Skip to main content

Development, validation and item reduction of a food literacy questionnaire (IFLQ-19) with Australian adults

Abstract

Background

Food literacy is theorised to improve diet quality, nutrition behaviours, social connectedness and food security. The definition and conceptualisation by Vidgen & Gallegos, consisting of 11 theoretical components within the four domains of planning and managing, selecting, preparing and eating, is currently the most highly cited framework. However, a valid and reliable questionnaire is needed to comprehensively measure this conceptualisation. Therefore, this study draws on existing item pools to develop a comprehensive food literacy questionnaire using item response theory.

Methods

Five hundred Australian adults were recruited in Study 1 to refine a food literacy item pool using principal component analysis (PCA) and item response theory (IRT) which involved detailed item analysis on targeting, responsiveness, validity and reliability. Another 500 participants were recruited in Study 2 to replicate item analysis on validity and reliability on the refined item pool, and 250 of these participants re-completed the food literacy questionnaire to determine its test–retest reliability.

Results

The PCA saw the 171-item pool reduced to 100-items across 19 statistical components of food literacy. After the thresholds of 26 items were combined, responses to the food literacy questionnaire had ordered thresholds (targeting), acceptable item locations (< -0.01 to + 1.53) and appropriateness of the measurement model (n = 92% expected responses) (responsiveness), met outfit mean-squares MSQ (0.48—1.42) (validity) and had high person, item separation (> 0.99) and test–retest (ICC 2,1 0.55–0.88) scores (reliability).

Conclusions

We developed a 100-item food literacy questionnaire, the IFLQ-19 to comprehensively address the Vidgen & Gallegos theoretical domains and components with good targeting, responsiveness, reliability and validity in a diverse sample of Australian adults.

Background

Vidgen & Gallegos [1] define food literacy as “… a collection of inter-related knowledge, skills and behaviours required to plan, manage, select, prepare and eat food to meet needs and determine intake” and the “… scaffolding that empowers individuals, households, communities or nations to protect diet quality through change and strengthen dietary resilience over time.” Their conceptualisation consists of 11 components organised under four inter-related domains of planning and managing, selecting, preparing and eating (see Fig. 1).

Fig. 1
figure 1

Domains and components of food literacy by Vidgen & Gallegos [1, 2]

Recently, an international consensus study [3] and a scoping review of 549 publications on food literacy [4] found that there is agreement internationally with the Vidgen & Gallegos [1] definition and conceptualisation of food literacy. This definition is the first to be empirically derived, and is proposed by researchers as the most current, predominant approach to food literacy that is able to significantly advance the field [5,6,7].

While food literacy surveys have been developed [8,9,10,11,12,13,14,15,16,17,18,19], these tend to be the result of expert consensus on food literacy rather than consultation with the general public. However, the Vidgen & Gallegos [1] conceptualisation was derived from an analysis of knowledge, skills and behaviours used by the general public to protect their diet quality through change. Where papers stated use of the Vidgen & Gallegos model [1] to guide tool development, analysis of these tools indicated that not all components in the model are measured; and thus, the domains or construct of food literacy are not comprehensively captured [20]. Being able to discernibly measure each component is likely to be more useful in guiding policy and practice than domain or overall scores which mask important differences across the spectrum of food literacy between individuals.

In order to address the limitations with existing surveys, work has been conducted by Vidgen and colleagues since 2014 to develop a comprehensive food literacy measure. This has included a four-part content validity study involving, 1) Expert feedback on the construct of food literacy and its components [2]; 2) A review of these constructs against interventions [2]; 3) Young people’s feedback on the construct [2] and; 4) An international item pool consensus study [3, 20]. Questionnaire development was further informed by reflections on what a measure needs to encompass [21, 22]. A face validity study using cognitive interviews was then conducted with a purposefully selected sample of the general public that was representative of demographic characteristics identified in earlier research [3, 23]. This series of studies generated a 171-item pool that reflected the 11 components proposed by Vidgen & Gallegos [1] (see Fig. 1). However, the reliability, validity and consistency of this item pool had not been assessed.

In existing food literacy surveys, psychometric properties are typically assessed using Classical Test Theory (CTT) methods. While CTT is widely used [24], it carries a number of assumptions that limit scale development and generalisability of findings [25,26,27].

Item Response Theory (IRT) offers one alternative to CTT. It has been defined as, “…a mathematical model that relates a test-taker’s latent trait or ability score with the probability of responding in a specific response category of an item” [28]. It consists of a family of models which are not bound by the same assumptions as CTT. In particular, IRT produces item-level information rather than scale-level information which overcomes a major limitation imposed by CTT. This is the assumption that all items contribute equally to the total questionnaire score; thus IRT is more nuanced, allowing individuals to be higher on one construct and lower on another [25]. IRT is increasingly used in public health research [29,30,31,32,33,34] and more specifically, in public health nutrition research [35, 36].

The purpose of this paper is to describe methods used to develop a food literacy questionnaire which comprehensively measures the four domains and 11 components of food literacy. This involved detailed item analysis on, 1) targeting: the extent to which the distribution in the sample matches the range measured by the scale [37]; 2) responsiveness: the ability of an instrument to detect change accurately when it has occurred [38]; 3) reliability: the extent to which results obtained by a scale can be replicated and scores are free from random error [39]; and 4) validity: the extent to which a scale measures what is intended, and is free of systematic error [3, 23, 39]. This aimed to develop a refined item pool (Study 1), replicate item analysis on validity and reliability on the refined item pool from Study 1 (Study 2); and determine test–retest reliability of the refined item pool from Study 1 (Study 3).

Methods

Study 1

Study participants and recruitment

This study purposefully sampled a diverse range of residents of Australia, over 18 years of age: participant quotas can be seen in Table 1 based on Australian census data [40, 41]. Participants were recruited via double-opt-in market research panels through Qualtrics [42] and approached via email, in-app or SMS notifications to participate in an online survey between the 25th November 2020 and 14th December 2020. A sample size of 500 participants achieved the upper boundary of recommended sample sizes [43,44,45]. If participants were less than 18 years of age or did not agree to participate, they were screened out.

Table 1 Demographic characteristics of study participants

Item pool

The 171-item pool was developed from previous research [3, 23] as described above. Each of the 11 components in the Vidgen & Gallegos food literacy model [1] were represented by 10–40 items (see Appendix 1). All items were on a 5-point likert scale of either: strongly disagree to strongly agree, never to always, not at all important to extremely important, not knowledgeable at all to extremely knowledgeable, or not at all sure to very sure. Twenty-seven items were reverse scored.

Data collection

Participants were asked to complete the item pool, which Qualtrics estimated would take 29-min. Participants were able to track their progress via a progress bar and to exit and re-enter the questionnaire and continue at any time prior to participant quotas being met. On completion of the food literacy questionnaire, the following demographic data (12 questions) was collected in addition to that obtained during screening: sex, ancestry, postcode, highest level of education obtained, employment status, income, number of people in their household (including children), who is primarily responsible for meal preparation and cooking in the household and how many times per week meals are prepared and cooked at home (see Table 1). Participants received compensation in the form of cash, vouchers or points through Qualtrics in accordance with the length of the questionnaire.

Data analysis

Participant responses were downloaded into an Excel spreadsheet [48] and data cleaning was undertaken to remove incomplete respondents. Participants who completed all items in the questionnaire were used in the analysis. Demographic information was summarised using descriptive statistics in Excel, and postcodes were analysed to determine socio-economic advantage or disadvantage using SEIFA [40] and level of remoteness [47].

Checking assumptions of methods

Statistical analyses were undertaken using R Studio, version 1.4.1717 and SPSS, version 27.0.1.0. As the likert-scales all contained five ordinal response categories, and one item had six response categories, the Partial Credit Model (PCM), a type of Item Response Theory (IRT) was chosen for this analysis.

Prior to running the analysis, the assumption of unidimensionality (that participants responses are based on the underlying concept, not previous items) needs to be met [49]. A Principal Components Analysis (PCA) was run using varimax rotation on each of the 11 components of food literacy in SPSS, version 27.0.1.0. First, sampling adequacy and the data’s suitability for reduction were assessed, where a Kaiser-Meyer Olkin (KMO) score of > 0.5 [50] and a significant Bartlett’s test confirmed a PCA was able to be appropriately conducted on the data. The eigenvalues, scree plot and rotated component matrix from the PCA were then assessed for the number of principal components and loadings. To determine a clear principal component structure, items were removed one at a time starting with the item with the lowest loading until all items loading < 0.32 were removed [51]. Then, items were removed one at a time from the highest cross loading until all items loading > 75% were removed [51]. A total variance explained of > 50% indicated adequate principal component structure [50]. Once a clear structure was developed, C.T. reviewed the principal components against the food literacy components and developed a questionnaire structure. Where one principal component did not comprehensively assess the theoretical food literacy component, the second or third principal component was included. Final decisions were discussed with H.A.V. and R.B. until agreement was reached. For food literacy components in which the PCA identified more than one principal component, these were separated and labelled as ‘statistical components’ which were taken forward for future analysis.

The PCM was then run for each statistical component following methods proposed by Wright & Masters [52], using the eRm package [53] and script adapted for R from Wind & Hua [54]. These methods are standard procedures when conducting IRT analysis, with further information available elsewhere [28, 29, 49, 55, 56].

The resultant PCM outputs were reviewed, where thresholds for the statistical tests are described below. However, it should be noted that these thresholds provide guidance for researcher judgements and are not strict criteria.

Assessment of targeting

Item thresholds were reviewed to determine if response categories were working as intended. Item thresholds should be sequentially ordered from most negative to most positive values. If they are not ordered, this suggests that participants are not reliability discriminating between response categories [57]. In these cases, disordered response categories were combined with subsequent response categories to meet ordering requirements [58]. This relationship was graphically described using Category Probability Curves (CPCs), with the width of a threshold indicative of the probability of the category response being chosen [59, 60].

To determine if items mapped out a discernible line of increasing intensity across the continuum of food literacy, item locations, the overall item difficulty were extracted [58]. Item locations should span equally from -2 to + 2 [55]; locations < -2 were considered too easy and >  + 2 too difficult and were removed. Further, item locations on the same logit (the item difficulty estimate determined using a simple logistic function) [55, 56] were also considered for removal as they are assessing the same level of food literacy. This was determined using scale-item maps [61].

Assessment of responsiveness

To determine if the food literacy items adequately represented respondent’s levels of food literacy, the range of item locations, thresholds and person locations (the person ability estimates) were compared [62]. Item locations and thresholds distributed within ± 1 logits of the person locations were considered consistent with expectations. Floor effects (< -1 logits of the item thresholds) and ceiling effects (> + 1 logits of the item thresholds) were assessed. The upper limit for the proportion of participants with floor or ceiling effects is 15% [63] and 20% respectively [64, 65]. Items with high floor or ceiling effects were considered for removal. This relationship was visualised using person-item maps.

Assessment of reliability

Person separation reliability were extracted to determine if the scale was sensitive in distinguishing between low and high performers [52]. Values of > 0.70 were considered as able to discriminate among people with differing levels of food literacy [56]. Item separation reliability were extracted to determine if the person sample was large enough to confirm item difficulty [62, 66]. Values of > 0.70 are indicative of high item separation reliability [56].

Assessment of validity

Item fit was analysed to determine if items appropriately measured the food literacy model. Outfit mean-squares (MSQs) between 0.5 and 1.5 were considered indicative of items that were productive for measurement. Values > 2.0 were considered to degrade the measurement and were considered for removal [56, 67].

Person fit was analysed using Z-statistics to identify atypical response patterns, such as participants who randomly selected responses, exaggerated or responded in a way that fluctuated across items [68]. A Zh value of < -2 indicates misfitting, where respondents typically select extreme responses, and a value of >  + 2 indicates overfitting, where respondents typically select the middle response: all values in between indicated well-fitting respondents [68]. Items with a high proportion of misfitting or overfitting were considered for removal.

Overall, decisions on whether items should be retained or removed were made with referral back to cognitive interview data [23] to ensure the voice of the general population was incorporated and appropriately reflected in the development of this questionnaire.

Study 2

Study participants and recruitment

Participant sampling, recruitment, screening and demographics were collected as per study 1, with 500 participants recruited between the 10th October 2021 and 17th October 2021. Participants involved in study 1 could not participate in study 2.

Item pool

The 100-item pool used in this study was taken directly from the IRT analysis conducted in Study 1. The allocation of the 100-items to each domain and component can be seen in Column A, Appendix 2. All items had response options on 5-point likert scales as described in Study 1, and eight items were reverse scored.

Data collection

Participants were asked to complete the 20-min item pool. Progress tracking, questionnaire time completion and compensation were the same as per Study 1.

Data analysis

Reliability and validity data extraction and analysis were repeated as per Study 1.

Study 3

Study participants and recruitment

All participants from Study 2 were approached to re-complete the food literacy questionnaire via the recruitment methods described for Study 1. The survey was closed to respondents once 250 participants were recruited. Participants were recruited [69] between the 1st November 2021 and 4th November 2021, two weeks after the study 2 data collection period closed. Participants completed the same screening items and demographic questions as in Study 1.

Item pool

The item pool described in Study 2 was used in this study.

Data collection

Participants were asked to complete the 20-min item pool. Progress tracking, questionnaire time completion and compensation were the same as per Study 1.

Data analysis

Test–retest reliability was assessed using intraclass correlation coefficients (ICC) using a two-way mixed effect with absolute agreement. The average rater ICC value was reported, and the interpretation of ICC values was as per Ratner, 2009 [70].

Calculation of food literacy scores

Raw scores, the sum of each response category in a set of items, are reported to inflate the extent of change reported by participants across administrations of a questionnaire due to the equal scoring of items. To address this, the Rasch method of scoring was conducted as it provides a more accurate measurement along the continuum of the underlying construct [71,72,73].

Raw score-to-measure tables were developed for each statistical component of the food literacy questionnaire. Raw scores represent the sum of participant responses to the likert scale (1–5) for the total number of items included within that component and correspond to a Rasch logit. These were calculated using Winsteps, version 5.2.4.0, following methods proposed by Linacre [74]. Rasch logits were transformed into “user friendly rescaling” such that all statistical components are assessed on a scale of 10–100 using methods proposed by Linacre [75]. Ten corresponds to the minimum possible statistical component score, while 100 corresponds to the maximum possible statistical component score.

Ethics

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 University Human Research Ethics Committee (UHREC) at QUT, Approval Number: 2000000004. Written informed consent was obtained from all participants.

Results

Study 1

Participant characteristics

The survey was opened 931 times. Forty-five respondents were screened out due to age (5%) and 48 did not agree to participate (5%). Six respondents did not provide their postcode (1%) and 328 did not complete the questionnaire (35%). Overall, 504 participants provided full data and were included in the analysis. The demographic characteristics of participants are reported in Table 1 and are reflective of the Australian population with relation to gender, education and state of residence [76] and fairly representative for age and annual individual income.

Principal Component Analysis (PCA)

Of the 11 theoretical components of food literacy, the assumption of unidimensionality was met for components 1.2, 2.1, 2.3, 3.1, 4.3. The remaining six theoretical components were split into two (1.1, 2.2, 3.2, 4.1) or three (1.3, 4.2) statistical food literacy components. This resulted in a total of 19 statistical components which are nested within the theoretical components and domains food literacy (see Table 2). Seventy-one items were removed as they were low loading, cross loading or did not fit within a clear principal components structure. The division of each item under the theoretical and statistical components of food literacy can be seen in Appendix 1. The KMO’s were > 0.5 and Bartlett’s test was significant for all statistical components of food literacy. The total variance explained by the statistical components was > 50% for 15 of the 19 statistical components.

Table 2 Assessment of unidimensionality using PCA and the resulting theoretical and statistical food literacy components

Item Response Theory (IRT)

Assessment of targeting

Across the 19 PCMs, 26 items were identified as having disordered thresholds and are shown in Column C, Appendix 2. Threshold 2 (t2) was the most frequently disordered (n = 24, 92%). To resolve this, ‘disagree’ and ‘neutral’ responses were combined. After this, all thresholds were ordered (see Column E–H, Appendix 2) and CPCs demonstrated acceptable patterns. Thresholds for the 100-item food literacy questionnaire ranged from -3.69 to + 6.11.

Assessment of responsiveness

Item locations ranged from < -0.01 to + 1.53 logits, within the recommended range of -2 to + 2 (see Column D, Appendix 2). Six of the 19 statistical components had items on the same logit level, with four of these reporting two items on the same logit (1.1.1, 1.2, 2.2.2, 4.1.1) and two reporting three items on the same logit (2.1, 4.2.1).

Floor effects were reported for 14 of the 19 statistical components of food literacy. These were between 0–3%, all below the upper limit of 15% (see Column K-L, Appendix 2). Ceiling effects were reported for all statistical components of food literacy (3–20%). All values were below 20%, with sub-component 3.2.1 at the upper limit. On average, 38 participants per sub-component of food literacy reported values that were outside of ± 1 logit from the lower and upper item threshold ranges; meaning that around 92% of participants gave responses consistent with expectations.

Assessment of reliability

The person separation reliability and item separation reliability for the statistical components of food literacy were high at > 0.99 for all components (see Column M, N, Appendix 2).

Assessment of validity

Outfit MSQs for the 100-item food literacy questionnaire ranged from 0.48 to 1.42 (see Column Q, Appendix 2). Ninety-nine items had MSQs between 0.5 to 1.5, while one item, (6_3.2) was just below the MSQ criteria, with an item fit of 0.48.

Person fit statistics are shown in Column P-R, Appendix 2. Between 11–41 participants had responses that were considered misfitting in the food literacy questionnaire, with sub-component 3.1 reporting the highest level of misfit (n = 41, 8%). Participant responses which overfit the food literacy questionnaire were only reported for component 4.2.1 (n = 14, 3%).

Study 2

Participant characteristics

The survey was opened 830 times. Sixteen respondents were screened out due to age (2%) and 15 did not agree to participate (2%). Two respondents did not provide their postcode (0.2%) and 294 only partially completed the questionnaire (35%). Overall, 503 participants provided complete data and were included in the analysis. The demographic characteristics of participants are summarised under Study 2 in Table 1. There were minimal differences in participant demographics between study 1 and study 2 administrations.

Item Response Theory (IRT)

Assessment of reliability

As in Study 1, the person separation reliability and item separation reliability for the statistical components of food literacy were high at > 0.99 (see Column S-T, Appendix 2).

Assessment of validity

Outfit MSQs for the 100-item food literacy questionnaire ranged from 0.518 – 1.362 (see Column Z, Appendix 2). All items had MSQs between 0.5 to 1.5.

Person fit statistics are shown in Column AB-AD, Appendix 2. Between 12–45 participants had responses that were considered misfitting in the food literacy questionnaire, with sub-component 2.1 reporting the highest level of misfit (n = 45, 9%). As before, participant responses which overfit the food literacy questionnaire were only reported for component 4.2.1 (n = 23, 4%).

Study 3

Participant characteristics

Of the 503 respondents from Study 2, 269 completed the food literacy questionnaire a second time and were included in the analysis. The demographic characteristics of participants are summarised in Table 1. There were no important differences in demographics between study 2 and study 3 participants.

Test–retest reliability

The intraclass correlation coefficients for the statistical components of food literacy are reported in Column AE, Appendix 2. Moderate reliability was reported for five statistical components of food literacy (1.1.2, 2.1, 2.3, 4.1.2, 4.2.2), while good reliability was reported for the remaining 14 statistical components.

Calculation of food literacy scores

The raw score-to-measure table can be seen in Appendix 3. All possible raw scores for each statistical component are listed in Column B, Rasch logits are listed in Column C and the user-friendly re-scaled scores are in Column D.

How to score the IFLQ-19

The International Food Literacy Questionnaire reflects the 19 statistical components of food literacy and thus, is called the IFLQ-19. The methods for scoring are seen in Table 3 below. The IFLQ-19 produces 19 separate scores for each of the statistical components.

Table 3 How to score each component of the IFLQ-19
Table 4 The food literacy questionnaire (IFLQ-19)

Discussion

This study described the methods used to develop a 100-item food literacy questionnaire, the IFLQ-19. This questionnaire comprehensively reflects the 11 theoretical components of food literacy by Vidgen & Gallegos [1], which gives valid, reliable and consistent results.

PCA

The principal components analysis was integral in determining the structure of the food literacy questionnaire. While the components of food literacy were theorised to be unidimensional, findings from this research rejected this theory in six cases. Components which had more comprehensive descriptions generally addressing multiple interrelated points (see Fig. 1) were often split, for example component 1.1, ‘Prioritise money and time for food’ was split into sub-component 1.1.1 addressing the ‘time’ aspect, while sub-component 1.1.2 addressed the context of ‘food’. The 19 statistical components resulting from the PCA were retained for the following reasons: 1) This research prioritised the voice of the general population obtained through cognitive interviews [23]. 2) Comprehensive measurement of the theoretical components required the statistical components to be retained. 3) Interventions targeting different aspects of food literacy can choose the appropriate statistical component to use in their evaluation. While it is theorised that the statistical components sit within the 11 theoretical components proposed by Vidgen & Gallegos, further statistical testing using a Multidimensional PCM (MPCM) is needed to verify this structure. The total variance explained was between 46–49% for four of the statistical components (1.1.2, 1.3.3, 2.2.2, 4.3). As this was just below the threshold (< 50%) and to ensure Vidgen & Gallegos [1] model was comprehensively addressed, these four statistical components were retained.

Targeting and responsiveness

Targeting and responsiveness were assessed using item thresholds, category probability curves, item locations and person-item maps. In the initial analysis, 26 item thresholds were not correctly ordered. The t2 category was most often disordered, meaning respondents were unable to reliably discriminate between the ‘disagree’ and ‘neutral’ categories. This can occur for two reasons: 1) there were too many response categories, and the ‘neutral’ option should be removed, and 2) there may be unequal category responses [56]. While the former is more difficult to confirm, the latter was identified in this situation, where respondents tended to select the upper categories (often/always, agree/strongly agree) compared to the lower options across the 26-items. The scale chosen for items, such as ‘strongly disagree’ to ‘strongly agree’ instead of ‘never’ to ‘always’ was also considered as a potential issue in the selection of response categories. However, this was not identified in cognitive interviews with participants on these items [23], suggesting that if there was an impact, it would have been minimal. Overall, disordered thresholds were low (26%) and category combining at the analysis phase resulted in ordered thresholds for all items; thus, response categories were working as intended and respondents were distributed across lower to higher food literacy levels [77]. No items were deleted, as they were all considered to be critical in assessing the statistical component of food literacy.

Item locations were all acceptable, though tended to cluster in the mid to higher end of the recommended range, from + 0 to + 1.5 logits. This suggests that while item locations were adequate, additional items that more equally spread across the -2 to + 2 range would have ensured differentiation between participants and their food literacy characteristics could be better captured [78]. Multiple items sat on the same logit across six statistical components, suggesting that items were assessing the same level of food literacy. However, all items were retained due to previous qualitative feedback [23] on ensuring the components of food literacy were comprehensively addressed.

Person-item maps identified that fewer respondents obtained the minimum possible score, meaning floor effects were negligible [79]. In contrast, at least 13 respondents obtained the highest possible score for each sub-section of the food literacy questionnaire indicating mild ceiling effects [79]. This suggests there may be a slight bias toward respondents with higher food literacy levels [80]. However, all values were below the upper limit for floor and ceiling effects, suggesting the food literacy questionnaire is an appropriate measurement model [81]. Overall, targeting was acceptable and responsiveness of the food literacy questionnaire was supported.

Validity

Validity was assessed using MSQs from item fit analysis and Zh statistics from person-fit analysis. In study 1, only one item had an MSQ value just below the 0.5 to 1.5 recommended criteria. While items with MSQs < 0.5 are considered to be less productive in measurement, considering item 6_3.2 reported an MSQ of 0.48, the impact on the overall questionnaire would have been negligible. In study 2, all MSQ values were within the 0.5 to 1.5 range. This suggests that the food literacy questionnaire measures what was intended, and that all items work well together and are appropriate for combined scoring [39].

With relation to person-fit, Z-statistics were used to determine misfitting and overfitting respondents. In study 1 and 2, respondents classified as ‘overfitting’ were only reported for sub-component 4.2.1, where respondents across this section predominantly selected 2, ‘somewhat sure’ and 3 ‘neutral’. All items in this section related to knowing food group recommendations from the Australian Dietary Guidelines [82] and Australian Guide to Healthy Eating [83]. If someone was unable to answer the first item, it was unlikely they responded in the extremes for similar items, thus explaining the higher number of ‘neutral’ responses for items in this sub-section. In study 1 there were between 11–41 and in study 2 between 12–45 respondents who were considered ‘misfitting’. A review of participants who were classified as misfitting, identified they tended to select 5, ‘strongly agree/always’ and 4 ‘agree/often’ responses rather than those at the lower end of the scale. This is unsurprising, considering findings from the ‘targeting and responsiveness’ statistics which suggest there was a mild ceiling effect in the questionnaire. Further, misfit differences across administrations is common due to latent trait distribution and item parameter drift [84, 85]. While misfitting and overfitting items may impact the validity of a questionnaire, this does not appear to be the case here [86]. Overall, validity for the food literacy questionnaire was high.

Reliability

Person separation and item separation reliability were very high in study 1 and 2 for all statistical components of food literacy. This suggests that the questionnaire was sensitive in distinguishing between low and high performers and that the sample was large enough to locate items on the latent variables [39, 52]. The ICCs for the statistical components were moderate to good, though none reported excellent ICCs. This may be due to some variability in participants circumstances (e.g. they may not run low on money or shop/eat at new restaurants) or interpretation on types of foods. Further, it may also be a result of the mild ceiling effects described above, whereby a less equal distribution of respondent scores has been reported to be associated with lower ICCs [87]. However, no statistical components reported ‘poor’ ICCs and thus, the questionnaire is still considered to have a good level of reliability.

Scoring the IFLQ-19

Re-scaled scoring was chosen as raw scores (grounded in CTT), assumes the distance between response options is the same. However, IRT methods, and in particular, our research found this was not the case. Therefore, re-scaled scoring provided a more precise indication of a person’s level on the statistical food literacy components and can be used to assess magnitude of change between questionnaire administrations and the relationship between other variables. However, this does increase the complexity of scoring. In acknowledgement of this, the authors developed a scoring guide for future users of the tool. This conversion table (Appendix 3) means practitioners do not need to conduct Rasch analysis every time to obtain Rasch transformed score and can refer to this table for ease of scoring.

Strengths and limitations

The strengths of this research include the rigorous and comprehensive development of the questionnaire and adherence to the Vidgen & Gallegos model [1]. The IFLQ-19 developed in this study is the result of several comprehensive, multi-step research studies over the course of several years. This included international consultation and feedback on the initial pool of food literacy items which was reviewed with general public and validated in a diverse sample of Australian adults [3, 23]. The international perspective and the use of IRT as a statistical method is a distinct advantage over previous approaches [8,9,10,11,12,13,14,15,16,17,18,19]. As IRT is sample-independent [88, 89], the questionnaire validated from this study can be re-administered in international populations. With previous international consultation [3] forming the basis of items within this questionnaire, we expect the IFLQ-19 to hold relevance across Western and middle to high income countries, with statistical testing now feasible across a broad range of cultures and economic levels internationally. This research favoured the perspective of the general population as obtained through early cognitive interview studies [23]. As the Vidgen & Gallegos conceptualisation [1] was designed with the general population to determine knowledge, skills and behaviours that encompass food literacy, it was integral the final questionnaire reflected this perspective over statistical cut-offs. Existing food literacy surveys have been limited due to their use of varying definitions or low adherence to the Vidgen & Gallegos conceptual [1] of food literacy. Recent research identified the Vidgen & Gallegos [1] model as the core conceptualisation of food literacy [4], thus, the development of a questionnaire which comprehensively addresses this framework was needed [4]. Finally, this food literacy questionnaire fills a substantial gap in that food literacy, a consumer behaviour, has been difficult to conceptualise and measure [90]. Thus, food systems monitoring and surveillance to date has been limited in assessing this construct [90]. This questionnaire aims to fill this gap by measuring the key components of acquisiting, preparation, meal practices and storage [90]. This allows for further investigation of the relationship between food literacy and the broader food system, including the impact on retail, marketing, food environments [90], dietary intake and food security [2].

The limitations of this research include the collapse of disordered categories during the IRT analysis and the length of the IFLQ-19. While the collapsing of response categories in Study 1 addressed threshold disordering, the questionnaire was not administered in Study 2 with the re-formed categories. As threshold ordering is based on an a priori structure, empirical review is recommended [55]. However, this was not conducted for pragmatic reasons, as forced choice, where too few categories are provided or missing intervals in response choices can cause respondents confusion, resulting in imprecise category choices [91,92,93].

The final IFLQ-19 developed from this research consisted of 100-items which is expected to take approximately 20-min to complete. In order to reduce participant burden, methods such as Computerised Adaptive Testing (CAT), where respondents receive a unique set of items from a larger item bank based on their responses, are recommended [49].

Future research should explore the use of CAT for the IFLQ-19, conduct the relevant statistics so the IFLQ-19 can be scored at the theoretical component, domain and overall level and examine associations with diet quality and food security to determine if higher scores on statistical components of food literacy are meaningfully associated with these constructs. Further, resources and videos will be available online or presented as part of workshops and at conferences on the use and scoring of the IFLQ-19 to assist with policy and practice implementation.

Conclusion

This study progressed the development of a comprehensive, validated food literacy questionnaire using item response theory. The resulting 100-item food literacy questionnaire reflected the 4 domains of planning and managing (6 statistical components), selecting (4 statistical components), preparing (3 statistical components) and eating (6 statistical components). The food literacy questionnaire had acceptable and supported targeting and responsiveness, high validity and good reliability in a diverse sample of Australian adults. This questionnaire fills a substantial gap in the conceptualisation, measurement, monitoring and surveillance of food literacy, a consumer behaviour, internationally: a key aspect within the broader food system.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Abbreviations

CPCs:

Category Probability Curves

CTT:

Classical Test Thoery

ICC:

Intraclass Correlation Coefficients

IFLQ-19:

International Food Literacy Questionnaire -19

IRT:

Item Response Theory

KMO:

Kaiser-Meyer Olkin

MPCM:

Multidimensional Partial Credit Model

MSQs:

Outfit Mean-Squares

PCA:

Principal Component Analysis

PCM:

Partial Credit Model

References

  1. Vidgen HA, Gallegos D. Defining food literacy and its components. Appetite. 2014;76:50–9.

    PubMed  Article  Google Scholar 

  2. Vidgen HA. Food literacy: what is it and does it influence what we eat? Queensland University of Technology. 2014.

    Google Scholar 

  3. Fingland D, Thompson C, Vidgen HA. Measuring food literacy: progressing the development of an international food literacy survey using a content validity study. Int J Environ Res Public Health. 2021;18:1141–58.

    PubMed  PubMed Central  Article  Google Scholar 

  4. Thompson C, Adams J, Vidgen HA. Are We Closer to International Consensus on the Term ‘Food Literacy’?: A Systematic Scoping Review of Its Use in the Academic Literature (1998–2019). Nutrients. 2006;2021:13.

    Google Scholar 

  5. Truman E, Lane D, Elliott C. Defining food literacy: A scoping review. Appetite. 2017;116:365–71.

    PubMed  Article  Google Scholar 

  6. Renwick K, Powell LJ. Focusing on the literacy in food literacy: practice, community, and food sovereignty. J Fam Consum Sci. 2019;111:24–30.

    Article  Google Scholar 

  7. Azevedo Perry E, Thomas H, Samra HR, Edmonstone S, Davidson L, Faulkner A, Petermann L, Manafò E, Kirkpatrick SI. Identifying attributes of food literacy: a scoping review. Public Health Nutr. 2017;20:2406–15.

    PubMed  Article  Google Scholar 

  8. Barbour L, Ho M, Davidson Z, Palermo C. Challenges and opportunities for measuring the impact of a nutrition programme amongst young people at risk of food insecurity: A pilot study. Nutr Bull. 2016;41:122–9.

    Article  Google Scholar 

  9. Begley A, Paynter E, Dhaliwal SS. Evaluation tool development for food literacy programs. Nutrients. 2018;10:1617–32.

    PubMed Central  Article  Google Scholar 

  10. Wijayaratne SP, Reid M, Westberg K, Worsley A, Mavondo F. Food literacy, healthy eating barriers and household diet. Eur J Mark. 2018;52:2449–77.

    Article  Google Scholar 

  11. Wallace R, Lo J, Devine A. Tailored nutrition education in the elderly can lead to sustained dietary behaviour change. J Nutr Health Aging. 2016;20:8–15.

    CAS  PubMed  Article  Google Scholar 

  12. Palumbo R, Annarumma C, Adinolfi P, Vezzosi S, Troiano E, Catinello G, Manna R. Crafting and applying a tool to assess food literacy: findings from a pilot study. Trends Food Sci Technol. 2017;67:173–82.

    CAS  Article  Google Scholar 

  13. Lahne J, Wolfson JA, Trubek A. Development of the Cooking and Food Provisioning Action Scale (CAFPAS): A new measurement tool for individual cooking practice. Food Qual Prefer. 2017;62:96–105.

    Article  Google Scholar 

  14. Hutchinson J, Watt JF, Strachan EK, Cade JE. Evaluation of the effectiveness of the Ministry of Food cooking programme on self-reported food consumption and confidence with cooking. Public Health Nutr. 2016;19:3417–27.

    PubMed  Article  Google Scholar 

  15. Amuta-Jimenez AO, Lo C, Talwar D, Khan N, Barry AE. Food label literacy and use among US adults diagnosed with cancer: results from a national representative study. J Cancer Educ. 2019;34:1000–9.

    PubMed  Article  Google Scholar 

  16. Beatrice AB, Elizabeth M, Meaghan RB, Lynn R, Rebecca T. The Ontario Food and Nutrition Strategy: identifying indicators of food access and food literacy for early monitoring of the food environment. Health Promot Chronic Dis. 2017;37:313–9.

    Article  Google Scholar 

  17. Krause CG, Beer-Borst S, Sommerhalder K, Hayoz S, Abel T. A short food literacy questionnaire (SFLQ) for adults: Findings from a Swiss validation study. Appetite. 2018;120:275–80.

    Article  Google Scholar 

  18. Méjean C, Hassen WS, Gojard S, Ducrot P, Lampuré A, Brug H, Lien N, Nicolaou M, Holdsworth M, Terragni L. Social disparities in food preparation behaviours: a DEDIPAC study. Nutr J. 2017;16:1–13.

    Article  Google Scholar 

  19. Poelman MP, Dijkstra SC, Sponselee H, Kamphuis CB, Battjes-Fries MC, Gillebaart M, Seidell JC. Towards the measurement of food literacy with respect to healthy eating: the development and validation of the self perceived food literacy scale among an adult sample in the Netherlands. Int J Behav Nutr Physc Act. 2018;15:1–12.

    Article  Google Scholar 

  20. Amouzandeh C, Fingland D, Vidgen HA. A scoping review of the validity, reliability and conceptual alignment of food literacy measures for adults. Nutrients. 2019;11:801.

    PubMed Central  Article  Google Scholar 

  21. Vidgen H. Food literacy: key concepts for health and education. New York: Routledge; 2016.

  22. International Society of Behavioural Nutrition and Physical Activity. Advancing Behavior Change Science. Hong Kong: In ISBNPA; 2018. p. 3.

    Google Scholar 

  23. Thompson C, Adams J, Vidgen HA. Progressing the development of a food literacy questionnaire using cognitive interviews. Public Health Nutr. 2021;25:1968–78.

    Article  Google Scholar 

  24. Jin X, Liu GG, Gerstein HC, Levine MA, Steeves K, Guan H, Li H, Xie F. Item reduction and validation of the Chinese version of diabetes quality-of-life measure (DQOL). Health Qual Life Outcomes. 2018;16:1–11.

    Article  Google Scholar 

  25. Streiner DL, Norman GR, Cairney J. Health measurement scales: a practical guide to their development and use. Oxford: Oxford University Press; 2015.

    Book  Google Scholar 

  26. Cappelleri JC, Lundy JJ, Hays RD. Overview of classical test theory and item response theory for the quantitative assessment of items in developing patient-reported outcomes measures. Clin Ther. 2014;36:648–62.

    PubMed  PubMed Central  Article  Google Scholar 

  27. Rusch T, Lowry PB, Mair P, Treiblmaier H. Breaking free from the limitations of classical test theory: developing and measuring information systems scales using item response theory. Inf Manag. 2017;54:189–203.

    Article  Google Scholar 

  28. Paek I, Cole K. Using R for item response theory model applications. New York: Routledge; 2020.

  29. Van der Linden WJ. Handbook of Item Response Theory: Three Volume Set. Boca Raton: CRC Press; 2018.

  30. Bode RK. Partial credit model and pivot anchoring. J Appl Meas. 2001;2(1):78–95.

    CAS  PubMed  Google Scholar 

  31. Fox C. An introduction to the partial credit model for developing nursing assessments. J Nurs Educ. 1999;38:340–6.

    CAS  PubMed  Article  Google Scholar 

  32. Jafari P, Bagheri Z, Safe M. Item and response-category functioning of the Persian version of the KIDSCREEN-27: Rasch partial credit model. Health Qual Life Outcomes. 2012;10:1–6.

    Article  Google Scholar 

  33. Sébille V, Challa T, Mesbah M. Sequential analysis of quality of life measurements with the mixed partial credit model. In: Advances in statistical methods for the health sciences. New York: Springer; 2007. p. 109–25.

  34. Tandon A, Murray CJ, Salomon JA, King G. Statistical models for enhancing cross-population comparability. In Health systems performance assessment: debates, methods and empiricism. Geneva: World Health Organization; 2003.

    Google Scholar 

  35. Santos TSS, Julian C, de Andrade DF, Villar BS, Piccinelli R, González-Gross M, Gottrand F, Androutsos O, Kersting M, Michels N. Measuring nutritional knowledge using Item Response Theory and its validity in European adolescents. Public Health Nutr. 2019;22:419–30.

    Article  Google Scholar 

  36. Nord M. Introduction to item response theory applied to food security measurement: Basic concepts, parameters, and statistics. In: Technical Paper Rome: FAO, vol. 2014. 2014. p. 18. https://www.fao.org/publications/card/en/c/577f6a79-9cbd-49f5-b606-500ea42bf88e/.

  37. Hobart J, Freeman J, Lamping D, Fitzpatrick R, Thompson A. The SF-36 in multiple sclerosis: why basic assumptions must be tested. J Neurol Neurosurg Psychiatry. 2001;71:363–70.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  38. Hobart J, Riazi A, Lamping D, Fitzpatrick R, Thompson A. How responsive is the Multiple Sclerosis Impact Scale (MSIS-29)? A comparison with some other self report scales. J Neurol Neurosurg Psychiatry. 2005;76:1539–43.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  39. Nunnally JC Jr. Introduction to psychological measurement. New York; 1970.

  40. Australian Bureau of Statistics. Census of Population and Housing: Socio-Economic Indexes for Areas (SEIFA), Australia; 2016. https://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/2033.0.55.0012016?OpenDocument.

  41. Australian Bureau of Statistics. Income data in the Census. https://www.abs.gov.au/websitedbs/censushome.nsf/home/factsheetsuid?opendocument&navpos=450.

  42. Qualtrics: Qualtrics ESOMAR 28. Amsterdam: Qualtrics; 2019. p. 1–7.

  43. Tsutakawa RK, Johnson JC. The effect of uncertainty of item parameter estimation on ability estimates. Psychometrika. 1990;55:371–90.

    Article  Google Scholar 

  44. Nguyen TH, Han H-R, Kim MT, Chan KS. An introduction to item response theory for patient-reported outcome measurement. Patient. 2014;7:23–35.

    PubMed  PubMed Central  Article  Google Scholar 

  45. Edelen MO, Reeve BB. Applying item response theory (IRT) modeling to questionnaire development, evaluation, and refinement. Qual Life Res. 2007;16:5–18.

    PubMed  Article  Google Scholar 

  46. Australian Bureau of Statistics. Australian Standard Classification of Cultural and Ethnics Groups (ASCCEG). https://www.abs.gov.au/statistics/classifications/australian-standard-classification-cultural-and-ethnic-groups-ascceg/latest-release.

  47. Australian Bureau of Statistics. Australian Statistical Geography Standard (ASGS), vol. 5. Remoteness Structure; 2016. https://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/1270.0.55.005July%202016?OpenDocument.

  48. Microsoft Excel. 2022. https://www.microsoft.com/en-au/microsoft-365/excel.

  49. Desjardins CD, Bulut O. Handbook of educational measurement and psychometrics using R. Boca Raton: CRC Press; 2018.

  50. Field A: Discovering statistics using IBM SPSS statistics. Sage. 2013.

  51. Samuels P. Advice on exploratory factor analysis. Birmingham: Birmingham City University; 2017. p. 1–7.

    Google Scholar 

  52. Wright BD, Masters GN. Rating scale analysis. Chicago: MESA press; 1982.

  53. Mair P, Hatzinger R. Extended Rasch modeling: The eRm package for the application of IRT models in R. 2007.

    Google Scholar 

  54. Rasch Measurement Theory Analysis in R: Illustrations and Practical Guidance for Researchers and Practitioners [https://bookdown.org/chua/new_rasch_demo2/]

  55. Andrich D, Marais I. A course in Rasch measurement theory. 2019.

    Book  Google Scholar 

  56. Bond TG, Fox CM. Applying the Rasch model: Fundamental measurement in the human sciences. New York: Psychology Press; 2015. p. 490.

  57. Andrich D. A rating formulation for ordered response categories. Psychometrika. 1978;43:561–73.

    Article  Google Scholar 

  58. Hobart J, Cano S. Improving the evaluation of therapeutic interventions in multiple sclerosis: the role of new psychometric methods. Health Technol Assess. 2009;13(12):214.

  59. Embretson SE, Reise SP. Item response theory. New York: Psychology Press; 2013.

  60. Böckenholt U, Meiser T. Response style analysis with threshold and multi-process IRT models: a review and tutorial. Br J Math Stat Psychol. 2017;70:159–81.

    PubMed  Article  Google Scholar 

  61. Rocque M, Posick C, Zimmerman GM. Measuring up: Assessing the measurement properties of two self-control scales. Deviant Behav. 2013;34:534–56.

    Article  Google Scholar 

  62. Eggert S, Bögeholz S. Students’ use of decision-making strategies with regard to socioscientific issues: an application of the Rasch partial credit model. Sci Educ. 2010;94:230–58.

    Google Scholar 

  63. McHorney CA, Tarlov AR. Individual-patient monitoring in clinical practice: are available health status surveys adequate? Qual Life Res. 1995;4:293–307.

    CAS  PubMed  Article  Google Scholar 

  64. Holmes WC, Shea J. Performance of a new, HIV/AIDS-targeted quality of life (HAT-QoL) instrument in asymptomatic seropositive individuals. Qual Life Res. 1997;6:561–71.

    CAS  PubMed  Article  Google Scholar 

  65. Andresen EM. Criteria for assessing the tools of disability outcomes research. Arch Phys Med Rehabil. 2000;81:S15–20.

    CAS  PubMed  Article  Google Scholar 

  66. Fisher W Jr. Reliability statistics. Rasch Measurement Transactions. 1992.

    Google Scholar 

  67. Fit diagnosis: infit outfit mean-square standardized [https://www.winsteps.com/winman/misfitdiagnosis.htm]

  68. Felt JM, Castaneda R, Tiemensma J, Depaoli S. Using person fit statistics to detect outliers in survey research. Front Psychol. 2017;8:863.

    PubMed  PubMed Central  Article  Google Scholar 

  69. Polit DF. Getting serious about test–retest reliability: a critique of retest research and some recommendations. Qual Life Res. 2014;23:1713–20.

    PubMed  Article  Google Scholar 

  70. Ratner B. The correlation coefficient: Its values range between+ 1/− 1, or do they? J Target Meas Anal Mark. 2009;17:139–42.

    Article  Google Scholar 

  71. Norquist JM, Fitzpatrick R, Dawson J, Jenkinson C: Comparing alternative Rasch-based methods vs raw scores in measuring change in health. Medical Care 2004:I25-I36.

  72. McHorney CA, Haley SM, Ware JE Jr. Evaluation of the MOS SF-36 physical functioning scale (PF-40): II. Comparison of relative precision using Likert and Rasch scoring methods. J Clin Epidemiol. 1997;50:451–61.

    CAS  PubMed  Article  Google Scholar 

  73. Wolfe F. Which HAQ is best? A comparison of the HAQ, MHAQ and RA-HAQ, a difficult 8 item HAQ (DHAQ), and a rescored 20 item HAQ (HAQ20): analyses in 2,491 rheumatoid arthritis patients following leflunomide initiation. J Rheumatol. 2001;28:982–9.

    CAS  PubMed  Google Scholar 

  74. Help for Winsteps Rasch Measurement and Rasch Analysis Software: User-friendly rescaling: zero point and unit [https://www.winsteps.com/winman/rescaling.htm]

  75. Help for Winsteps Rasch Measurement and Rasch Analysis Software: SCOREFILE= person score file [https://www.winsteps.com/winman/scfile.htm]

  76. Australian Bureau of Statistics. 2016 Census. https://www.abs.gov.au/websitedbs/censushome.nsf/home/2016.

  77. Gothwal VK, Bharani S, Reddy SP. Measuring coping in parents of children with disabilities: a Rasch model approach. PLoS One. 2015;10:e0118189.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  78. Chen T-A, O’Connor TM, Hughes SO, Beltran A, Baranowski J, Diep C, Baranowski T. Vegetable parenting practices scale. ITEM Response Modeling Analyses Appetite. 2015;91:190–9.

    PubMed  Google Scholar 

  79. An M, Yu X. A Rasch analysis of emerging adults’ health motivation questionnaire in higher education context. PLoS One. 2021;16:e0248389.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  80. Kramer B, McLean S, Shepherd Martin E. Student grittiness: A pilot study investigating scholarly persistence in EFL classrooms. J Osaka Jogakuin Coll. 2018;47:25–41.

    Google Scholar 

  81. Miller KJ, Pollock CL, Brouwer B, Garland SJ. Use of Rasch analysis to evaluate and refine the community balance and mobility scale for use in ambulatory community-dwelling adults following stroke. Phys Ther. 2016;96:1648–57.

    PubMed  PubMed Central  Article  Google Scholar 

  82. National Health and Medical Research Council. The Guidelines. https://www.eatforhealth.gov.au/guidelines.

  83. National Health and Medical Research Council. Australian Guide to Healthy Eating. https://www.eatforhealth.gov.au/guidelines/australian-guide-healthy-eating.

  84. Zhao Y, Hambleton RK. Practical consequences of item response theory model misfit in the context of test equating with mixed-format test data. Front Psychol. 2017;8:484–95.

    PubMed  PubMed Central  Article  Google Scholar 

  85. Park YS, Lee Y-S, Xing K. Investigating the impact of item parameter drift for item response theory models with mixture distributions. Front Psychol. 2016;7:255–72.

    PubMed  PubMed Central  Google Scholar 

  86. Meijer RR, Sijtsma K. Methodology review: Evaluating person fit. Appl Psychol Meas. 2001;25:107–35.

    Article  Google Scholar 

  87. Mehta S, Bastero-Caballero RF, Sun Y, Zhu R, Murphy DK, Hardas B, Koch G. Performance of intraclass correlation coefficient (ICC) as a reliability index under various distributions in scale reliability studies. Stat Med. 2018;37:2734–52.

    PubMed  PubMed Central  Article  Google Scholar 

  88. Gordon RA. Measuring constructs in family science: how can item response theory improve precision and validity? J Marriage Fam. 2015;77:147–76.

    PubMed  PubMed Central  Article  Google Scholar 

  89. Somaraki M, Ek A, Sandvik P, Byrne R, Nowicka P: How do young children eat after an obesity intervention? Validation of the Child Eating Behaviour Questionnaire using the Rasch Model in diverse samples from Australia and Sweden. Appetite. 2021:105822.

  90. Fanzo J, Haddad L, McLaren R, Marshall Q, Davis C, Herforth A, Jones A, Beal T, Tschirley D, Bellows A. The Food Systems Dashboard is a new tool to inform better food policy. Nature Food. 2020;1:243–6.

    Article  Google Scholar 

  91. Foddy WH. Constructing questions for interviews and questionnaires: Theory and practice in social research. Cambridge: Cambridge University Press; 1993. p. 246. https://doi.org/10.1017/CBO9780511518201.

  92. Kelsey JL, Whittemore AS, Evans AS, Thompson WD: Methods in observational epidemiology. Monographs in Epidemiology and; 1996.

  93. Choi BC, Pak AW. Peer reviewed: a catalog of biases in questionnaires. Prev Chronic Dis. 2005;2:1–13.

    Google Scholar 

Download references

Acknowledgements

The authors wish to thank all participants of the research.

Funding

C.T. is supported by an Australian Government Research Training Program Scholarship and the King and Amy O’Malley Trust postgraduate scholarship. The panel recruitment was supported by Queensland University of Technology, Faculty of Health, Centre for Children’s Health Research funding. J.A. is supported by the UK Medical Research Council (grant number MC_UU_00006/7). The funding bodies above had no role in the design, collection, analysis or interpretation of this study.

Author information

Authors and Affiliations

Authors

Contributions

All authors conceptualised the study; all authors contributed to the methodology. C.T. analysed the data and prepared the original manuscript. All authors reviewed and edited the manuscript, all authors have read and approved the final manuscript.

Authors’ information

C.T. is a PhD Candidate at Queensland University of Technology in Public Health Nutrition. J.A. is an Associate Professor and Programme Leader of the Population Health Interventions programme in the MRC Epidemiology Unit at the University of Cambridge. R.B. is a Senior Lecturer at the Queensland University of Technology. H.A.V. is an Associate Professor in Public Health Nutrition at the Queensland University of Technology.

Corresponding author

Correspondence to Courtney Thompson.

Ethics declarations

Ethics approval and consent to participate

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 University Human Research Ethics Committee (UHREC) at QUT, Approval Number: 2000000004. Written informed consent was obtained from all participants.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Thompson, C., Byrne, R., Adams, J. et al. Development, validation and item reduction of a food literacy questionnaire (IFLQ-19) with Australian adults. Int J Behav Nutr Phys Act 19, 113 (2022). https://doi.org/10.1186/s12966-022-01351-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12966-022-01351-8

Keywords

  • Food literacy
  • Survey
  • Item response theory
  • Rasch measurement
  • Partial credit model
  • Test–retest reliability
  • Validity