Consumers are becoming more health conscious and prefer to have nutrition information available during food purchasing(Reference Lando and Labiner-Wolf1). Diet and nutrition are important factors to lower the risks of many chronic diseases of lifestyle(Reference Rosell, Appleby and Key2), which endorses the link between nutrition and health(Reference Reid3). Labels educate and inform consumers to make nutritionally appropriate choices(Reference Wills, Schmidt and Pillo-Blocka4), promoting healthy dietary behaviour. It is thus important that consumers have sufficient knowledge and interpretation skills regarding label information to make healthy dietary choices(Reference Alfieri and Byrd-Bredbenner5), as limited nutritional knowledge could cause consumers problems in reading labels(Reference Cowburn and Stockley6) and prevent them from benefiting from label information.
Knowledge obtained through experience or education can be defined as information that was evaluated and organised to result in a convinced understanding of subject matter (nutrition information on labels) with the skills to utilise it with a specific intention (to make a purchasing decision)(7). The nutritional knowledge acquired through labels correlates with the nutritional quality of consumers’ diets(Reference Borra8). Amplified information required on food labels(Reference Gaschler, Mata and Störmer9) could however confuse and overload consumers with information that they consequently ignore during their purchasing decisions(Reference Norgaard and Brunso10), resulting in uninformed choices and unhealthy lifestyles. In order to improve the credibility(Reference Van der Merwe and Venter11) and accuracy of label information(12), South African (SA) Labelling Regulations (R642) were revised and promulgated in 2007(13).
Research has indicated a link between consumers’ food label knowledge and demographics(Reference Mahgoub, Lesoli and Gabotswang14). As a result of socio-political changes, previously disadvantaged, price-sensitive consumers were allowed to enter the emerging economy and consumer markets(Reference D'Andrea, Ring and Aleman15). Since the estimated 50·6 million SA citizens(16) have different socio-economic backgrounds, lifestyles and cultures(Reference Gothan and Erasmus17), and international consumer research is not directly relevant to the unique SA environment, the behaviour and knowledge of SA consumers thus need to be researched. Although different international studies have focused on consumers’ knowledge of nutrition label information(Reference Alfieri and Byrd-Bredbenner5, Reference Byrd-Bredbenner18–Reference Grunert, Wills and Fernández-Celemín20), only Anderson and Coertze(Reference Anderson and Coertze21) have determined consumers’ knowledge of nutrition information on labels in the SA context.
The aim of the present study was to determine consumers’ knowledge regarding food label information in a SA context. The first three objectives were to determine the ability of consumers to: (i) locate and manipulate information; (ii) assess the accuracy of nutrient content claims and which health/nutrient claims are allowed; and (iii) identify symbols on labels. Also, associations pertaining to use and knowledge regarding label information were determined for consumers from different demographic and related groups.
Methods
Research design
A descriptive quantitative approach was employed, using structured questions. A face-to-face survey was conducted to obtain a better response rate and to clarify unclear aspects to respondents(Reference Maree and Pietersen22).
Sampling
Purposive sampling was employed, recruiting male and female respondents who complied with four inclusion criteria(Reference Babbie23): (i) the respondent, his/her spouse or partner did not work as a dietitian, consumer scientist, nutritionist or pharmacist, as these professions are likely to be more knowledgeable about nutrition and health; (ii) the respondent was his/her household's primary food purchaser (>50 % of shopping) to ensure exposure to label reading; (iii) the respondent was literate and able to read label information without visual impairments(Reference Alfieri and Byrd-Bredbenner5); and (iv) the respondent was aged 18 years or older.
Respondents were recruited at public locations such as post offices, health-care centres or municipal offices in Potchefstroom, North West Province, thereby attempting to incorporate aspects of representativeness into the non-probability sample.
Data collection instrument
Data were collected through a 15 min, seven-part interviewer-administered questionnaire (n 229). Existing instruments used in nutrition labelling surveys(Reference Alfieri and Byrd-Bredbenner5, Reference Byrd-Bredbenner, Alfieri and Wong19) were adapted to also include questions about food labels in general.
Part I (‘Opening’) explained the research aim and inclusion criteria; Part II collected demographics while Part III (‘General information’) also included information regarding respondents’ shopping behaviour (frequency and time spent shopping). Four-point Likert scales allowed respondents to evaluate the nutritional quality of their diet, their health status (1 = poor; 4 = excellent) and their informedness about nutrition and label information (1 = not informed at all; 4 = well informed). Part IV (‘Nutrition is important beliefs’) only included two Likert-scale statements: ‘Eating nutritiously can help prevent certain diseases’ and ‘I eat what I want regardless of what is good for me’ (1 = strongly disagree; 4 = strongly agree). Part V included questions regarding the frequency of label use, how often it affects respondents’ purchasing decisions and label use during first-time purchasing (1 = never; 4 = always).
Part VI contained ‘label knowledge’ questions to represent SA food labels, including a question on the ‘identification of symbols’. Internal reliability of ‘label knowledge’ questions was determined with the Kuder–Richardson 20 reliability coefficient, yielding the following values: ‘locating information’ = 0·74; ‘identifying symbols’ = 0·68; ‘health/nutrient claims’ = 0·53; ‘manipulating information’ = 0·47; and ‘nutrient content claims’ = 1·39. Internal consistency for the first three factors was acceptable, although the third should be interpreted with caution; while item-discrimination values for items within these factors were good (>0·3), indicating acceptable item discrimination. The average inter-item correlations of these factors were 0·26, 0·31 and 0·27, respectively, which indicates acceptable reliability. ‘Manipulating information’ and ‘nutrient content claims’ showed low reliability and were interpreted as separate items.
Reliability of the instrument was furthermore ensured and improved by discussions with field workers, careful construction and pre-testing of the questionnaire(Reference Steyn24). Cross-tabulations indicated a practically significant (ω = 0·45) association showing that respondents’ rating of their informedness regarding nutrition increased with their informedness about label information. The frequency that label information affected respondents’ purchasing decisions showed medium to practically significant associations with their frequency of label reading (ω = 0·43) and with label reading during first-time purchasing (ω = 0·42), also confirming excellent reliability(Reference Steyn24).
Face validity was determined to indicate whether the instrument truly measures respondents’ knowledge about labels. Content validity ensured that the measuring instrument represented the entire content of the study's constructs(Reference Pietersen and Maree25) and was established by having the adapted questionnaire refereed by a panel (n 8) of research experts in consumer and food sciences. Statistical Consultation Services of the North-West University scrutinised the questionnaire. The present study served as an exploratory study for larger-scale application.
Part VII contained questions regarding labelling preferences and information sources, expanded to include more options.
Statistical analysis
Data were analysed using the statistical software package PASW Statistics 18 (SPSS Inc., Chicago, IL, USA). Descriptive statistics (means and frequencies) and statistically significant differences in means were determined with ANOVA and Tukey's test. Associations between variables were determined with cross-tabulations(Reference Ellis and Steyn26). Because a non-probability sample was used, P values are not reported, since practical significance was determined. Cohen's effect sizes (d values) indicated whether differences were important in practice, where d = 0·2 indicated small, d = 0·5 medium and d = 0·8 large effect sizes. Effect sizes of ω = 
$$\[-->$<>\sqrt{{\tf="MPi-OneI"\char120}^2/n}<$>
<!--\]$$, where χ 2 is the usual chi-square statistic for the contingency table and n is the sample size, were used to test associations in cross-tabulations, where values of 0·1, 0·3 and 0·5 indicated small, medium and large effects, respectively(Reference Ellis and Steyn26). Only large effect sizes are reported and considered to indicate practically significant associations between variables.
Ethical considerations
Ethical approval for the study was obtained from the North-West University (reference number: NWU-00024-09-S1) and all ethical measures were practically applied.
Results and discussion
Demographic and general characteristics
The demographic profile of the study population (Table 1) showed that the majority were female, Afrikaans-speaking, had a tertiary education and had no children. The results confirmed that females are responsible for most of the household food purchases(Reference Byrd-Bredbenner18) and that nutritional educational programmes should firstly be directed at females(Reference Vijayapushpam, Menon and Rao27). The higher number of educated, Afrikaans-speaking respondents with no children may be explained by Potchefstroom being an academic environment(28), the North-West University (Potchefstroom Campus) using Afrikaans as the basic medium of teaching and attendance at tertiary educational institutions postponing parenthood(Reference Byrd-Bredbenner18). The age groups were moderately evenly represented, while 47 % belonged to a middle income group(29).
Table 1 Characteristics of the study population: respondents (n 229) recruited through purposive sampling in Potchefstroom, North West Province, South Africa
General information; nutrition is important beliefs and label use
According to the results, 46 % of respondents did their main grocery shopping once per month and 62 % spent less than an hour in doing so. Respondents might have been women in the early stages of their career facing time constraints, who mostly did not purchase food frequently and spent less time shopping. This infrequent shopping by respondents was in contrast with more frequent purchasing results found by Jacobs et al.(Reference Jacobs, De Beer and Larney30), while respondents in both studies spent little time shopping. In the current study, most respondents rated the nutritional quality of their diet (66 %) and health (83 %) as being good to excellent as compared with 88 % and 95 % of American consumers(31); while consistent with American consumers(32), 85 % rated themselves as being somewhat informed about nutrition and 68 % as being well informed about labels. These respondents were thus health, nutrition and label conscious; understood the link between food and health; and used nutrition information to improve healthy eating(Reference Rosell, Appleby and Key2).
Most respondents (Table 2) indicated that they learned to read labels with the assistance of relatives and friends, newspapers/magazines and television, similar to American consumers(31). These sources are thus useful to assist with consumer label education, although previous results suggests that consumers attach more value to printed media(Reference Ellis and Glanville33).
Table 2 Percentage of respondents (n 229) consulting different information sources in learning about food labels, Potchefstroom, North West Province, South Africa
The majority of respondents were aware of food label legislation (69 %), read labels during first-time purchasing (70 %) and sometimes to always read labels (73 %), confirming earlier research(Reference Jacobs, De Beer and Larney30). Label reading sometimes to always affected respondents’ purchasing decisions (62 %), which is lower than that of American consumers (88–92 %)(Reference Byrd-Bredbenner, Alfieri and Wong19, 32). Since most respondents were regular label readers, only 30 % or fewer selected reasons for not often using food labels.
Although almost all respondents (95 %) agreed to strongly agreed that ‘eating nutritiously can help prevent certain diseases’, only 24 % reported special diets for themselves or family members, of which 93 % were for health reasons. Thus, consumers focused more on overall health maintenance than on a particular health condition(32). Despite respondents’ reported health and label awareness and consciousness, 56 % reported ‘I eat what I want regardless of what is good for me’ since consumers tend not to use nutrition labels, even though they say that they do.
Food label knowledge
A total of 86 % of respondents were able to locate label information (Table 3), which corresponds with respondents’ claims of being health and label conscious. Most respondents (97 %) could correctly identify symbols, while only 19 % were able to correctly identify health/nutrition claims allowed on labels. Similarly, a label front panel icon representing a healthy product also received favourable responses from participants in a previous study(Reference Lando and Labiner-Wolf1).
Table 3 Frequencies of correct responses to questions regarding food label knowledge among respondents (n 229), Potchefstroom, North West Province, South Africa
The items regarding the manipulation of information were correctly calculated by most respondents although they seemed to be more focused on ingredient and nutrient content than on the number of servings. Previous research also showed that respondents had difficulty with some label information questions(Reference Stuart, Schröder and Hughes34) and performing calculations regarding portion sizes(Reference Rothman, Housam and Weiss35).
Almost all respondents were able to correctly identify the ‘high in vitamin A’ and ‘trans-fat free’ claims, while fewer (<5 %) could identify false claims regarding vitamin E and saturated fat (Table 3). The respondents’ knowledge regarding trans fat might be due to the association of trans fat being unhealthy(Reference Ellis and Steyn26, 32). The low knowledge scores relating to vitamin E and saturated fat might indicate consumers’ lack of understanding of nutrients and their inability to evaluate nutritional label information(Reference Jacobs, De Beer and Larney30). Consumers thus need opportunities to learn about complex criteria(Reference Byrd-Bredbenner, Alfieri and Wong19) and set parameters(Reference Byrd-Bredbenner, Wong and Cottee36) for ‘low’ or ‘high’ nutrient content claims. The lack of knowledge about health and nutrient claims confirms former research(Reference Alfieri and Byrd-Bredbenner5) which also indicates a need for consumer education regarding nutrition, labelling and use of labels to enable informed decisions(Reference Van der Merwe and Venter11).
Respondents’ knowledge of food label information as explained by associations with demographic, general and nutrition beliefs and label use items
A high percentage of respondents (87 %) provided correct responses to almost 80 % of the knowledge questions, and therefore only the following practically significant associations were found between questions (Table 4).
Table 4 Comparison of practically significant associations of knowledge among different demographic groups of respondents (n 229), Potchefstroom, North West Province, South Africa
a,bMean values within a group with unlike superscript letters were practically significantly different with a large effect size (d ≥ 0·8).
1. Afrikaans- (d = 0·92 and d = 0·96, respectively) and English-speaking respondents (d = 0·84 and d = 0·82, respectively) were practically significantly more knowledgeable with regard to the factors ‘locating information’ and ‘identifying symbols’ than respondents speaking an African language. These differences between the label knowledge of respondents from different language groups might be explained by cultural differences.
2. Younger respondents (18–24 years) were practically significantly (d = 0·91) more knowledgeable concerning the factor ‘locating information’ than older respondents (≥55 years). Although previous research confirms that senior citizens are less knowledgeable about label information, they are more interested in healthy eating and in verifying nutrition labels than younger respondents(Reference Grunert, Wills and Fernández-Celemín20).
3. Concerning the factors ‘locating information’ and ‘identifying symbols’, respondents with a tertiary (d = 1·21 and d = 1·27, respectively) and grade 12 qualification (d = 0·88 and d = 0·94, respectively) were practically significantly more knowledgeable than those with qualifications below grade 12, as confirmed by previous research(Reference Cowburn and Stockley6). A lack of education results in infrequent label reading(Reference Jacobs, De Beer and Larney30) and a limited ability to make informed purchasing decisions(Reference Van der Merwe and Venter11).
4. A practically significant (ω = 0·45) association indicated that as respondents’ rating of their informedness regarding nutrition increased, their informedness about label information also increased and vice versa.
5. Respondents’ frequency of label reading (ω = 0·43) and label reading during first-time purchasing (ω = 0·42) showed medium to practically significant associations with the frequency that label information affects their purchasing decision, as confirmed by Jacobs et al. (Reference Jacobs, De Beer and Larney30). Food labels thus serve as an important medium that consumers may use during purchasing decisions(Reference Mahgoub, Lesoli and Gabotswang14).
The differences in the label knowledge of respondents of different demographic backgrounds indicate a need to target food label educational programmes at specific subgroups of the SA population and to particular areas of label education, as found for nutrition education programmes(Reference Hendrie, Coveney and Cox37).
A few associations of medium and small effect sizes were also found but are not reported due to their lack of practical significance.
Conclusions
The results confirmed the health and label awareness of most respondents, strengthened by their knowledge of food labels. Respondents’ informedness about nutrition shaped into informedness about label information, label reading practices and label-influenced purchasing decisions. Most respondents showed the ability to locate and manipulate label information, identify symbols and some nutrient content claims, but an inability to identify some permissible health/nutrient claims and false claims.
Despite some respondents’ excellent knowledge, areas for improvement were identified as well as differences in label knowledge of respondents from different demographic backgrounds, which indicate a need for consumer education. Food label educational programmes utilising different media should therefore start with sufficient nutritional background on certain label aspects, focusing on specific demographic subgroups to motivate frequent label reading, to apply label knowledge during purchasing decisions.
This exploratory research is valuable due to the lack of research regarding SA consumers’ food label knowledge. Considering the small non-probability sample used it is evident that future research should aim to represent the diversity of the SA population, since practically significant associations between respondents’ knowledge and their demographics were evident. Such research should be conducted on a larger scale, thereby providing information to consumer educators concerned with food label knowledge.
Acknowledgements
Sources of funding: The study was funded by the National Research Foundation. Conflict of interest: There has been no conflict of interest. Authors’ contributions: D.v.d.M. acted as the project manager, initialized the study, obtained the funding, and organized the literature collection and application, adapting and refining of the questionnaire, data interpretation and writing of the manuscript. M.B. was co-responsible for literature collection and application, adapting and refining the questionnaire, data interpretation and writing the manuscript. S.E. was responsible for statistical analysis and interpretation. H.d.B. and A.M. assisted with the screening of the questionnaire and reviewing of the manuscript. Acknowledgements: The assistance of the Consumer Sciences Honours Class of the North-West University of 2010 in data collection and literature collection is acknowledged with gratitude.
