European Union. Eur J Clin Nutr 1997;51(Suppl 2):S41–6.
September • October 2001
265
42. DeVellis RF. Scale development: theory and applications. Newbury Park, CA: Sage, 1991. 43. Prochaska JO. Strong and weak principles for progressing from precontemplation to action on the basis of twelve problem behaviors. Health Psychol 1994;13:47–51.
33. Dillman DA. Mail and telephone surveys: the total design method. Toronto: John Wiley & Sons, 1978.
44. Prochaska JO, DiClemente CC,Velicer WF, Ginpil S, Norcross JC. Predicting change in smoking status for self-changers. Addict Behav 1985;10:395–406.
34. Ling AM, Horwath CC. Self-efficacy and consumption of fruit and vegetables—validation of a summated scale. Am J Health Promotion 1999;13:290–8.
45. Brug J, Glanz K, Kok G. The relationship between self-efficacy, attitudes, intake compared to others, consumption and stages of change related to fruit and vegetables.Am J Health Promotion 1997;12:25–30.
35. Ministry of Health, Singapore,Training and Health Education Department. Guidelines for a healthy diet—recommendations of the National Advisory Committee on Food and Nutrition. Singapore: Ministry of Health, 1989. 36. Ling AM, Horwath CC. Defining and measuring stages of change for dietary behaviors: readiness to meet fruit, vegetable and grain guidelines among Chinese Singaporeans. J Am Diet Assoc 2000;100: 898–904. 37. Hair JF,Anderson RE,Tatham RL, Black WC. Multivariate data analysis with readings. Englewood Cliffs, NJ: Prentice-Hall, 1995. 38. Jöreskog K, Sörbom D. Lisrel 8: user’s reference guide. Chicago: Scientific Software International, 1996. 39. Hu LT, Bentler PM. Evaluating model fit. Structural equation modeling: concepts, issues and applications. Newbury Park, CA: Sage, 1995:76–99. 40. Cohen J, Cohen P.Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum, 1983. 41. SPSS 8.0 for Windows. Chicago: SPSS, 1998.
46. Marcus BH, Owen N. Motivational readiness, self-efficacy and decision-making for exercise. J Appl Soc Psychol 1992;22:3–16. 47. Cox DN, Anderson AS, McKellar S, Reynolds J, Lean MEJ, Mela DJ. Vegetables and fruits: barriers and opportunities for greater consumption. Nutr Food Sci 1996;5:44–7. 48. Harnack L, Block G, Lane S. Influence of selected environmental and personal factors on dietary behavior for chronic disease prevention: a review of the literature. J Nutr Educ 1997;29:306–12. 49. Harnack L, Block G, Subar A, Lane S. Cancer prevention-related nutrition knowledge, beliefs and attitudes of U.S. adults: 1992 NHIS cancer epidemiology supplement. J Nutr Educ 1998;30:131–8. 50. Singapore Telecom.The Singapore phone book, residential listing. July 1997. 51. Department of Statistics. General household survey report. Statistical Release I. Singapore: Ministry of Health, 1995.
Corrections in: Sobal J. Sample Extensiveness in Qualitative Nutrition Education Research. J Nutr Educ 2001;33:184–192. On page 184, under “Results”, the abstract should have stated “Ten studies used observation/fieldwork, and eight used other types of qualitative research, with mixed patterns of sample extensiveness in those studies.” On page 185, 2nd paragraph, it should have stated “In contrast to quantitative research, qualitative research is grounded in nonpositivist paradigms that include interpretivist, constructionist, constructivist, phenomenological, post-modernist, and other perspectives that focus on gathering and analyzing non-numeric data in the form of text, images, sounds, and objects.” On page 189, the last paragraph under “Discussion” should have stated “The author is also examining sample extensiveness in other journals in other fields.”
Perceived Benefits and Barriers of Increased Fruit and Vegetable Consumption: Validation of a Decisional Balance Scale A N N I E M E I C H UA N L I N G 1 A N D C A RO L I N E H O RWAT H 2 1
2
Department of Nutrition, Level 5, Institute of Health, Singapore 168937; Department of Human Nutrition, University of Otago, Dunedin, New Zealand
ABSTRACT Objective: To develop and validate scales to assess perceived benefits and barriers (decisional balance) for increasing fruit and vegetable consumption. Design:
A cross-sectional mail and telephone survey was conducted.
Subjects/Settings: A total of 1200 Chinese households were randomly selected from the Singapore residential telephone listings, and 71% responded to the mail survey; 390 males and 406 females participated (mean age = 39.3). Main Outcome Measures:
Decisional balance, stage of change, and fruit and vegetable consumption were measured.
Statistical Analyses Performed: Using a split-half sample approach, developmental sample responses were analyzed by principal-components analysis and validation sample responses by confirmatory factor analysis.Analyses of variance were used to examine stage differences in decisional balance. Results: Principal-components analysis indicated two components representing benefits (or pros) (Cronbach’s α = 0.86) and barriers (or cons) (α 0.79) of change. Confirmatory factor analysis strongly supported the two-component structure (Goodness of Fit Index = 0.97).There was a shift from cons to pros being more important across the stages.The increase in pros across the stages of change (p < .0001) corresponded to a medium effect size, and the decrease in cons (p < .01) corresponded to a small effect size. Implications:
Decisional balance scales may be used to guide interventions to influence fruit and vegetable consumption.
KEY WORDS:
benefits, barriers, stage of change, fruit, vegetables (JNE 33:257–265, 2001)
INTRODUCTION
Relatively little, however, is known about what might motivate people to eat more fruits and vegetables. Understanding how motivational considerations can influence the process of changing fruit and vegetable consumption may be enhanced by applying the decisional balance construct of Janis and Mann’s Decision-Making Model.9 Decision making was regarded as a comparative process involving balancing four categories of gains (instrumental gains for self and others and approval for self and others) against four categories of costs (instrumental costs for self and others and disapproval from self and others). Transtheoretical Model (TTM) researchers have used the decisional balance construct to examine how motivational considerations are related to stage of readiness to change.10 According to the model, behavior change progresses through
The dietary guidelines of many countries1 encourage fruit and vegetable consumption since there is compelling evidence that higher intakes are associated with reduced cancer risk.2–4 In Singapore, as in the United States5 and Britain,6 fruit and vegetable consumption falls well below recommended levels.7,8 ................................................... This work was sponsored by the National Medical Research Council, Singapore and the International Life Sciences Institute, South East Asia. Address for correspondence: Annie Ling, B.Sc., M.Sc., Ph.D., Department of Nutrition, Level 5, Institute of Health, 3 Second Hospital Avenue, Singapore 168937; Fax: (65) 438-3605; E-mail: [email protected]. ©2001 SOCIETY FOR NUTRITION EDUCATION
257
258
Ling and Horwath/PERCEIVED BENEFITS AND BARRIERS OF INCREASED FRUIT AND VEGETABLE CONSUMPTION
five stages. During the early stages of precontemplation and contemplation, change is conceptualized as cognitive (moving from no intention of changing to considering it).Those in the preparation stage plan to change in the near future, those in the action stage have recently adopted the new behavior, and those in the maintenance stage are working at sustaining the change. Research across diverse health behaviors has demonstrated two orthogonal factors representing the perceived benefits (pros) and barriers (cons) of change11–16 rather than the eight-factor structure hypothesized by Janis and Mann.9 These and other studies17,18 have revealed cons to outweigh pros for precontemplators, whereas pros outweigh cons for those in the action and maintenance stages. Longitudinal studies of smoking cessation have demonstrated decisional balance to be a strong predictor of movement, especially through the early stages of change.19 Dietary applications of the decisional balance construct within a TTM framework have focused on dietary fat reduction.14,15,18,20,21 This is one of the first studies to examine decisional balance in relation to stage of readiness to increase fruit and vegetable consumption. In a study of African Americans, Campbell et al.22 reported that barriers to change were more prevalent among precontemplators than those contemplating or planning change. Perceived benefit, assessed using one item, was generally high across all stages. Qualitative research using focus groups has suggested the main perceived benefits of a higher fruit and vegetable intake to be nutritional value, a sense of well-being (e.g., digestive regularity, feeling light/refreshed, and doing some good for oneself), health, and weight loss.23–27 Common barriers to greater intake appeared to be cost in time and money, effort to prepare, difficulty in changing habits, preference for other foods, lack of availability, concern for pesticides, and lack of awareness of what constitutes sufficient amounts to be consumed. Quantitative research examining perceived benefits of fruit and vegetable consumption has involved measures limited in coverage of the construct domain mainly to items concerning nutrient value and health benefits.22,28,29 Scales have often involved general belief statements (e.g.,“fruits and vegetables are good sources of fiber”),28 rather than personalized items with clear statement of the goal behavior (e.g.,“eating more fruits and vegetables would help me avoid constipation”), and were more likely to tap knowledge than value appraisal of outcomes.Although a few studies have reported low (0.48) to moderate (0.70) alpha coefficients for the benefit and barrier scales,28,30 none has established the factor structure of decisional considerations for fruit and vegetable consumption. Nevertheless, in one study, barriers accounted for the greatest variability in consumption28; another found those in the highest intake tertile to have more positive attitudes,30 and yet another failed to find an association between beliefs about the healthfulness of fruits and vegetables and intake levels.29 This study aims to fill a gap in research by developing a scale to measure the pros and cons of decision making for eating more fruits and vegetables. Using a split-sample technique,
developmental sample responses were analyzed by principalcomponents analysis to determine an initial dimensional structure, and validation sample responses were analyzed by structural modeling procedures to confirm the structure.The scales were also evaluated for their ability to differentiate individuals at varying stages of readiness to change. As nutrition beliefs and attitudes have been reported to vary among men and women,31,32 gender differences were also examined.
METHODS Sample and study design. The Singapore residential telephone listings were used to randomly select 1200 Chinese households. Each household was mailed both Chinese and English versions of a self-administered questionnaire with a cover letter and postage-paid return envelope. Up to three reminders were sent to nonrespondents at 10 days, 3 weeks, and 5 weeks after the initial mailing,33 including replacement questionnaires with the second reminder. Cover letters were randomly assigned to requesting participation by either a male or female adult aged 21 to 60 years. If no one of the specified age and gender was present in the household, a person of the opposite gender was invited to participate.To encourage participation, $10 telephone cards were promised on return of the completed questionnaire. For those who responded, a short telephone interview followed to assess stage of change. Each household was called up to seven times. Measures. As part of a larger study investigating factors influencing fruit, vegetable, and grain consumption,34 the questionnaire included a validated fruit and vegetable food frequency questionnaire (FFQ),8 a decisional balance scale, and questions on demographics, supplement use, general beliefs, meal venues, and health conditions. Stage of change algorithm. Based on the combined fruit and vegetable estimates from the FFQ, subjects were asked the length of time they had been at their current level of intake (for those having adequate numbers of servings) or whether they intended to increase intake to meet the Singapore dietary guideline of four servings per day (excluding juices).35 Subjects consuming adequate servings but for less than 6 months were classified as being in the action stage, distinct from those in the maintenance stage, who had been at that level of intake for more than 6 months. Those falling short of recommended intake were classified as being in the precontemplation stage if they had no intention of increasing intake over the next 6 months, as being in the contemplation stage if they intended to increase intake in the next 6 months, and as being in the preparation stage if they planned to increase intake over the next 30 days and were consuming at least two servings daily. Two servings per day correspond to the median fruit and vegetable consumption.7,8 Intentional criteria and timeframes used in stage definitions were the same as those used by Prochaska et al.10
Journal of Nutrition Education Volume 33 Number 5
In a substudy of 101 subjects undertaken 3 months after the original survey, the performance of the algorithm relative to three 24-hour dietary recalls was examined. From preaction to post-action, there was a significant increase in the mean number of servings of fruits and vegetables, with those classified as being in the action and maintenance stages having mean intakes that met the serving recommendation.36 Decisional balance scale. Drawing on previous research and focus group interviews, items were written to represent the eight categories of Janis and Mann.9 Three focus group sessions were conducted to elicit perceived benefits and barriers to eating more fruits and vegetables. Discussions were tape recorded and transcribed.Thirty-five items were generated with approximately equal numbers of pros and cons. These items were pretested for clarity, relevance, and understanding in focus groups involving another 30 Singaporean subjects. Several potentially ambiguous items were excluded and reworded. A comparison of two different 5-point response formats suggested that “important-not important” ratings were more likely than “agree-disagree” ratings to reflect value evaluations rather than knowledge. Two nutrition researchers at the University of Guelph (Ontario, Canada) familiar with the decisional balance construct sorted the final 29 items according to the eight decisional balance categories. Only items for which sorting was congruent were retained.The final 28 items specified clear outcomes associated with performing the behavior change. Subjects were asked to rate on a 5-point scale from extremely important (1) to not important (5) the importance of each item in their decision to (or not to) eat more fruits or vegetables. Statistical analyses. Exploratory analysis. The data were randomly split into halves.An initial principal-components analysis with varimax rotation was performed on one sample.This was considered appropriate because the components were assumed to be minimally related as suggested by theory9 and previous research.11,12,16 Both the scree plot and eigenvalues were employed to determine the number of components to retain. Items were considered to represent a component if they achieved loadings of 0.4 or greater and did not load on another component.37 Individual components were assessed for internal consistency using item-scale correlations and alpha coefficients. Confirmatory analysis. The component structure was crossvalidated in the second sample using principal-components analysis followed by structural modeling procedures. As item scores were ordinal, the Weighted Least Squares method of the Lisrel 8.0 program38 was used to assess data fit for four nested models. Given that there is no one best index of fit for evaluating structural equation models,39 five different indices were examined.The chi-square statistic, root mean square residual (RMR), and Goodness of Fit Index (GFI) are absolute measures of fit.The RMR is a measure of the average residuals that
September • October 2001
259
result from fitting the model-predicted correlations to the observed correlations. Hence, small values (less than 0.1) are desirable.The chi-square difference tests measure the change in chi square relative to degrees of freedom and provide a useful means of comparison between the models. The NonNormed Fit Index (NNFI) and the Comparative Fit Index (CFI) are relative fit indices that measure fit relative to a null model.The GFI, NNFI, and CFI can range from 0 to 1.0, with a minimum value of 0.90 indicating acceptable model fit. Relationships with demographics/dietary practices/general beliefs. Summated pro and con scores were calculated such that higher scores indicated greater importance. Using Mann-Whitney U and Kruskal Wallis tests, the skewed pro and con scores were examined for their associations with demographic, dietary, and general belief variables. Assessing construct validity. Pro and con scores were converted into T-scores (mean = 50, SD = 10). Due to the small numbers in action (n = 4, all women), these were combined with those in maintenance to form one post-action (A/M) group.Thus, the following analyses were based on four stage groupings. A multivariate analysis of covariance (MANCOVA) was performed using a combination of pro and con scores as the dependent variable and stage as the independent variable. MANCOVA allows testing for correlations among the composite dependent variables that may go undetected by examining each dependent variable separately in univariate tests.When cell sizes vary across stage groupings, as in the present study, an assumption of particular importance is the equivalence of variances of the dependent measures between groups. Results of the Box’s M test suggested that the variance-covariance matrices of the collective dependent variables were not different (p = .38) across cells. Thus, use of MANCOVA is appropriate. Following a significant combined difference across stage groupings, separate ANCOVAs were performed for each dependent variable. Like the Box test, the univariate Levene tests indicated nonsignificant heterogeneity of variances (p = .06 for pros, p = .10 for cons). Significant univariate effects were further evaluated using the Least Significant Difference post hoc test. The covariates used in the MANCOVA and ANCOVA were identified from the bivariate analyses in the previous section, followed by multiple regression analyses to allow assessment of the joint and individual effects of the covariates. To determine the proportion of variance in the pro and con ratings that were attributable to stage groupings, effect size was calculated separately for men and women as d2 η2 = d2 + 4 where d is the difference between two means, expressed in standard deviation units.40 All analyses were performed using the SPSS 8.0 program.41
260
Ling and Horwath/PERCEIVED BENEFITS AND BARRIERS OF INCREASED FRUIT AND VEGETABLE CONSUMPTION
RESULTS Subjects. The mail survey response rate was 70.8% (21 nondeliveries due to change or incorrect address; 835 questionnaires returned).Thirty questionnaires had more than a page of questions unanswered and were discarded. Questions with missing or ambiguous responses were clarified in telephone interviews. Nine respondents were excluded as they exceeded the specified age limit. Of the final sample of 796 (49% men, 51% women; mean age 39.3 years, range 21–60; SD = 9.5), the majority had secondary, primary, or no education (60.9%) and lived in four to five-room governmentsubsidized flats (51.0%). Of those who responded, interviewers were able to assess by telephone 716 (49% men; 51% women) for stage of readiness to meet fruit and vegetable recommendations. There were no demographic differences between those interviewed and those who could not be contacted by telephone. Scale development and validation. Factor structure. Exploratory principal-components analysis revealed that a two-component solution best represented the data: gains/approval relating to self and losses/disapproval relating to self and others. After excluding six items, the remaining items were subjected to another round of principal-components analysis. The first component (13 items; mean loading = 0.66) represented the benefits of eating more fruits and vegetables (labeled “pros”).The second component (labeled “cons”) consisted of nine items describing barriers to eating more fruits and vegetables (mean loading = 0.64) (Table 1). Both components had good internal consistency of 0.89 and 0.83, respectively, and accounted for 45% of total item variance. Cross-validation and scale refinement. Repeating the principal-components analysis on the 22 items in the validation sample replicated the results observed with the developmental sample, with item loadings ranging from 0.49 to 0.77 (mean = 0.62) (see Table 1). To achieve parsimony and balance, the scales were reduced to seven items per component by maximizing breadth of construct coverage and minimizing item content overlap. The final pro and con scales demonstrated good internal consistency (0.86 and 0.79, respectively) (Table 2). Factor loadings ranged from 0.62 to 0.79 (mean = 0.69), with pros accounting for 31% and cons 19% of variance in the solution.The two scales were weakly correlated (r = .22, p < .001). Subscale relationships. Comparison of fit indices across the four nested models suggested that the two-factor correlated model best described the relationships in the data (p < .0001) (Table 3). Standardized loadings of scale items ranged from 0.62 to 0.81 (mean = 0.74), with all parameter estimates significant (p < .01).The square multiple cor-
relations for the items ranged from 0.38 to 0.69 (mean = 0.55), indicating that a good proportion of variance in the items was accounted for by the two-factor correlated model. Decisional balance and demographics, dietary practices, beliefs. Pro scores positively correlated (Spearman correlation = 0.24, p < .01) and con scores negatively correlated (Spearman correlation = –0.10, p < .01) with daily servings of fruits and vegetables. Furthermore, although cons did not differ significantly, pros for eating more fruits and vegetables were higher among women (p < .05), older adults (p < .0001) and fiberproduct users (i.e., those who took wheat bran, fiber products, or whole grains) (p < .0001). Similarly, for certain belief statements, those who expressed agreement (e.g., would rather drink juices than eat more fruit [p < .01], can do without fruit or vegetables on some days [p < .0001]) or disagreement (e.g., have made it a habit to eat plenty of fruits and vegetables since childhood [p < .0001]), rated the pros as less important but showed no difference in con ratings.These observations were empirically meaningful, providing evidence of construct validity for the scales.42 Decisional balance and stages of change. Of all of the variables considered (stage classification, demographic, and dietary variables), only stage emerged in multiple regression analyses as significantly associated with con ratings. Several variables, however, emerged as significant predictors of pro ratings, including age, fiber-product use, opinion of whether fruits and vegetables are daily essentials, and whether the habit of eating plenty of fruits and vegetables has been cultivated since childhood.Thus, these variables were included as covariates in MANCOVA. The proportions of explained variance in the pro and con ratings attributable to the stage variable were 18% (p < .001) and 9% (p < .05), respectively. A significant main effect was found for the composite pro and con scores across stage groupings (Wilk’s Λ = 0.96, F = 4.80, df = 6, 1414, p < .0001). Follow-up ANCOVAs revealed significant differences across stage groupings for both pro (F = 5.64, df = 3, 708, p < .001) and con scores (F = 2.73, df = 3, 708, p < .05). Subjects in preparation reported significantly higher pros than those in the precontemplation (p < .0001) and contemplation (p < .01) stages (Fig. 1). For cons, mean ratings were significantly lower among subjects in post-action than among those in the precontemplation (p < .01), contemplation (p < .05), and preparation (p < .05) stages. Decisional balance and gender. When performed separately for each gender, principal-components analysis and confirmatory factor analysis by structural modeling led to replication of the two-component structure. All 14 items achieved loadings greater than 0.55 on the appropriate component. Pro and con scores for men and women varied in a similar pattern across stage groupings (data not shown).
Journal of Nutrition Education Volume 33 Number 5 Table 1.
September • October 2001
Decisional considerations for eating more fruits and vegetables: subscales, scale items, and factor loadings.
Subscale and Items
Loading
Loading
Loading
Content
(Exploratory
(Confirmatory
(After Scale
Category
Sample)
Sample)
Refinement)
Pros 1 Eating more fruits or vegetables would give me more vitamins and minerals
Gain, self
0.69
0.75
0.75
Gain, self
0.68
0.72a
—
Approval, others
0.62
0.60a
—
Gain, self
0.56
0.57a
—
Approval, self
0.73
0.72
0.77
Approval, others
0.65
0.63
0.64
Gain, self
0.59
0.60a
—
Gain, others
0.58
0.58a
—
Approval, others
0.61
0.66
0.68
Gain, self
0.74
0.75
0.78
Approval, self
0.75
0.77
0.79
Gain, self
0.61
0.67a
—
Gain, self
0.72
0.68
0.70
Loss, self
0.60
0.63
0.65
Loss, self
0.54
0.49a
—
Disapproval, self
0.68
0.56
0.57
Loss, self
0.56
0.59
0.62
Disapproval, others
0.68
0.68a
—
Loss, self
0.69
0.64
0.66
Loss, others
0.69
0.75
0.77
Loss, self
0.59
0.64
0.66
Disapproval, others
0.73
0.68
0.67
2 Eating more fruits or vegetables would help to “cleanse” my body 4 I would be following the advice of my doctor or nurse if I ate more fruits or vegetables 7 Eating more fruits or vegetables would help me cut down calories 9 I would feel good about looking after my health by eating more fruits or vegetables 10 My family would be pleased if I ate more fruits or vegetables 11 I enjoy the taste of fruit 12 Other family members would worry less if I looked after my health by eating more fruits or vegetables 13 I would be following the advice of the government’s National Healthy Lifestyle Campaign if I ate more fruits or vegetables 15 Eating more fruits or vegetables every day would help to keep me regular (avoid constipation) 16 I would feel good about eating more fruits as they are fresh 19 Eating more fruits or vegetables would mean that I’m less likely to get cancer 24 Eating more fruits or vegetables would help me look better Cons 17 Eating more fruits would be expensive 18 Ordering more fruits or vegetables when I eat out wouldn’t give me value for money 20 I would feel I was overeating if I ate more fruits 21 I would worry about pesticides if I ate more fruits or vegetables 22 My family would think I was fussy if I ate more vegetables instead of other foods such as meat or seafood 25 Preparing and cooking vegetables would be time consuming 26 It would make meal planning difficult for my family if I asked for more vegetable dishes 27 I would get a bad reaction (e.g., cough, wind, cramps, etc.) if I ate more of certain fruits or vegetables 28 Others would think I was fussy if I kept worrying about having more fruits or vegetables a
261
Items that were excluded in the process of shortening the pro and con scales.
262
Ling and Horwath/PERCEIVED BENEFITS AND BARRIERS OF INCREASED FRUIT AND VEGETABLE CONSUMPTION
Table 2.
Summary statistics for the pro and con decisional balance scales.
Scale
No. of Items
Mean Score a
SD
Coefficient Alpha
Skewness
Spearman Correlation b
7
0.22**
Pros 24.89
6.16
0.86
–0.43
Men (n = 390)
Total sample
24.40
6.46
0.87
–0.40
Women (n = 406)
25.37
5.82
0.83
–0.42
1.18
Cons Total sample
a
12.35
4.88
0.79
Men (n = 390)
7
12.59
5.03
0.79
1.20
Women (n = 406)
12.13
4.72
0.79
1.15
The decisional balance scale consisted of seven pro and seven con items scored from 1 (not important) to 5 (extremely important). For the pro scale,
scores ranged from 7 to 35 and for the con scale from 7 to 33. b
Spearman correlation was computed as the distributions were skewed.
**p < .01.
DISCUSSION
efits of eating more fruits and vegetables than on reducing the barriers.10 Results from a number of studies15–18 suggest that intervention should target increasing the pros to get people to think about change, followed by decreasing the cons to allow the behavior to change. In this study, the cons remained relatively constant, but pro ratings became increasingly positive from precontemplation to preparation. Among those meeting action criteria, both the pros and the cons appeared to recede in importance. Due to combination of the action and maintenance stages, it is not possible to determine where the reduced saliency occurs. Decisional factors may be less relevant for individuals in the later stages as they have already decided that there are more pros than cons to performing the change.44 Less positive attitudes have also been reported among those in maintenance for eating more fruits and vegetables45 and for reducing dietary fat.21 It is possible that people in maintenance already have a relatively high intake and are well aware of the benefits of fruits and vegetables but
The 14-item scale developed in this study consisted of two orthogonal scales representing the pros and cons for increasing fruit and vegetable consumption.This is consistent with application of the decisional balance construct to other health-11,12,13,17 and diet-related behaviors.15,16 The two-factor structure reliably emerged across two split-half samples and gender and demonstrated good internal consistency. Evidence of construct validity was provided by replication of the hypothesized crossover pattern from cons to pros being more important across the stages. The larger change in pros across the stages (a medium effect size)40 relative to the change in cons (a small effect size)40 supports Prochaska’s43 contention of the strong and weak principles of behavior change. Although confirmation is required in prospective studies, this suggests that in order to stimulate movement from the early stages, a greater emphasis may be needed on raising perceptions of the ben-
Table 3.
Comparison of model fita among four nested models (n = 398). χ2
df
6847
91
398
latent variables) Two factor correlated (allows for correlated latent variables)
Model Null (assumes all manifest variables independent)
RMR
GFI
NNFI
CFI
77
0.23
0.94
0.94
0.95
276
77
0.12
0.96
0.97
0.97
236
76
0.09
0.97
0.97
0.98
One factor (assumes all manifest variables loaded on one latent variable) Two factor uncorrelated (assumes no correlation among two
a
Fit indices are based on the Weighted Least Squares method. The relatively high values may be a result of the estimation method used. Compari-
son of indices, however, should be relative across models rather than absolute. The chi-square difference test between the two-factor correlated model and the next alternative model, the uncorrelated model, gives ∆χ2 = 40.38, ∆df = 1 (p < .001). The other indices (RMR, GFI, NNFI, and CFI) consistently perform best for the two-factor correlated model when compared with the two-factor uncorrelated and one-factor models. RMR = root mean square residual, GFI = Goodness of Fit Index, NNFI = Non-Normed Fit Index, CFI = Comparative Fit Index.
Journal of Nutrition Education Volume 33 Number 5
Figure 1. Pro and con scores by stages of change. Scores were presented as standardized T-scores (mean = 50, SD = 10). Pros: P > PC (p < .0001); P > C (p < .01). Cons: PC > A/M (p < .01); C, P > A/M (p < .05). Plotted for four stage groupings, with action and maintenance stages combined into the post-action stage. PC (precontemplation, n = 210), C (contemplation, n = 234), P (preparation, n = 207), A/M (post-action, n = 65).
might not see any added benefit from increasing their intake any further.45 The pros and cons crossed over between contemplation and preparation. Depending on the health behaviors examined, the crossing over from the pros to the cons being more important generally occurred between the contemplation and preparation stages.15–17 Some authors have reported the crossover to occur between the preparation and action stages.18,46 These findings suggest that people would need to perceive higher pros than cons prior to taking action.17 Previous work12,15,16 has also demonstrated that the eight decision-making considerations were not equally salient and that the final item pool reflected the nature of the behavior studied. In this study, the pro items represented self-approval, instrumental gains for self, and approval from others. None of the items relating to instrumental gains for others were included by the selection process. Loadings for the pro items were also weighted toward the “self ” items, suggesting that social influence may be less salient in determining perceived benefits to eat more fruits and vegetables. On the other hand, items representing perceived social barriers loaded high on the con scale. In Singapore society, where eating out and sharing of communal dishes are common, it is likely that fussing over having more fruits and vegetables or requesting extra vegetable dishes would inconvenience others. It is of interest to note that considerations of taste, price, and inconvenience did not emerge as particularly salient in driving consumption decisions. In the 5 A Day baseline survey, most respondents reported that they liked the taste of fruits (81.7%) and vegetables (71.4%).5 In a nationwide U.K. survey,47 taste, healthfulness, nutrition, family liking, and fit-
September • October 2001
263
ting into eating habits emerged as the five most important beliefs for eating more fruits and vegetables. However, in a recent review of environmental and personal influences on dietary behaviors for chronic disease prevention,48 it was noted that widely held perceptions such as taste or time constraints are not necessarily predictive of dietary behaviors. Consumer research conducted for the 5 A Day for Better Health campaign27 also found that taste, price, and convenience are both benefits and barriers. Although focus group participants mentioned (good) prices, taste, and convenience as benefits derived from buying and eating fruits and vegetables, complaints about texture, flavor, availability, and preparation time were also common reasons given for not eating more of these foods. Similarly, a recent study examining cancer-related nutrition beliefs and attitudes49 revealed that Americans may believe that healthful foods taste good, but these were not necessarily the best-tasting foods. Although necessary, taste may not be sufficient to motivate dietary change. Importantly, decisional balance items represented immediate gains or losses rather than long-term outcomes.This is consistent with the findings of Balch et al.,27 whose focus group participants considered cancer a remote concern beyond their control and rarely taken into account when making food choices. The present study achieved a fairly good response rate of 71%, which is close to the reported average of 73% for general public surveys that used similar protocols.33 Although practically every household in Singapore has a telephone, the sampling approach excludes the 14.5% of the residential telephone numbers not listed.50 However, housing type (indicative of socioeconomic status in Singapore) and educational level characteristics of the study sample were comparable with national averages data.51 The administration of study instruments was also scheduled to reduce the likelihood of response bias.The 3-month delay between the mail survey and the 24-hour dietary interviews employed to validate the staging algorithms reduced the likelihood that prior completion of the food frequency and decisional balance questions influenced subjects’ dietary recalls. Moreover, subjects were unaware that they would be asked to recall the specific day’s diet, and the interviews assessed all food and drinks, not just fruits and vegetables.
IMPLICATIONS FOR RESEARCH AND PRACTICE To effect change, we will probably need to apply both individual change processes and public health policies.43 The results of this study suggest that to raise fruit and vegetable consumption, it may be most effective to target those in the early stages of change with messages about more immediate benefits directly relevant to the self. For those in the later stages, public health policies (e.g., promoting availability of fruits and vegetables at subsidized prices) may help to reduce
264
Ling and Horwath/PERCEIVED BENEFITS AND BARRIERS OF INCREASED FRUIT AND VEGETABLE CONSUMPTION
barriers and allow these individuals to take action and maintain the behavior change. The robustness of the decisional balance and stage constructs in a non-white population, rarely investigated elsewhere,22 is encouraging. Further research is needed to determine the longitudinal shift in pros and cons as people progress through the stages and the influence of intervention trials on decisional balance. Further testing of the scale is also needed to examine its generalizability to an ethnically diverse population.
ACKNOWLEDGMENTS This work was undertaken as part of the Ph.D. dissertation of the first author. Ethical issues related to the study were considered as part of the grant-approving process by the National Medical Research Council, Singapore.The authors are grateful to the Department of Nutrition, Singapore, for support and assistance with survey implementation; Professor Rob Lawson and Mrs. Sheila Williams at the University of Otago for statistical advice; Dr. Pauline Gulliver for assistance with data analysis; and Drs. Stephanie Ounpuu and Judy Sheeshka for feedback on the scales.
REFERENCES 1. World Health Organization. Diet, nutrition and chronic diseases: report of a WHO study group. Technical Report Series 797. Geneva: WHO, 1990. 2. Steinmetz KA, Potter JD.Vegetables, fruit and cancer. I. Epidemiology. Cancer Causes Control 1991;2:825–57. 3. Steinmetz KA, Potter JD.Vegetables, fruit and cancer. II. Mechanisms. Cancer Causes Control. 1991;2:427–41. 4. Block G, Patterson B, Subar A. Fruit, vegetables and cancer prevention: a review of the epidemiological evidence. Nutr Cancer 1992;18:1–29. 5. Krebs-Smith SM, Cook A, Subar AF, Cleveland L, Friday J. US adults’ fruit and vegetable intakes, 1989 to 1991: a revised baseline for the Healthy People 2000 Objective. Am J Public Health 1995;85:1623–9. 6. Ministry of Agriculture, Fisheries and Food. National food survey, 1995. London: HMSO, 1996. 7. Ministry of Health, Singapore, Food and Nutrition Department. Fruit and vegetable consumption—survey of shoppers. Singapore: Ministry of Health, 1990. 8. Ling AMC, Horwath C, Parnell W. Validation of a short food frequency questionnaire to assess consumption of cereal foods, fruit and vegetables in Chinese Singaporeans. Eur J Clin Nutr 1998;52:557–64. 9. Janis IL, Mann L. Decision-making: a psychological analysis of conflict, choice and commitment. New York: Collier MacMillan, 1977. 10. Prochaska JO, Redding C, Evers K. The Transtheoretical Model of behavior change. Health behaviour and health education: theory, research and practice. 2nd Ed. San Francisco: Jossey-Bass, 1997:60–84. 11. Velicer WF, DiClemente CC, Prochaska JO, Brandenburg N. Decisional balance measure for assessing and predicting smoking status. J Pers Soc Psychol 1985;48:1279–89.
12. O’Connell D,Velicer WF. A decisional balance measure and the stages of change model for weight loss. Int J Addict 1988;23:729–50. 13. Marcus BH, Rakowski W, Rossi JS.Assessing motivational readiness and decision making for exercise. Health Psychol 1992;11:257–61. 14. Steptoe A,Wijetunge S, Doherty S,Wardle J. Stages of change for dietary fat reduction: associations with food intake, decisional balance and motives for food choice. Health Educ J 1996;55:108–22. 15. Ounpuu S.The pros and cons of reducing fat as intermediate outcome variables in the Transtheoretical Model: validation of a measurement model. Unpublished Ph.D. dissertation, University of Guelph, Guelph, Ontario, Canada, 1997. 16. Gulliver P, Horwath CC. Decisional balance and stage of change for dairy consumption.Abstract presented at the 20th Annual Convention of the Society for Behavioral Medicine, San Diego, CA, March 1999. 17. Prochaska JO,Velicer WF, Rossi JS, et al. Stages of change and decisional balance for 12 problem behaviors. Health Psychol 1994;13:39–46. 18. Herrick AB, Stone WJ, Mettler MM. Stages of change, decisional balance and self-efficacy across four health behaviors in a worksite environment. Am J Health Promotion 1997;12:49–56. 19. Prochaska JO, DiClemente CC,Velicer WF, Ginpil S, Norcross JC. Predicting change in smoking status for self-changers. Addict Behav 1985;10:395–406. 20. Rossi SR, Greene GW, Reed G, Rossi JS, Prochaska JO, Velicer WF. Cross-validation of a decisional balance measure for dietary fat reduction. Abstract presented at the 14th Annual Convention of the Society for Behavioral Medicine, Boston MA, April 1994. 21. Brug J, Hospers HJ, Kok G. Differences in psychosocial factors and fat consumption between stages of change for fat reduction. Psychol Health 1997;12:719–27. 22. Campbell MK, Symons M, Demark-Wahnefried W. Stages of change and psychosocial correlates of fruit and vegetable consumption among rural African-American church members. Am J Health Promotion 1998;12:185–91. 23. Reicks M, Randall JL, Haynes BJ. Factors affecting consumption of fruits and vegetables by low-income families. J Am Diet Assoc 1994;94:1309–11. 24. Brug J, Debie S, van Assema P,Weijts W. Psychosocial determinants of fruit and vegetable consumption among adults: results of focus group interviews. Food Qual Pref 1995;6:99–107. 25. Treiman K, Fremuth V, Darmon D, et al.Attitudes and behaviors related to fruits and vegetables among low-income women in the WIC programme. J Nutr Educ 1996;28:149–56. 26. Loughrey KA, Balch GI, Lefebvre RC, et al. Bringing 5-a-Day consumers into focus: qualitative use of consumer research to guide strategic decision making. J Nutr Educ 1997;29:172–7. 27. Balch GI, Loughrey KA,Weinberg L, Lurie D, Eisner E. Probing consumer benefits and barriers for the national 5-a-Day campaign focus group findings. J Nutr Educ 1997;29:178–83. 28. Dittus KL, Hillers VN, Beerman KA. Benefits and barriers to fruit and vegetable intake: relationship between attitudes and consumption. J Nutr Educ 1995;27:120–6. 29. Krebs-Smith AM, Heimendinger J, Patterson BH, Subar AF, Kessler R, Pivonka E. Psychosocial factors associated with fruit and vegetable consumption. Am J Health Promotion 1995;10:98–104. 30. Brug J, Lechner L, deVries H. Psychosocial determinants of fruit and vegetable consumption. Appetite 1995;25:285–96.
Journal of Nutrition Education Volume 33 Number 5 31. Lappalainen R, Saba A, Holm A, Mykkanen H, Gibney MJ. Difficulties in trying to eat healthier: descriptive analysis of perceived barriers for healthy eating. Eur J Clin Nutr 1997;51(Suppl 2):S36–40. 32. Zunft HJF, Friebe D, Seppelt B, et al. Perceived benefits of healthy eating among a nationally-representative sample of adults in the European Union. Eur J Clin Nutr 1997;51(Suppl 2):S41–6.
September • October 2001
265
42. DeVellis RF. Scale development: theory and applications. Newbury Park, CA: Sage, 1991. 43. Prochaska JO. Strong and weak principles for progressing from precontemplation to action on the basis of twelve problem behaviors. Health Psychol 1994;13:47–51.
33. Dillman DA. Mail and telephone surveys: the total design method. Toronto: John Wiley & Sons, 1978.
44. Prochaska JO, DiClemente CC,Velicer WF, Ginpil S, Norcross JC. Predicting change in smoking status for self-changers. Addict Behav 1985;10:395–406.
34. Ling AM, Horwath CC. Self-efficacy and consumption of fruit and vegetables—validation of a summated scale. Am J Health Promotion 1999;13:290–8.
45. Brug J, Glanz K, Kok G. The relationship between self-efficacy, attitudes, intake compared to others, consumption and stages of change related to fruit and vegetables.Am J Health Promotion 1997;12:25–30.
35. Ministry of Health, Singapore,Training and Health Education Department. Guidelines for a healthy diet—recommendations of the National Advisory Committee on Food and Nutrition. Singapore: Ministry of Health, 1989. 36. Ling AM, Horwath CC. Defining and measuring stages of change for dietary behaviors: readiness to meet fruit, vegetable and grain guidelines among Chinese Singaporeans. J Am Diet Assoc 2000;100: 898–904. 37. Hair JF,Anderson RE,Tatham RL, Black WC. Multivariate data analysis with readings. Englewood Cliffs, NJ: Prentice-Hall, 1995. 38. Jöreskog K, Sörbom D. Lisrel 8: user’s reference guide. Chicago: Scientific Software International, 1996. 39. Hu LT, Bentler PM. Evaluating model fit. Structural equation modeling: concepts, issues and applications. Newbury Park, CA: Sage, 1995:76–99. 40. Cohen J, Cohen P.Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum, 1983. 41. SPSS 8.0 for Windows. Chicago: SPSS, 1998.
46. Marcus BH, Owen N. Motivational readiness, self-efficacy and decision-making for exercise. J Appl Soc Psychol 1992;22:3–16. 47. Cox DN, Anderson AS, McKellar S, Reynolds J, Lean MEJ, Mela DJ. Vegetables and fruits: barriers and opportunities for greater consumption. Nutr Food Sci 1996;5:44–7. 48. Harnack L, Block G, Lane S. Influence of selected environmental and personal factors on dietary behavior for chronic disease prevention: a review of the literature. J Nutr Educ 1997;29:306–12. 49. Harnack L, Block G, Subar A, Lane S. Cancer prevention-related nutrition knowledge, beliefs and attitudes of U.S. adults: 1992 NHIS cancer epidemiology supplement. J Nutr Educ 1998;30:131–8. 50. Singapore Telecom.The Singapore phone book, residential listing. July 1997. 51. Department of Statistics. General household survey report. Statistical Release I. Singapore: Ministry of Health, 1995.
Corrections in: Sobal J. Sample Extensiveness in Qualitative Nutrition Education Research. J Nutr Educ 2001;33:184–192. On page 184, under “Results”, the abstract should have stated “Ten studies used observation/fieldwork, and eight used other types of qualitative research, with mixed patterns of sample extensiveness in those studies.” On page 185, 2nd paragraph, it should have stated “In contrast to quantitative research, qualitative research is grounded in nonpositivist paradigms that include interpretivist, constructionist, constructivist, phenomenological, post-modernist, and other perspectives that focus on gathering and analyzing non-numeric data in the form of text, images, sounds, and objects.” On page 189, the last paragraph under “Discussion” should have stated “The author is also examining sample extensiveness in other journals in other fields.”