Study area
The study was conducted in selected places in South Wollo (Dessie, Kombolcha, and Wereilu), and Oromia (Kemissie and Bati) zones. South Wollo and Oromia zones are in the Amhara regional state with the geographic coordinates of 10.8997° N, 38.9877° E, and 10.3959° N, 40.0000° E, respectively. South Wollo and Oromia zones are situated in the north-eastern part of Ethiopia, 401, and 327 km away from Addis Ababa (the capital city of Ethiopia), respectively. South Wollo and Oromia zones cover the area of 17,067.45 km2, and 286,612 km2, respectively.
Study population
South Wollo and Oromia zones have a total population number of 2,518,862 and 457,278 respectively [11]. The study population was consumers of raw beef in selected raw beef restaurants. For the eating behavior assessment, all age groups greater than 18 years old and both sexes were included. A total of 570 raw beef consumers were interviewed. The majority (70.18%) of the participants were from South Wollo (35.09% in Dessie, 26.32% in Kombolcha, and 8.77% in Wereilu) and the remaining (29.82%) were from the Oromia zone (17.54% in Kemissie, and 12.28 in Bati).
Study design
A cross-sectional type of study (a study that investigates a situation at a point in time) was carried out from January 2021 to September 2021 in selected cities and towns of South Wollo and Oromia zones for assessing the raw beef-eating behavior of raw beef consumers. In this study, both descriptive and inferential statistics were used.
Sample size and data collection techniques
The sample size for eating behavior was done based on the suggestions of Taherdoost’s formula [12]. Taherdoost and his research team suggested that for every type of cross-sectional survey the following formula is more appropriate than others.
$$n=\frac{p\ \left(100-p\right)\ {z}^2}{e^2}$$
Where n = is the required sample size.
p = is the percentage occurrence of a state or condition.
z = is the value corresponding to the level of confidence required.
e = is the percentage maximum error required.
Since there was no preceded raw beef-eating behavior assessment conducted in the study areas, 50% for p-value, 95% (1.96) for z-value, and 5% for e-value were taken. As a result, the sample size was calculated as follows.
$$n=\frac{50\left(100-50\right){1.96}^2}{5^2}$$
$$=384\;\mathrm{minimum}\;\mathrm{samples}\;\mathrm{were}\;\mathrm{required}$$
Even if the minimum sample size is 384, the researchers collected a higher number of samples (570). The total sample size from Dessie, Kombolcha, Kemise, Bati, and Werielu were 200, 150, 100, 70, and 50, respectively.
Structured questionnaire interviews were conducted to assess the raw beef-eating behavior of raw beef consumers. The tables in the randomly selected raw beef restaurants were chosen randomly and any raw beef consumer in the selected table of each restaurant was invited for an interview. All the selected restaurants have sold both raw and processed (roasted, cooked, and fried) beef. Only raw beefeaters in the raw beef restaurants at the time of the interview who were volunteering to be interviewed were used and processed meat consumers were excluded. Those raw beef consumers who were not volunteer for an interview in the selected table were excluded from sampling. Lunchtime was purposively selected for the interview and one raw beef consumer was interviewed from 30 minutes to 1 hour depending on how fast the raw beef consumer understood the questions. The interview continued until the data or information saturation was attained. All the questions in the questionnaire were close-ended. The questionnaire has five sections and different sets of questions. The first section was about the general demographic characteristics of the raw beef consumers and the second section of the questionnaire was about the general raw beef eating habit of raw beef consumers while the third and fourth sections were about the choice of raw beef for consumption and their intention to change or minimize raw beef-eating, respectively. The fifth section of the questionnaire was about the perception of raw beef consumers toward raw beef-eating. The questions in sections three, four, and five enabled the researcher to understand the choice, intention, and perception of raw beef consumers, respectively. The questionnaire was composed of 34 questions/variables. Seven questions were used for each section of demographic characteristics, and general eating habits, eight questions about the perception of eating raw beef and six questions were used for each choice and intention of raw beef consumers.
All the questions concerned on the choice, intention, and perception of the raw beef consumers were pooled into a single variable, which had two categories. These two categories were favourable or unfavourable for choice, and favourable or unfavourable for both intention and perception of the raw beef consumers.
The consumers’ choice of raw beef eating was assessed based on the food choice conceptual model [13]. Six questions that are related to the consumers food preparation preference, the food type usually consumed, reason for the usual consumption of specific food item, feelings if consumers did not eat the usual food item, daily frequency of eating the usual food item, and the mealtime consumers eat their usual food item.
The intention of the raw beef consumers was assessed based on the theory of planned behavior [14]. Six questions (intention to reduce raw beef eating, knowledge on the health risk of raw beef-eating, intention to improve their knowledge on raw beef-eating health risk, willingness to stop raw beef-eating if consumers know raw beef-eating health risk, easiness to stop raw beef-eating, and obstacles to stop raw beef-eating) were used to investigate the raw beef consumers’ intention to stop eating raw beef.
The perception of raw beef consumers towards the safety of raw beef-eating was assessed based on Likert’s scale [15]. The agreement of the raw beef consumers on the exposure to diseases from raw beef, the fatality of diseases originated from raw beef, the benefits of raw beef-eating, the effect of spices and alcohol on the raw beef borne pathogens, the effect of heating/cooling on raw beef borne pathogens, the contamination of raw beef with dangerous pathogens, the raw beef’s potential to transmit diseases to humans and the respondents’ belief in the safety of raw beef-eating were the items used for the assessment of raw beef consumers perception.
The data set prepared from the 34 questions and the dependent variables of choice, intention, and perception of raw beef consumers were analyzed using bivariate logistic regression with SPSS version 25.
Data analysis
After the target sample size was collected, it was administered in Microsoft Excel 2013. Based on the answer of each choice, intention, and perception related questions, dependent binary variables were created for each choice, intention, and perception assessments of raw beef consumers. The participants whose answers were an indicator of raw beef-eating choice was categorized as “unfavourable choice” and whose answers were an indicator of not choosing raw beef-eating were grouped into favourable choice. Likewise, all the participants who intended to stop eating raw beef were grouped into favourable intentions, and those whose intentions was the opposite was categorized into the unfavourable intention category. In the same with choice and intention, the participants who perceive the health risks of eating raw beef were grouped into favourable perceptions, and those who perceive the opposite were categorized into unfavourable perceptions.
Based on the p-value of the logistic regression, the predictive explanatory variables for the result, favourable choice or unfavourable choice, favourable intention or unfavourable intention, and favourable perception or unfavourable perceptions were identified. The investigations of the participants’ choice, intention, and perception were conducted in three steps. The first step was assessing the relationship between potential predictor variables with the participants’ choice, intention, and perception one by one. Secondly, the relationship for the potential confounding effects was adjusted. Finally, the possibility of an interaction effect among the variables was considered.
To have initial insight into the structure of the data, cross-tabulations were used in SPSS version 25. From this basic descriptive tool, it is possible to see the proportions of each response category, which were indicative of the level of participants’ choice, intention, and perception of raw beef-eating.
After descriptive investigations using crosstabs, the association between the dependent binary variables (choice, intention, and perception) and each predictive variable was conducted. Probability values were used to see the association between these dependent binary variables and predictive variables (variables produced from each question). The effect levels of predictive variables on choice, intention, and perception of the participants were shown by the odds ratio (OR 95%CI).
