ABSTRACT: Computer-based training is increasingly favored by food companies for training workers due to convenience, self-pacing ability, and ease of use. The objectives of this study were to determine if personal hygiene training, offered through a computer-based method, is as effective as a face-to-face method in knowledge acquisition and improved attitude toward food safety. Employees from four food processing facilities (n= 94) were randomly assigned to a control group, a face-to-face training group, or a computer-based training group. Evaluation instruments consisted of a pretest and a posttest to measure knowledge gain, a step scale to measure attitude after training and retrospectively before training, and a brief set of interview questions. Results for both treatments and the control group indicated increases in knowledge. Analysis of the posttest scores, when controlling for the pretest scores, indicated the difference was significant (P≤ 0.05); however, post hoc analysis did not indicate which treatment was superior. There was a significant increase in attitude scores for both groups using paired t-tests, although the difference between groups was not significant when controlling for attitude scores before training. Issues that may limit application of the technology used in this study include low literacy and language barriers among employees, and time constraints within companies. These and other variables should be taken into account in future research studies addressing comparisons of training methods. This study is the first to demonstrate the effectiveness of computer-based training aimed at food handlers who work in food processing establishments.
Observed decreases in the incidence of food-borne infections over the past several years have been attributed to new government regulations, voluntary control measures put into place by industry, and increased attention of food safety issues from consumer groups and the media, as well as enhanced food safety educational efforts aimed at consumers and workers who handle food (Vugia and others 2004). A variety of instructional methods are available for training employees in the workplace such as face-to-face classroom lectures, videos, distance education materials, and printed brochures and manuals. Another method that has gained in popularity within the last several years and has been demonstrated to be effective when compared to traditional methods (Culbertson and Smith 2003) is the use of computers to deliver training materials, known as computer-based instruction (CBI).
In Pennsylvania, approximately 75000 workers are employed by food product manufacturing businesses (Pennsylvania Center for Workforce Information and Analysis 2005). Of these businesses, 25% employed less than 4 and 59% employed less than 50 workers (PennTAP 2004). CBI food safety training courses could offer these small businesses flexibility in scheduling, a portable method of instruction, and possible reduction in training costs since the materials could be used multiple times. Therefore, the objectives of this study were to (a) develop a CBI training module on personal hygiene for food processing line workers and (b) compare its effectiveness with face-to-face (FTF) training in terms of knowledge gain and changes in attitude. Data on selected demographic characteristics of the participants were also gathered.
Materials and Methods
Learning objectives were written that covered issues related to health, handwashing, and general personal hygiene practices in food processing facilities. Learners were to be able to: (1) understand why personal hygiene is important when handling food; (2) know the meaning of the following terms: hand sanitizer, jaundice, asymptomatic diseases; (3) recognize the symptoms that exclude a person from working with food; (4) know when and how to wash hands properly; (5) know when and how to use gloves; (6) understand why the use of tobacco in food processing areas can be hazardous to food; and (7) recognize how proper work clothes prevent food contamination.
Text for the module relative to the specific learning objectives described above was developed and transformed into a scripted Microsoft PowerPoint® (Microsoft Corporation, Redmond, WA 2000) presentation for FTF delivery. A parallel version (Figure 1) of the presentation was developed for CBI using Macromedia Flash® (Macromedia, Inc., San Francisco, Calif., U.S.A. 1999) software. The CBI version was loaded onto a CD-ROM and then copied onto laptop computers located at the training sites. Both versions contained the same text either on the slides of the CBI module or in the script of the FTF version, since the information on the Powerpoint slides was presented in bullet form. Both versions also contained identical graphics, photographs, and video clips. The Flesch-Kincaid Grade Level Scale (Kincaid and others 1975) was used to compare the relative reading comprehension level of each version. Scores for the CBI and the FTF versions were 8.0 and 6.5, respectively, indicating that the material could be comprehended by individuals with at least an 8th-grade education. The grade level was lower for the FTF version since the text on slides was not presented in complete sentences. Before the material was presented, each participant signed the informed consent form approved by the Pennsylvania State Univ.'s Institutional Review Board.
Four evaluation instruments were developed: a demographic questionnaire, a pretest/posttest, a “post-then-pre” attitude assessment, and a brief set of interview questions for the CBI group. The demographic questionnaire was administered before the pretest. Questions were designed to gather data on the selected characteristics of the participants. The pretest/posttest was used to measure the knowledge gained as a result of the training. Each test consisted of 18 multiple-choice questions that were based on the 7 learning objectives. A “Don't know” answer was included as a choice for the questions to reduce the possibility of guessing. The “post-then-pre” questionnaire consisted of 6 questions that measured participant attitudes regarding personal hygiene after the training and, retrospectively, before the training. This method of evaluation allowed participants to make a more accurate assessment of their knowledge level, since it gave them a baseline level of knowledge for reference (Rockwell and Kohn 1989). In the current study, this method was adapted for attitude assessment. A step scale with 7 steps was used to quantify answers. For the purposes of analysis, each step was assigned a number value, with 1 meaning “not important” and 7 meaning “very important.” At the end of each training session, participants placed a “B” on the step that reflected their attitude regarding the subject before training and then also placed an “N” on the step that reflected their attitude regarding the subject now that training was completed (Figure 2). Using this system, a total of 42 points each were possible on the “before” and “after” questions. Open-ended interview questions were asked of participants who completed the CBI training to gather opinions on the suitability of this training method in the workplace. Due to time limitations, these questions were not asked of the control group because they did not receive training prior to evaluation.
The demographic questionnaire, pretest/posttest, and attitude assessment evaluation tools were reviewed by a panel of experts consisting of 4 food safety specialists and 1 evaluation specialist for content validity and face validity (Norland 1990) and field-tested in a small ready-to-eat meat processing establishment. In addition to reviewing them, the participants (n= 13) in the field test also were asked to complete the CBI module and evaluation tools.
A modified randomized pretest/posttest control group design (Figure 3) was used in this study. Pintauro and others (2005) recently implemented a similar experimental design to compare the knowledge gain and attitudes of students using 3 instructional methods. Four United States Dept. of Agriculture (USDA)–inspected meat and poultry processors in Pennsylvania were used in this study. According to the USDA classification system, they were “small” meaning that they had 10 to 500 employees. Participants were required to be at least 18 y of age, work directly with food, and to read, write, and speak English. Each company had a food safety training program in place that could include on-the-job training with coworkers; internal and external training sources; written and verbal instruction from management; Good Manufacturing Practices (GMP) document that was read by employees twice a year; and a quiz or video training.
Before the study was conducted, a contact person at each company was established to provide preliminary information on the number of employees at the company, the current status of training programs at the company, the number of shifts, and the ready-to-eat meat or poultry products produced. The contact person at each company compiled a list of participants who met the inclusion criteria and forwarded this list to the researchers. At three of the four companies, all of the line employees were included, while at one company, only line employees on the day shift participated. Supervisory employees participated in the training sessions in the case where they worked directly with food on production lines. Participants from each company were randomly assigned to control, FTF, or CBI groups. Prior to data collection, the company contact person received a set of instructions for the training sessions. The demographic questionnaire and pretest were administered to the control group approximately 1 wk prior to training at each facility. During the week between the pretest and the posttest, the control group did not receive any additional materials or training by the researchers. The control group members were not segregated and may have conversed with other employees or management during this time. All data were collected during November, 2004.
On the day of training at a facility, those randomly assigned to the control group completed the posttest and attitude assessment prior to receiving training. Participants in the FTF and the CBI groups completed the demographic questionnaire and pretest, were immediately trained, and then completed the posttest and attitude assessment. The interviews were conducted with the CBI group on completion of the posttest and attitude assessment. To allow all participants to receive the instruction, the control group attended the FTF training session, but did not complete additional evaluation tools at the conclusion of the training.
All data were analyzed using the Statistical Package for Social Sciences (SPSS), release 11.0 for Windows (SPSS Inc., Chicago, Ill., U.S.A. 2000). Data from participants (N= 4) who did not complete all of the questions on the pretest and posttest were removed from further analysis. Attitude data were analyzed when all questions were answered, with the exception of the control group in which only data from responses on attitudes “before” training were analyzed. Since this approach was taken, the participants were not able to accurately assess the change due to training because they had not completed the training.
The demographic data were split according to group assignment into control, FTF, or CBI. Data including age, level of education, and years worked at the company were recoded into larger categories that encompassed two or more of the multiple choice answers for analysis.
Reliability was calculated for the knowledge data using the split-half method (Fraenkel and Wallen 2003). Descriptive statistics included the mean, the standard deviation, and the range. Further analyses included paired t-tests, independent t-tests, 1-way analysis of variance (ANOVA), and analysis of covariance (ANCOVA). Scheffé test (Garson 2002) was used for post hoc analysis when the variances were equal according to Levene's test (Garson 2002) for equal variances, while the Games-Howell test was used when variances were not equal (Garson 2002). The attitudinal data were analyzed in a similar manner except the calculation of the reliability was conducted using Cronbach's alpha (Fraenkel and Wallen 2003). Interview data were summarized and commonalities were noted.
Results and Discussion
The participants (n= 94) represented 4 companies geographically distributed throughout Pennsylvania with 1 company in the west, 2 in the south-central region, 1 in the eastern-central region, and 1 in the southeast. The demographic profile of the participants indicated that 68% (n= 62) were male. The majority of participants (69%, n= 65) were white, while 20% (n= 19) were African American. Just over 7% (n= 7) were Hispanic and 3 considered themselves to be of another race. The majority of the participants, 84% (n= 74), indicated they had completed high school or technical school, had some college, or had a college degree. Age was evenly distributed among participants according to a chi-square analysis of age distribution by category (not shown). The mean age was 39 ± 12 y old. Most participants had not worked for this company for more than 5 y, since 31% (n= 29) indicated that they worked for the company for 0 to 1 y and 38% (n= 35) indicated that they had worked for the company for 2 to 5 y. Of those responding, only 6 participants indicated they had never been trained in food safety practices. Of those 6 participants, 3 were in the FTF group and 3 were in the CBI group.
When asked about computer usage habits, slightly more than one-third (35%, n= 32) indicated they had never used a computer, while slightly less than one-third of the participants (31%, n= 28) indicated they used a computer daily. More than one-half of the participants (64%, n= 57) reported that they were somewhat comfortable, moderately comfortable, or very comfortable using a computer.
The maximum score possible on the knowledge assessment was 18, which was equal to the number of questions. A correct response was assigned 1 point while an incorrect response was assigned 0 point. The reliability score of the knowledge assessment was 0.786 (n= 94), which falls above the acceptable standard of 0.70 (Fraenkel and Wallen 2003). Knowledge gain, examined by comparing pretest and posttest scores using paired t-tests, was significant for the control group, the FTF group, and the CBI group (Table 1). Increases in control group audit scores have been noted by Soneff and others (1994) for reasons such as high staff turnover, lack of support from management to implement change, and failure of the treatment group to read the educational materials. The pretest scores were not equal among the three groups; therefore analysis of posttest scores was conducted by controlling for pretest scores using ANCOVA. The results indicated a significant (P < 0.05) difference among groups; post hoc analysis did not indicate any significant differences in pair-wise comparisons. Based on these findings, we cannot conclude that the results from one instructional method were significantly different from the other method.
Table 1—. Pretest and posttest scores from knowledge assessment
at value from paired t-test of pretest and posttest scores.
bF value from ANCOVA of posttest scores by group assignment with pretest scores as the covariate.
15.1 ± 1.9
15.8 ± 2.1
P < 0.05
12.6 ± 3.8
15.7 ± 3.0
P < 0.05
12.8 ± 4.4
14.5 ± 4.5
P < 0.05
P < 0.05
Average scores of knowledge assessment increased as the participants' self-perceived abilities to read, write, and speak English increased. These data were gathered from the demographic questionnaire. This assessment was made by splitting the data for each group according to self-perceived ability scores and examining the test scores for each group. This increase occurred regardless of the assignment of participants to the control, FTF, or CBI groups. Knowledge was not significantly affected by any other demographic characteristics, which was consistent with the findings of Costello and others (1997).
A total of 7 points was possible on each of the 6 attitude assessment items, making a maximum possible score of 42. The reliability score of the attitude assessment was 0.783 (n= 84). Both the FTF and the CBI groups demonstrated a positive change in attitude regarding personal hygiene from “before the training” to “after the training” (Table 2). When attitude scores “after” training were compared by group and controlled for scores “before” training, there was no significant difference. It is of note that the “before” scores for the control group were equivalent to the “after” scores for both treatment groups; therefore the “post-then-pre” method may have allowed a more accurate assessment of attitudes before training for the experimental groups. It also is possible that the better attitudes toward personal hygiene of the control group reflected their pretest scores. Using the same method of evaluation, Barrett and others (1998) demonstrated improved attitude scores following food safety training. In this study, changes in attitude scores were not related to any of the recorded demographic characteristics.
Table 2—. Before and after instruction attitude assessment scores
NS = no significant difference
at value from paired t-test of attitude scores before and after training
bF value from ANCOVA of attitude scores after training by group assignment with attitude scores before training as the covariate
40.3 ± 2.5
36.4 ± 5.6
40.3 ± 3.4
P < 0.05
36. 5 ± 5.3
40.6 ± 3.5
P < 0.05
During the interviews of the CBI group, participants indicated that they liked the CBI training because “it was different, quick, easy, short,” and “it allowed them to proceed at their own pace.” Most participants indicated that they felt computers were effective as a training medium and that the material covered during instruction also was effective. Most participants felt their coworkers would be receptive to CBI and that usage was self-explanatory.
We noted several problems during this study, which should be remedied during future research of this nature. Participating companies volunteered their time and employees. Therefore, employees were not randomly selected from a large pool that met the inclusion criteria. A ceiling effect, similar to that observed by Eckerman and others (2004) and Anger and others (2004), also occurred in knowledge score. With only 18 questions on the posttest, some participants had little room to improve since they scored high on both the pretest and posttest. A threat due to testing also may have been an issue as the control group improved their scores significantly between the pretest and the posttest, despite lack of instruction. The improvement in scores could be attributed to previous exposure to the testing instrument. It is also plausible that the participants in this group could have sought out correct information from management, coworkers, or company Good Manufacturing Practices (GMP) documents between the time of the pretest and the posttest. Using cluster sampling to assign all employees from a particular worksite to a group (Soneff and others 1994; Costello and others 1997) could reduce the chances of knowledge diffusion through communication, which may have contributed to the observed increase in control group scores. Another possible remedy is to increase the time between the administration of the pretest and posttest to the control group, since the tests were separated by 1 wk in the current study.
Another problem encountered was low literacy of some participants and poor ability to respond verbally or in writing to open-ended questions. As future educational materials are developed, the reading level of the material should be kept at the appropriate level for the target audience. In cases where participants have difficulty with the English language, an option put forth by other researchers was adding an audio component (Anger and others 2004) to the module. The collection of interview data from the FTF group may have provided more insight into the workers' preferences and views toward training methods.
In the present study, management allowed most employees no more than 1 h of work time to participate in the training. This time frame became an issue since most workers completed the training in less than 1 h, while some workers appeared to have difficulty reading the materials and took close to 1.5 h to complete the training. Milheim (1993) and Eckerman and others (2004) also observed that studies conducted in the workplace were subject to similar stipulations from management. Milheim (1993) further asserted that adults in the workplace are a difficult target audience to train. Walker and others (2003) demonstrated that employees in the food industry were difficult to train because they were often from lower socioeconomic classes and lower education levels, and were subject to rapid turnover of employees, especially when seasonal staff was employed. Interestingly, these findings were consistent with observations of the current study.
Our research demonstrated that CBI may be a useful tool for providing food safety training to employees in the food processing industry. Compared to FTF instruction, CBI offers greater flexibility in scheduling, consistent presentation of material, minimal requirement of instructor time, and reusable materials. We found no significant difference in food safety knowledge gained through either the FTF instruction method or the CBI method. However, knowledge increases in the control group prevented us from determining if one method was significantly superior to the other. We also found that both methods were equally effective in increasing participant attitudes toward personal hygiene. Studies conducted in an active workplace environment illustrate inherent difficulties due to constraints from management on which workers are permitted to attend training at a given time, the location of training sessions, and the number of training sessions permitted at a facility, time restrictions, turnover of employees, and diversity within the workforce including varying literacy levels and languages. These and other variables should be taken into account in future research projects addressing comparisons of training methods. Although this research has demonstrated no significant difference between the CBI and FTF methods, further research is needed to confirm the effectiveness of CBI for long-term knowledge gain and behavior change among food processing employees.
The authors would like to thank the Pennsylvania companies and employees who participated in the study. This research was funded, in part, by the Dept. of Food Science and The College of Agricultural Sciences, Pennsylvania State Univ., University Park, PA.