Food security is defined as ‘when all people at all times have physical, social and economic access to food, which is safe and consumed in sufficient quantity and quality to meet their dietary needs and food preferences, and is supported by an environment of adequate sanitation, health services, and care, allowing for a healthy and active life’ by the FAO in 2012(1).
Food security is a multidimensional phenomenon and is covered under four dimensions. These are availability, access, utilisation and stability. Availability is related to the supply of food and means the availability of sufficient quality and variety. Access covers the access to these foods in economic and physical terms. Utilisation is how the body uses nutrients and includes topics such as food preparation or cooking. Stability, however, briefly expresses the continuity of these three dimensions(2).
According to the State of World Food Security and Nutrition report published in 2021, undernourishment increased from 8·4 % to 9·9 % from 2019 to 2020, which means that in 2020 – 161 million more people than in 2019 – an average of 763 million people in the world will face hunger. One in three people (2·37 billion) worldwide did not have access to adequate food in 2020(3). Ending hunger, achieving food security, improving nutrition and promoting sustainable agriculture are also part of sustainable development goals. Therefore, measuring and monitoring food security in valid and reliable ways is critical in tracking progress towards the Sustainable Development Goal (SDG) 2 for zero hunger(Reference Mollier, Seyler and Chotte4,Reference Savary, Waddington and Akter5) .
Many methods exist to measure food insecurity(6,Reference Swindale and Bilinsky7) , including experience-based food insecurity measurement tools(Reference Carlson, Andrews and Bickel8,9) . Experience-based food insecurity scales emerged from the problems in the analysis of food consumption data and the need to address the concept of food insecurity more broadly. Experience-based food insecurity scales offer advantages in that they offer the possibility to directly capture the food insecurity situation and analyse it from a behavioural perspective(Reference Cafiero, Melgar-Quiñonez and Ballard10).
The eighteen-item Household Food Security Survey Module (HFSSM), developed in the USA(Reference Swindale and Bilinsky7), later began to be implemented in surveys conducted for the US Department of Agriculture (USDA) in 1995(Reference McGuire and Sperling11). Apart from HFFSM, there are shorter scales developed specifically for adults and children. The Adult Food Security Survey Module (AFSSM), which consists of ten items, does not provide specific information on children’s food security, but the screening keeps the respondent burden to the minimum needed to get reliable data(12). The US Food Security Survey Modules have been adapted for use in a wide variety of cultural and linguistic settings worldwide(Reference Kleve, Gallegos and Ashby13–Reference Harrison, Stormer and Herman16). When we searched the literature, it was seen that there was no Turkish validity and reliability study of the AFSSM. In addition, we did not find a similar scale that measures food security at the individual level in Turkey. Therefore, due to the importance of the subject, this study was conducted to adapt the AFFSM, a food security scale, to the Turkish language culture and to test its validity and reliability.
Materials and methods
Aim
This study aimed to adapt the AFFSM to the Turkish language culture and evaluate its psychometric properties.
Study design
This study established the AFSSM through a four-step process: (1) testing the validity (content, construct and language) of the scale (2), testing the reliability of the scale (3), exploratory factor analysis (EFA) and (4) confirmatory factor analysis (CFA). Data for the study were collected between May 2022 and June 2022.
The AFSSM was translated from English to Turkish by two experts with proficiency in English and then from Turkish to English by another expert to ensure language validity. The draft form was created by considering the suggestions in the translations. For the content validity of the draft form, the expert opinions of ten lecturers in the Nutrition and Dietetics department and two psychologists were consulted. Davis technique was used for content validity(Reference Grant and Davis17). After expert opinions, the scale was revised based on their comments, and the Turkish version of the scale was created by testing. The flow chart of the adaptation stages is shown in Fig. 1.
Participants
The population of the study consists of university students. Study participants were enrolled using the following inclusion criteria: (a) being 18 years and older, (b) being enrolled as a student at a university, (c) using the Turkish language for communication and (d) volunteering to participate in the study. The exclusion criteria were as follows: (a) international students, (b) those under the age of 18 years and (c) non-university students.
Before the study, the participants were informed about the study, and their consent was obtained. According to the literature, the sample should be at least five or ten times the number of scale items to make factor analysis in validity and reliability studies(Reference Joseph, Barry and Rolph18,Reference DeVellis and Thorpe19) . Therefore, about 100 university students attained by calculating ten times the number of items (ten items) on the scale. For the pilot study, the AFSSM questionnaire was completed and then applied to twenty university students to check intelligibility. For the test–retest, 117 participants completed the AFSSM again 2 weeks after the first administration.
Adult Food Security Survey Module
The Adult Food Security Survey Module consists of ten questions in total. Food security status of each household has been researched in the last 12 months. It is assessed through responses to ten closed-ended questions about food-related behaviours, experiences and conditions, which are known to characterise households that have difficulty meeting their food needs. The total score varies between 0 and 10 and is obtained from the sum of the number of affirmative answers given to the scale questions. Food insecurity is used to refer to both low and very low food security categories(12). The items of the scale are presented in Table 1.
AFSSM, Adult Food Security Survey Module; DK, don’t know.
Ethical considerations
The Ethics Commission at Gazi University approved this study (code: 2022-625/10-05-2022). Before starting the study, an informative text about the study was presented to the participants and the consent of the participants was obtained. The study was conducted by the principles of the Declaration of Helsinki.
Data analysis
In the study, the validity and reliability of the Turkish version of AFSSM were tested. After checking the translation of the scale from English to Turkish, factor analysis was performed for its validity. The suitability of the dataset to the normal distribution was examined using the skewness and kurtosis values, and it was seen that the dataset was suitable for the normal distribution. The Bartlett test was used to decide the suitability of the data for factor analysis and the Kaiser–Meyer–Olkin test was used to test the suitability of the sample size for the factor analysis. EFA was used to analyse the construct validity of the scale. The dataset suitable for factor analysis was analysed using the principal component analysis method and the varimax rotation, which is one of the vertical rotation techniques, assuming that the factors are unrelated to each other based on the theoretical background in line with the recommendations of the literature(Reference DeVellis and Thorpe19–Reference Worthington and Whittaker21). Then, the number of factors of the tested structure and which items load which factors were analysed. CFA was used to analyse the compatibility of the subdimensions with the original scale and test the confirmability of the construct that emerged as a result of the EFA. The scale structure was examined in line with the fit indices suggested by the literature and the criterion values of these indices. The following indices were used to determine the model fit: χ 2 test, comparative fit index (CFI), goodness-of-fit index (GFI), normed fit index (NFI), adjusted goodness-of-fit Index (AGFI), root mean square error of approximation (RMSEA) and standardised root mean square residual (SRMR). The validity analysis was performed with the R program.
To determine the reliability of the scale, item analysis was performed for internal consistency, the reliability coefficient (Cronbach α) was calculated and the item-total test correlations of the items were examined. Factor scores were interpreted with analysed Cronbach’s α values. The reliability coefficient is between 0 and +1, and reliability was considered to be acceptable for 0·7 ≤ Cronbach’s α < 0·8, good for 0·8 ≤ Cronbach’s α < 0·9 and excellent for Cronbach’s α ≥ 0·9. Values close to 1 indicate that the reliability and internal consistency between the items are high(Reference Davis22–Reference Kline24).
Mean and standard deviation values were given for descriptive statistics related to the demographic characteristics. SPSS 24.0 program was used in the evaluation of the data, and the level of significance was accepted as P < 0·05. The test–retest reliability of the scale was re-evaluated 2 weeks later.
Results
In this study, 117 university students (10 males and 107 females) participated in the construct validity of AFSSM. Students studying at Gazi University, Department of Nutrition and Dietetics, were included in the study. The baseline characteristics of the participants are presented in Table 2. The mean age of students was 22·7 ± 4·1 years.
₺: Turkish Lira.
Participants of 71·8 % had average body weight, and their mean BMI was 22·15 ± 3·58 kg/m2. Participants of 70·9 % were residing in the city, 57·3 % of their fathers working and their mothers not working. Participants of 4·3 % are working and, the income means were 1711·97 ± 1815·30 Turkish Lira (₺). Their food expenditures were 777·86 ± 880·57 ₺, and food expenditures/income ratio means were found 53·2 %.
Validity
Content validity
For a content validity evaluation, a lecturer in the Nutrition and Dietetics department experienced scale development, and two psychologists were asked to comment on the grammar, wording and scoring of the AFSSM items with the Davis technique(Reference Grant and Davis17). The scale was revised based on their comments.
Explanatory factor analysis
Bartlett’s test value was 450·653 (P = <0·001), and Kaiser–Mayer–Olkin test value was 0·686. Two items (I) (I7, I8) that the item load value is <0·50 and have a low factor load value were excluded from the scale. After two questions were removed from the scale, the process was repeated on the remaining eight questions. As a result of EFA, three factors were obtained with factor loadings below 0·50 and eigenvalues above 1. The eigenvalues of these three factors were 3·99, 1·54 and 1·25, respectively, and they explained 77·4 % of the total variance of the AFSSM (Table 3).
AFSSM, Adult Food Security Survey Module.
The resulting factors were named ‘Factor-1: Inadequate Nutrition’, ‘Factor-2: Economic Concern’ and ‘Factor-3: Hunger’. The factor loadings, eigenvalues and explained percentages of variance of the items constituting these factors are given in Table 4. It is seen that the ‘Factor-1’ dimension consists of ‘I4, I5, I6’, ‘Factor-2’ dimension consists of ‘I1, I2, I3’ and ‘Factor-3’ dimension consists of ‘I9, I10’ items. The factor load, which shows the relationship of each item with the total score, was over 0·70, and three factors accounted for 65 % of the variance.
AFSSM, Adult Food Security Survey Module; F, factor; I, item.
Based on the factor levels obtained as a result of the EFA, the AFSSM has a mixed 3, 4, and 5 Likert types, eight-item, and three-factor structure.
Confirmatory factor analysis
In this study, before starting the CFA, it was checked whether the data showed normal distribution and whether there were any empty cells or missing data in the database. The covariance matrix was produced, and analyses were made since the data followed the normal distribution. The process of generating the matrix was done by the maximum likelihood estimation method by the R program. Then, the fit indexes of the model were evaluated(Reference Kline25).
In the model, it was observed that the criterion values were met and the model created is given in Fig. 1. In addition, since there was no item with standardised factor loadings below 0·5, no items were removed from the scale. The fit of the final model obtained after this process was tested with fit indices such as χ 2/s d, AGFI, CFI, GFI, NFI, RMSEA and SRMR(Reference Browne and Cudeck26). The results of the CFA and GFI indices are shown in Table 5.
AFSSM, Adult Food Security Survey Module.
As a result of the analysis, the GFI indices were as follows: χ 2/s d = 0·235, AGFI = 0·986, CFI = 0·992, GFI = 0·994, NFI = 0·986, RMSEA = 0·034 and SRMR = 0·041.
Based on the data obtained, it was concluded that the model fit values were at an acceptable level, and since most of the fit indices showed a high degree of fit, no change was made in the model structure. The CFA model of AFSSM is presented in Fig. 2.
Convergent validity
Convergent validity was used for the construct validity of the scale. Convergent validity means that statements about variables are related to each other and to the factors they generate. The composite reliability coefficient (composite reliability-CR) is used to evaluate convergent validity. The average variance extracted (AVE) value is the convergent validity between the items that make up a factor. For convergent validity, all CR values for the scale are expected to be greater than the AVE values and the AVE value to be greater than 0·5. In addition, the standardised factor loads of the items should be above 0·5 and the CR value should be higher than 0·7. The AVE value is obtained by dividing the sum of the squares of the item loads of the factor by the number of items(Reference Hair, Black and Babin27,Reference Bagozzi and Yi28) .
As a result of CFA, when convergent validity is examined (Table 6), it is seen that the standardised factor loads of the items are between 0·75 and 0·95; the CR and Cronbach’s α values are higher than 0·7 and the AVE values are higher than 0·5. Therefore, all three constructs have convergent validity. These results indicate that the scale used in the research has convergent validity.
SFL, standardised factor loading; CR, composite reliability; AVE, average variance extracted; I, item.
* SFL (λ).
† CR.
‡ AVE.
Reliability
Cronbach’s α coefficient – internal consistency
Internal consistency is based on the assumption that items measuring the same construct should be related to each other, and the most commonly used statistic to evaluate internal consistency is the Cronbach’s α coefficient. Cronbach’s α coefficients are between 0 and 1. A coefficient close to 1 means the scale is reliable and reflects better internal consistency(Reference Kimberlin and Winterstein29,Reference Polit and Beck30) .
In this study, Cronbach’s α coefficient was determined as a result of the items (eight items) obtained after the EFA was found to be 0·769, supporting the scale reliability. Cronbach’s α values for ‘inadequate nutrition’, ‘economic concern’ and ‘hunger’ were found to be 0·761, 0·742 and 0·750, respectively, and ranged from 0·72 to 0·77 for the three dimensions (Table 7).
AFSSM, Adult Food Security Survey Module.
Test–Retest
When the scale is applied to the same individuals at different times, getting similar answers indicates the stability of the scale. In this context, test–retest reliability is the most common method used to evaluate the stability of a scale(Reference Weir31). To calculate the test–retest reliability of the scale, the questionnaire was applied again 2 weeks after the first application for the retest application. When the test–retest analysis correlations of the AFSSM dimension and total scores were examined, it was found that all dimensions and total scores of the scale showed a statistically significant positive correlation (P < 0·01; P < 0·001). When the test–retest results were examined, it was concluded that the scale was reliable and test–retest reliability was acceptable. The correlation coefficients are given in Table 8.
AFSSM, Adult Food Security Survey Module; F, factor; r, Pearson’s correlation coefficients.
* P < 0.01.
** P < 0.001.
Scoring
The eight items of the AFSSM were scored on a 3, 4 and 5-Likert scale. The highest score of each question is 6, and the highest score of the scale is 48. The total score of the scale is between 0 and 48. The cut-off point of the scale is 10. The cut-off point of the scale was determined as a result of the Receiver Operating Characteristics (ROC) curve analysis. (0–10 = have food security and 11–48 = have no food security). The ROC curve analysis is presented in Table 9.
SEN, sensitive; SPC, specificity.
* P < 0.05.
Discussion
In the study, the validity and reliability of the adaptation of AFFSM to the Turkish language and culture were tested. A total of 117 individuals were included in this study.
The AFFSM(Reference DeVellis and Thorpe19) is a food security measurement tool used around the world by adapting it to various languages. The results of the scale adaptation studies have shown that the scale has good internal validity, and its use in different regions is reliable(Reference Kleve, Gallegos and Ashby13–Reference Harrison, Stormer and Herman16).
The validity and reliability study of the short version of the Household Food Security Survey Module (HFFSM-SF)(12) for Turkey was conducted by Emiral et al. in 2017(Reference Emiral, Önsüz and Metintaş32). However, based on our literature research, we found no scale measuring food security at the individual level in Turkey. Therefore, we tested the validity and reliability of the AFSSM by adapting it to the Turkish language and culture.
The scale includes differences according to the US methodology. According to the EFA, it was seen that the scale had a three-factor structure. One of the factors focuses on the relationship between food insecurity and inadequate nutrition, and the other focuses on its relationship with hunger. This distinction can help us learn more about individuals’ coping strategies for food insecurity. The other factor highlights the impact of economic concern and poverty on food access.
The original scale followed a categorisation method (very low food security, low food security and marginal food security) according to the presence and severity of food insecurity by scoring only on affirmative options. The US version calculates food insecurity based on yes questions only; this new scale improves sensitivity within frequencies and gives different scores to ‘yes’, ‘often’, ‘sometimes’, ‘almost every month’ and ‘some months but not every month’ answers. However, in some reports, it is reported that they are divided into two separate categories and defined as food-secure and food-insecure(12). As in the results of the study of Wehler et al.(Reference Wehler, Scott and Anderson33), the first study in which categories were discussed after Radimer et al.(Reference Radimer, Olson and Campbell34), we also similarly found that the answers were not compatible for more than one category according to income status. Therefore, we propose, presenting the results in two categories, food-secure and food-insecure.
We are aware that the answers do not express results that are equidistant from each other. However, since we think that not every affirmative answer is equally important, we have adopted a scoring system based on the predominance of points for questions that have frequency options (Q1, Q2, Q3, Q5 and Q10). Our scale score range (0–48) thus differed compared with the original scale’s score range (0–10). However, thanks to this situation, we hope that more precise and sensitive results can be obtained in studies to be carried out, especially among samples with similar food security degrees.
Another difference between these two scales stems from the number of questions. Although the original scale consisted of ten questions, this adapted new scale consisted of eight questions. One of the questions with a low item load was ‘In the last 12 months, did you lose weight because there wasn’t enough money for food?’. We think that the reason why this problem has a low factor load is that the relationship between food security and body weight is still controversial(Reference Farrell, Thow and Abimbola35–Reference Dhurandhar38). Although the item load was not very low (0·491), the other question we asked was ‘In the last 12 months, were you ever hungry but didn’t eat because there wasn’t enough money for food’. We chose to exclude this question to obtain more consistent results due to the significant difference between the other item loads.
We could not perform a reference validity analysis because there is no other parallel scale measuring food security at the individual level in our country. However, this scale will allow future studies to perform this test. In addition, the population participating in the survey represents a limited population section regarding age and education level. Although we think there is ethnic diversity because university students are from various regions of the country, we recommend that the scale be re-evaluated for older and/or less educated people.
Conclusion
The results of the study show that the AFFSM, which we have adapted to the Turkish language and culture, is valid and reliable. Thus, we brought the first Turkish Adult Food Security Scale, which measures food security individually, to the literature. We hope this scale will be a qualified tool to determine food insecurity in adults living in Turkey and reveal the impact of food insecurity on other parameters. Finally, we hope to support further studies to identify food security or conditions associated with food security.
Limitations
Among the limitations of this study is that the participants included in the study are students of the Nutrition and Dietetics departments.
Acknowledgements
The authors would like to thank the study participants for taking the survey and the Gazi University Academic Writing Application and Research Center for proofreading the article.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Conflict of interest
There are no conflicts of interest.
Authorship
YA, BSK and MD conceptualised the study. YA, BSK and MD undertook the analysis. All authors contributed to the interpretation of the results. YA, BSK and MD led the writing of the manuscript, with contributions from all authors.
Ethics of human subject participation
This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the The Ethics Commission at Gazi University (code: 2022-625/10-05-2022). Written informed consent was obtained from all subjects/patients.