Development and validation of Transtheoretical model-based questionnaire on micronutrients for adolescents: Psychometric properties

Abstract

BACKGROUND:

The Transtheoretical model (TTM) is one of the most commonly used theories in assessing behavioral change and designing more effective interventions. The study aimed to develop and validate a questionnaire based on the TTM to assess the behavioral change of adolescent girls related to dietary micronutrient intake.

MATERIALS AND METHODS:

A TTM-based questionnaire (Items 36) was validated through 150 adolescent girls aged 13–18 years. The questionnaire is made on four constructs: stages of change, processes of change, self-efficacy, and decision balance. The content validity and face validity were assessed by expert panels and participants, respectively. The confirmatory factor analyses (CFA) were used to test construct validity. The Cronbach’s alpha coefficient and Intraclass Correlation Coefficient (ICC) were used to analyze the internal consistency and stability of the questionnaires.

RESULTS:

Results showed that all content validity Index values of items in questionnaires were within the range (0.8-1). Based on face validity results, most of the items were clear, easy, and understood by participants. The CFA results included a comparative factor index of 0.99 for all constructs; a goodness of fit index ranging from 0.88 to 0.951, and a root mean square error of approximation ranging from 0.039-0.069, suggesting good model fit indices. The reliability analysis with Cronbach’s alpha of 0.737-0.953 and ICCs of 0.737-0.845 was acceptable.

CONCLUSION:

These psychometric analysis results provided evidence for the appropriateness, reliability, and validity of the developed questionnaire. Hence, this questionnaire may help to develop customized nutrition intervention programs based on behavioral changes in the context of micronutrient intake.

Introduction

The period of adolescence is characterized by biopsychosocial changes, and it involves special nutritional needs.[1] Adolescents tend to unhealthy eating practices, which include low consumption of fruits and vegetables, high intake of energy-dense beverages, fast foods, chocolates, and skipping breakfast.[2,3] Micronutrients are essential throughout life. Still, the adolescent period is crucial due to the rapid physiological growth and development that occurs during this time, and their deficiency is prevalent in this age group.[4,5] There are many initiatives and programs, such as fortification, micronutrient supplementation, diet diversification, and nutrition education, implemented by the Government of India, to prevent the problem of micronutrient malnutrition.[6] Studies based on educational and/or behavioral interventions, either alone or in combination, have been done to improve dietary intake. The findings of these studies indicate that interventions that focused only on nutrition education were less effective than interventions that applied behavioral change techniques.[7] It is necessary to understand why this age group does not consume adequate amounts of micronutrient-rich sources and how the intake of micronutrients can be increased. This is where the planning of model-based intervention and behavioral change theories can be beneficial to change their behavior. One of the key components under the Prime Minister’s Overarching Scheme for Holistic Nutrition (POSHAN) Abhiyaan in India is behavior change to promote positive health behavior among populations, which can bring required dietary modification and hence diversification.[8,9]

The Transtheoretical model (TTM) is one of the most commonly used theories in assessing behavioral change and designing more effective interventions. It was first proposed by Prochaska and DiClemente[10] to promote behavioral change in the context of cigarette smoking, and after that, it was applied to fruit and vegetable consumption by Di Noia et al.[11] The TTM appears to be an effective nutritional intervention strategy meant to improve dietary intake in adolescents, and its application in various contexts shows that the TTM can be implemented in numerous settings because of its potential and flexibility.[12] The model is composed of four constructs, including stages of change (SOC), processes of change (POC), self-efficacy (SE), and decision balance (DB). This model provides an integrative framework for understanding the process of health behavior, which involves progress through five SOC, ten POC, two DB, and five SE [Figure 1].[13] The model has been extensively utilized for health education interventions, including promoting physical activity,[14,15] encouraging fruit and vegetable consumption,[16] reducing fat intake,[17] and reducing fast food consumption among obese and overweight female adolescents,[18] changing eating habits in adolescents with obesity,[19] a diet behavior modification of pregnant women with iron deficiency anemia,[20] all of which have been demonstrated to be effective. To the best of our knowledge and based on an extensive literature review, no prior behavioral change intervention study based on the TTM model for micronutrients among adolescent girls has been done. The TTM questionnaire will be useful for assessing and monitoring behavior change as well as designing more effective interventions to improve the micronutrient intake among adolescents. Therefore, the present study aimed to develop and validate a questionnaire for micronutrients.

Figure 1.

Constructs of transtheoretical model

Materials and Methods

Study design and setting

This was a cross-sectional study conducted as part of one project. In that, the questionnaire was developed and validated to achieve one of the objectives of the project, which aimed to develop a comprehensive nutrition intervention for the prevention and control of micronutrient deficiency among adolescent girls. A study was conducted at one purposively selected private school in Navi Mumbai, India.

Study participants and sampling

A total of 150 school-going adolescent girls of the age group 13–18 years were included. We randomly selected 30 students each from the eighth to twelfth grades.

The sample size was calculated separately for each validity test. For construct validity, the sample size was decided based on the available literature, suggesting 3–20 times the number of items.[21,22] In the POC, SE, and DB questionnaires, a total of 32 items were formulated, so the required sample size was a minimum of 96. For reliability analysis, fifty participants were included, as per the literature on sample size determination for internal consistency and test-retest analysis.[23,24]

Data collection tool and technique

The first section of the questionnaire contains basic demographic information on age, education of the participant, education, and occupation of parents, dietary habits, disease history of the past six months, food allergy, and monthly household income, along with menstrual history. Socioeconomic status was assessed by using the Revised Kuppuswamy’s scale, 2023.[25]

Development of behavioural change questionnaire

This questionnaire development was done using the following five steps: Item generation, assessing face, content, construct validity, and reliability analysis [Figure 2].

Figure 2.

Questionnaire development and validation process. EFA = Exploratory factor analysis, CFA = Confirmatory factor analysis, CVI = Content validity index, ICC = Intraclass correlation coefficient

Phase 1: Item generation

The TTM questionnaire was developed by Prochaska and colleagues for smoking cessation.[26,27] and a few other researchers have used this model for nutrition-related studies to assess health-related behavioral changes.[11,20,28,29] Using these questionnaires as the base model, the items were formulated for our study questionnaire. The TTM questionnaire construct is as follows:

1. Stages of Change: These questions assess the current condition of the individuals’ dietary habits related to micronutrients. This construct included 4 questions related to micronutrient-rich food consumption at present and the readiness to incorporate these food items into their diet. Based on the response, the participants will be grouped into one of the following 5 SOC namely, Pre-Contemplation, Contemplation, Preparedness, Action, and Maintenance. We considered the SOC construct as a dependent variable, therefore, we did not perform a reliability and construct validity analysis for it.[30]

2. Processes of Change: The Process of change construct represents the processes that people take to modify their attitudes, beliefs, and actions on a certain issue. It comprises two categories—Experiential and Behavioural changes—with 10 techniques that measure the subject’s process of change in terms of experiences, thoughts, and behavior. Based on these aspects, the POC construct was developed with 8 statements to assess experiential changes and 10 statements on Behavioural changes related to dietary micronutrient intake. Each of the questions/statements was to be rated from 1 to 5 according to the experience and thoughts of the participants. The scores were 1 for never, 2 for seldom, 3 for occasionally, 4 = often, and 5 for repeatedly.

3. Self-Efficacy: The third construct of the model is self-efficacy, which measures a person’s capacity and degree of confidence in their ability to adjust and engage in new health behavior in response to environmental cues such as challenging circumstances, positive social interactions, and negative affect. The developed questionnaire included 2 items in each of the cues, with a total of 6 questions assessing the self-efficacy of the participants for choosing micronutrient-rich foods. A 3-point rating scale was used for scoring each of the cues. The scores were 1 for not confident, 2 for somewhat confident, and 3 for confident.

4. Decision Balance: Decision Balance measures an individual’s assessment of the pros and cons of adopting a new health behavior. This questionnaire included 2 questions related to pros and 6 questions to cons. Similar to the SE, a 3-point Likert scale is used for assessing the decision-balance capability of the participants. The scores for pros were 1 for don’t agree, 2 for somewhat, and 3 for agree, whereas scores for cons were 1 for agree, 2 for somewhat, and 3 for disagree.

Validation

Phase 2: Content validity

The developed questionnaire was sent to six expert panels. Two panel members had experience using the TTM, and the remaining 4 had worked on behavior change assessments in various age groups. All the panel members were asked to rate the relevance of each of the items in each construct (SOC, POC, SE, and DB) using a 4-point Likert scale (1 = not relevant, 2 = somewhat relevant, 3 = relevant, and 4 = highly relevant). The Content Validity Index (CVI) was calculated as the number of experts (who gave a score of 3 or 4 to each item for relevancy) divided by the total number of experts. The panels were also asked to review the items for clarity and understanding. Items with a CVI higher than 0.8 were retained, and those < 0.8 CVI were removed.[31]

Phase 3: Face validity

A subsample of fifty participants was asked to evaluate each item for readability, understanding, clarity of wording, and difficulty level by giving the following options: good, moderate, and poor. Participants were also asked to identify any difficult words or phrases.

Phase 4: Construct validity

The Exploratory factor analysis (EFA) was conducted for POC, SE, and DB questionnaires. The number of factors to extract was decided based on the TTM theory. The principal component analysis was assessed by the factor analysis method to assess which observed variables were associated with which factor, and the varimax method was applied to rotate all the factors. The Kaiser-Meyer-Olkin (KMO) test was used for sampling adequacy (>0.6), and Bartlett’s test of sphericity (P ≤ 0.05) was considered for suitability of conducting factor analysis. Factor loadings >0.4 were used to select an item associated with a factor. After extracting the factors from EFA, confirmatory factor analysis (CFA) was used for POC with factors (experiential and behavioral), DB (pros and cons), and SE. The CFA was performed to determine whether data supported the TTM theory and factors. Various fit indices were utilized to assess the model fit: Goodness-of-fit index (GFI), Comparative Fit Index (CFI), and Root Mean Square Error of Approximation (RMSEA). The values of the CFI >0.95 and from 0.90 to 0.95 were considered as excellent model fit and acceptable, respectively. For the GFI, the values > 0.90 were considered as a good model fit, whereas those values between 0.85 and 0.90 were represented as acceptable. The values of the RMSEA < 0.05 reflected an excellent model fit, and values from 0.05 to 0.08 were considered acceptable.[32]

Phase 5: Reliability

The internal consistency of the items in each construct was assessed using Cronbach’s α. Items with Cronbach’s α value ≥ 0.7 were retained.[33] To check for the stability of the questionnaire, a test-retest was conducted among the 50 participants at 15-day intervals, after which the Intraclass Correlation Coefficient (ICC) was obtained. An ICC value > 0.70 suggests good reliability.[34]

Ethical consideration

Before data collection, detailed information about the study was given to the participants and their parents using an information sheet. Then, written assent was obtained from all the participants, and consent was obtained from their parents. The Institutional Ethics Committee fully reviewed and approved this study (Approval code: RES./02/2022/178).

Statistical analysis

Data was analysed using IBM^® SPSS version 25 (SPSS, Chicago IL, USA). For categorical variables, frequency, percentages, and continuous variables, the mean and standard deviation values were calculated. The reliability analysis and the EFA using principal component analysis and varimax Rotation for construct validity were done on SPSS. The CFA was done on the IBM SPSS Amos 26.0, Microsoft Windows, Multilingual. For CVI estimation, Microsoft Excel was used.

Results

A total of 150 participants were included in the pilot test to evaluate the validity, internal consistency, and test-retest reliability of the questionnaire. The mean age of the participants was 14.78 ± 1.40 years. More than two-thirds of the participants (64.2%) were from the upper-lower class. The majority of girls (77.3%) attained menarche at the age of 13–14 years. More than sixty percent of the girls (66%) had regular periods, and 57.1% had dysmenorrhea (painful periods).

Content validity

The mean CVI scores for relevance were 0.86, 0.941, 0.917, and 0.938 for SOC, POC, SE, and DB constructs, respectively. All CVI values of items in the questionnaires were within the range. Based on the comments and recommendations given by the experts, 1 item in the SOC questionnaire was modified, and rich sources of micronutrients were specified for better understanding. At the end of the content validity, a total of 4 items in SOC, 18 in POC, 6 in SE, and 8 in DB were included.

Face validity

Based on face validity results, most of the items were clear, easy, and understood by participants. Only a few items were slightly revised based on participants’ responses.

Construct validity

1. Processes of Change: The EFA results showed that the KMO was adequate (0.944) and Bartlett’s test of sphericity was found to be significant (χ² = 5026.871, P = 0.000). Prochaska and Velicer[13] identified two-factor structures that measure experiential and behavioral in the original POC metric. The results of EFA showed that eight items were loaded on the experiential POC, which was expected to load on the same as per theory, whereas the remaining items were loaded on the behavioral POC as expected. The experiential and behavioral POC factors explained 55.98% and 86.61% of the item’s variance, respectively [Table 1]. To confirm the two-factor model, the CFA for the POC questionnaire was conducted. Initially, modification indices suggested that an improvement in the model fit would be substantial if the error terms of three items were allowed to adjust. Thus, two pairs of error variances were allowed to be correlated with each other. There was improvement shown in the model after adjustment, and it reflected an acceptable fit to the data (CFI = 0.99, GFI = 0.88, and RMSEA = 0.053) [Figure 3].

2. Self-Efficacy: For the self-efficacy questionnaire, EFA showed that KMO was adequate (0.890) and Bartlett’s test of sphericity was found to be significant (χ² = 1739.202, P = 0.000). About 92.93% variance was explained by this component in the questionnaire [Table 1]. For the confirmation of the single-factor model derived from EFA, CFA was conducted. For self-efficacy, modification indices also suggested improvement in the model. After adjustment in the error of the two items, the model was improved and indicated as an acceptable fit to the data (CFA = 0.99, RMSEA = 0.039, GFI = 0.991) [Figure 4].

3. Decision Balance: Both EFA and CFA were applied to develop the DB questionnaire. The PCA with the varimax rotation method was used for EFA. The results showed that the KMO was adequate (0.807) and Bartlett’s test of sphericity was found to be significant (χ² = 1254.461, P = 0.000). Similar to the POC questionnaire, the number of factors to rotate was decided based on TTM theory, which indicates two-factor structures: pros and cons. The results of EFA showed that six items were loaded on one factor ‘cons’ and two items were loaded on another factor ‘pros’ as expected. The cons factor accounted for 61.23%, whereas the pros factor explained 82.08% of the items ‘variance [Table 1]. Further, these two factors derived from the EFA were confirmed through CFA. Similar to POC and SE, modification indices suggested an improvement in the model fit. There was improvement shown in the model after adjustment and it reflected as an acceptable fit to the data (RMSEA 0.069, CFI = 0.99, GFI 0.951) [Figure 5].

Table 1.

Exploratory factor analysis of TTM constructs

Factors		Initial eigenvalues						Extraction of sums of squared loadings						Range of factor loadings
		Total		% of variance		Cumulative percentage		Total		% of variance		Cumulative percentage
Process of change
Behavioural		10.08		55.98		55.98		10.08		55.98		55.98		0.666–0.972
Cognitive		5.51		30.63		86.61		5.51		30.63		86.61		0.868–0.977
Self-efficacy		5.58		92.93		92.93		5.58		92.93		92.93		0.940–0.986
Decision balance
Cons		4.90		61.23		61.23		4.90		61.23		61.23		0.745–0.946
Pros		1.67		20.85		82.08		1.67		20.85		82.08		0.946–0.962

Extraction method: Principal component analysis, Rotation method: Varimax with Kaiser Normalization, Factor loading cut off >0.50

Figure 3.

Two model factors of the processes of change (POC) questionnaire. The results obtained from the confirmatory factor analysis

Figure 4.

One model factor of Self Efficacy (SE) questionnaire. The results obtained from the confirmatory factor analysis

Figure 5.

Two model factors of the decision balance (DB) questionnaire. The results obtained from the confirmatory factor analysis

Overall, the analysis’s findings showed that the theory-driven component structure of the POC, DB, and SE questionnaires was an acceptable fit to the data and gave supportive evidence for their construct validity [Table 2].

Table 2.

Details of items used in the questionnaire according to TTM constructs

TTM constructs		Sub factors		No of items		Item description
Stages of change				1		Do you currently consume the micronutrient-rich foods daily?
				2		Do you intend to consume micronutrient-rich foods daily in the next 6 months?
				3		Do you intend to consume micronutrient-rich foods daily in the next 30 days?
				4		Have you been consuming micronutrient-rich foods daily for the past 6 months?
Processes of change		Experiential
		Consciousness raising		1		I look for information on sources of micronutrient-rich foods
				2		I try to explore the articles/advertisements on how to eat more micronutrient-rich foods
		Dramatic relief		3		I am afraid of the consequences to my health if I do not eat micronutrient-rich foods
		Environmental re-evaluation		4		If I eat micronutrient-rich foods, it will reduce healthcare costs
				5		I realize that if I don’t eat micronutrient-rich foods, I may get ill and be a problem to people close to me
		Self-re evaluation		6		I feel healthier, happier if I eat micronutrient-rich foods
		Social liberation		7		I have noticed that many people are buying micronutrient-rich foods
				8		Supermarkets and online stores are making it easier for consumers to buy a variety of micronutrient-rich foods
		Behavioral
		Self-liberation		9		I tell myself I can choose to eat more micronutrient-rich foods
				10		I make commitments to decrease my health risks by eating micronutrient-rich foods
		Reinforcement management		11		I try to eat micronutrient-rich foods as a promoter to increase my overall performance
		Helping relationship		12		I have someone/family/friends who encourage me to have micronutrient-rich foods
				13		I have someone at home who prepares micronutrient-rich foods for me
TTM constructs		Sub factors		No of items		Item description
		Counterconditioning		14		I purchase micronutrient-rich foods instead of unhealthy/junk foods
				15		I find ways to include micronutrient-rich foods in place of less healthy/junk food
		Stimulus control		16		I try to put posters/pictures of micronutrient-rich foods at home to remind me to eat them
				17		I maintain a food diary to keep a check on micronutrient-rich food intake
				18		I avoid eating outside foods where healthy options are unavailable
Self-efficacy				1		When I am in a hurry, I prepare meals rich in micronutrients
				2		When I have the chance to choose, I can eat foods rich in micronutrients
				3		I can shop for a variety of micronutrient-rich foods
				4		When I feel restless or bored, I can eat foods rich in micronutrients
				5		When I celebrate with others, I can eat plant foods rich in micronutrients
				6		When I eat outside the home, I can eat foods rich in micronutrients
Decision balance		Pros		1		It is good for my health if I eat foods rich in micronutrients
				2		Eating foods rich in micronutrients will improve my health status.
		Cons		3		It is difficult to find foods rich in micronutrients that I like.
				4		I don’t have time to shop the foods rich in micronutrients
				5		I worry about the safety of chemicals used in foods rich in micronutrients
				6		It is expensive to purchase foods rich in micronutrients
				7		It would be too confusing for me to follow all the recommendations about eating foods rich in micronutrients
				8		I don’t think I know enough about foods rich in micronutrients

Reliability

The total Cronbach’s α values for POC, SE, and DB were 0.953, 0.737, and 0.898 respectively. All items of each questionnaire had Cronbach’s α above 0.7, which indicates a good internal consistency. All three questionnaires were administered twice to participants at intervals of 15 days. The mean ICC Values for the POC, SE, and DB questionnaires were 0.845, 0.773, and 0.737, respectively, which are satisfactory.

Final questionnaire

Based on the results of validity and reliability, 4 SOC, 18 POC, 6 SE, and 8 DB items were finalized for the questionnaire. The details of items under each construct are given in Table 2.

Discussion

The results showed that it was a valid questionnaire for measuring behavior change toward micronutrient intakes among adolescent girls. The methods that we used for validation in the present study are similar to other studies that validated TTM-based questionnaires. We assessed the psychometric properties of the questionnaire using several validation methods, such as content, face, and construct validity. Reliability analysis was performed using internal consistency and test and retest methods. A study by Sarbandi et al.[30] examined the psychometric properties of the Iranian version of TTM-based four questionnaires for smoking cessation using similar methods. Another study used similar methods for the psychometric testing of four TTM-based questionnaires for behavior, including completing a healthcare proxy.[32]

The SOC is the most explored construct of the TTM model, and it serves as an index of motivational readiness to change.[35] It helps to identify at what point in time people are ready to change their dietary behavior. We used confirmatory factor analysis to determine whether there was coherence between the data and the theoretical framework. As expected, the POC revealed two subscales: behavioral and experiential. Similar structures were found in other studies.[30,32] It is a key construct of the TTM and plays an important role when developing questionnaires for assessing behavior change. The contribution of these types of subscales in the POC suggests that while creating interventions, care should be taken to create the content that is necessary for the target audience to go from the stage of preparation to that of action. The SE showed a single-factor model, which was developed to assess an individual’s confidence in eating micronutrient-rich sources in different challenges. These findings confirmed the underlying structure found in the previous studies on self-efficacy_.[32,36] The DB confirmed two subscales: pros and cons, which showed that the questionnaire was well-differentiated benefits and barriers involved in making decisions to incorporate micronutrient-rich sources in the diet. A similar underlying structure was found in other studies.[35,36] Pros and cons are an important part of the decision balance scale. Before attempting to change behavior in target populations, it is important to consider the specific pros and cons of that behavior.

The TTM strategy is a successful strategy for nutritional interventions to improve dietary intake. A study by Gur et al.[16] found that an intervention based on the TTM model increased fruit and vegetable consumption in adolescents. Another study by Ibrahim et al.[20] showed that the use of four constructs, SOC, POC, SE, and DB, in the intervention program improved the diet behavior of pregnant women.

The strength of the present study is that there has not been a study that developed and validated TTM-based questionnaires to change behavior and adopt healthy dietary practices to prevent micronutrient deficiency in adolescents. As the questionnaire is developed for specific micronutrients such as vitamins A, C, D, B12, folate, iron, zinc, and calcium in the present study, we have covered all the aspects related to these nutrients. A review by Nakabayashi et al.[12] on TTM-based nutritional interventions in adolescents showed that most of the nutrition-related studies have applied the TTM strategy only to improve fruit and vegetable consumption, promote physical activity, and reduce dietary fat consumption and obesity. The SOC has been widely used in TTM constructs to develop tailored intervention programs in previous studies, and the use of other constructs of the TTM with SOC in the intervention to improve its effectiveness is needed.[12,37] In the present study, we have used all four constructs of TTM to develop a questionnaire on micronutrients, which can be used to assess the change in the behavior of adolescents as well as to design the nutrient intervention program. The TTM’s affordability, adaptability in solving many issues, and flexibility to customize the intervention based on the participant’s level of readiness for change are some of its advantages. It has been found that tailored interventions are known to be more successful in changing health-related behaviors.

This study also has some limitations of generalizability because the study was done in girls alone, used a purposive sampling method, and the selection of an urban setting to assess the psychometric properties of the questionnaire. Further studies with different diverse settings are needed to validate this questionnaire.

Conclusion

This present study showed the psychometric evidence for a valid and reliable questionnaire to assess behavior and promote behavioral change in the context of micronutrient intake. In the questionnaire, the inclusion of stages of change, the experiential and behavioral process of change, the capacity and confidence level of participants, and advantages and disadvantages of the adoption of new behavior can be helpful to give more understanding about the behavioral aspect of individuals. The behavior change model has been incorporated into health policies as one of the strategies for the betterment of nutritional status and prevention of nutritional deficiencies. The use of questionnaires would enable policymakers and health educators to more effectively design, target, and monitor the progress of interventions for increasing micronutrient consumption in adolescents.

Conflicts of interest

The authors declare no conflict of interest.

Ethical approval

Ethical approval was obtained from the Institutional Ethics Committee, MGM Institute of Health Sciences.

Funding Statement

This research is part of a project, which is granted by the Indian Council of Medical Research (Project ID: F.N.5/9/1441/2021-22Nut).