Food Protection Trends - September/October 2020 - 299

into scale items such as "I will help protect my restaurant from
liability for foodborne illnesses" and "I will help protect the
reputation of my restaurant."
The questionnaire (as shown in the appendix) consisted of
six subscales to measure the four constructs of the SCT, food
safety behavioral intentions, and self-reported food safety
behaviors. Eleven demographic and operational information
items were included to describe and understand the sample.
The self-efficacy scale included 10 items using a 5-point
Likert-type scale (1 = cannot do at all to 5 = can do for sure)
adapted with modifications (16, 17). The environmental
determinants scale included 10 items measured on a 5-point
Likert scale (1 = strongly disagree to 5 = strongly agree)
adapted from (2) and (21).
The outcome expectations scale included eight items
measured on a 5-point Likert scale (1 = strongly disagree
to 5 = strongly agree), adapted with modification (69). The
outcome expectations items included social, physical, and
self-evaluative consequences of the behavior (10).
The self-regulation scale consisted of three subscales, selfmonitoring, goal setting, and self-evaluation (8). The three
subscales were measured with three items, each on a 5-point
Likert scale (1 = strongly disagree to 5 = strongly agree)
adapted with modification (66).
Food safety behavioral intentions were measured with six
items on a 5-point Likert scale (1 = strongly disagree to 5 =
strongly agree) adapted from (36) and (46). Twelve items
were used to measure self-reported food safety behavior,
using a frequency response scale (1 = never to 5 = always)
adapted from (60) and (73). The food safety behavior
scale focused on three food safety behaviors implicated in
recurring incidents of foodborne illnesses in restaurants:
handwashing, use of a thermometer, and proper handling of
food and work surfaces (37, 59, 70, 73).
Different descriptors/anchors were used in the 5-point
scales throughout the questionnaire to mitigate the potential
impact of common method bias associated with the crosssectional self-reporting method used and to alleviate
potential social desirability bias caused by commonalities in
scale endpoints (61). In addition, anonymity of respondents
and their answers were assured to reduce the effect of
evaluation apprehension.
The questionnaire was translated to Chinese and Spanish,
and back translated to English to ensure that the correct
wording was used. To assess content validity, an experienced
researcher in food safety behavior reviewed the questionnaire
for face validity and adequacy of the scales' content. The
research protocol (proposal number: 9170) was approved by
the Kansas State University Institutional Review Board prior
to data collection.
Data collection
The questionnaire was pilot tested with a sample of 25
food handlers from independent Chinese and Mexican

restaurants prior to data collection. Restaurant owners/managers were contacted in person to request the participation
of their employees. After owners'/managers' approval had
been received, the questionnaire was administered onsite to
employees who wished to participate. Each participant who
completed the pilot test received $5 as a token of appreciation. No major issues were noted during the pilot testing,
although minor modifications to a few questions were made.
For instance, unnecessary spaces between questions were
adjusted to reduce the number of pages, thus reducing the fatigue caused by participants completing long questionnaires.
In addition, item D in the question about participants' level
of education in the Spanish version was changed from "Algo
de universidad" to "Algunas clases universitarias," which
means completing some college-level classes and is more
consistent with the English version. Reliability of the measurement scales was initially determined during pilot testing
and then was verified after the final data was collected.
The main study consisted of 50 restaurants from the three
counties included in the study, with the goal of sampling
five food handlers randomly from each of the 50 restaurants.
If a selected restaurant owner/manager declined the
request for participation of the restaurant's food handlers, a
replacement was taken from another independent Chinese
or Mexican restaurant in the counties until the target
number of participants was obtained. To ensure consistency,
participants were chosen based on two selection criteria:
participants had to be at least 18 years of age at the time of
recruitment and had to be food handlers.
Each employee received a cover letter that included
information about the study. Participants who completed the
questionnaire after having read the letter of information were
presumed to have given informed consent. Each participant
who completed the questionnaire received $5 as a token of
appreciation for their participation in the survey.
Data analysis
Data was analyzed using SPSS for Windows (Version 25).
Cronbach's alpha coefficient was computed to determine
the internal consistency/reliability for each scale. A cutoff point of 0.70 was used to demonstrate consistency
(58). Frequencies, means, and standard deviations were
computed to summarize and screen the data. Multiple
regression analysis was used to examine the ways in
which the constructs act alone and together to influence
behavioral intentions and self-reported food safety behaviors.
Mediation analysis was performed using the procedures of
the PROCESS macro developed for SPSS (35), including
bootstrap confidence intervals to examine the indirect effect
of food safety behavioral intentions. All significance levels
were set at P < 0.05. Confirmatory factor analysis (CFA) was
conducted using AMOS for Windows (Version 25) to test
the constructs' validity in the proposed model.

September/October Food Protection Trends

299

Food Protection Trends - September/October 2020

Table of Contents for the Digital Edition of Food Protection Trends - September/October 2020