Food Protection Trends - January/February 2018 - 30

dents were ambiguous because of inconsistent responses
that made it impossible to definitively determine the type of
milk products consumed. All data analyses were completed
on the remaining 454 respondents.
Online survey and data analysis
The survey contained 94 items other than the three for
inclusion criteria. The questions originated from previous
studies (6, 7, 9, 13, 14, 15) with modifications to the specific
needs of the present survey. Testing and validation of items
are described in these references. The survey was converted
to an online format using SelectSurvey.NET (ClassApps,
Release 2.0, Overland Park, KS). Data were analyzed by ChiSquare tests (non-parametric for cell counts and percentages), ANOVA (parametric variables), discriminant analysis
(continuous variables only), and HLM (all variables). Data
analysis was accomplished using the Software Package for the
Social Sciences (SPSS, version 24, Chicago IL). Significant differences were declared if probability levels were at least 95% or
greater. In some cases, principal component analysis was used
to identify the set of data with the best internal consistency,
defined for this study as Cronbach α > 0.600 (18).
Model components
The first part of the mental models followed the Risk
Information Seeking and Processing (RISP) model (5).
Constructs were: Individual Characteristics, Perceived
Hazard Characteristics, Affect, Informational Subjective
Norms, Information Insufficiency, information channel
beliefs, Information Gathering Capacity, Information
Seeking Behavior, and Information Processing Behavior.
The second part of the mental models was based on constructs of the Theory of Planned Behavior (TBP) (1). The
constructs were: Attitude, Behavioral Subjective Norms,
Perceived Behavioral Control, and Milk Consumption
Behavior. Each construct was assessed with one or more
survey items, some of which were modified from the
literature (6, 7, 9, 13, 14, 15), or modified from the results
of item analysis with the current dataset. For all variables,
specific survey item stems are shown in each table and response choices for each item are detailed in the tables and
footnotes. Marking the "choose not to answer" response
was designated as missing data.
Individual characteristics
Individual characteristics included demographic descriptors (Gender, Age, Education, Income, and Ethnicity) and
were nominal-choice items, with one exception: respondents
were asked to record their actual age. Continuous data for
Age were then coded to four categories: 18 - 29 years, 30
- 39 years, 40 - 49 years, and 50 - 77 years, based on the
distribution of responses. Primary and secondary school
responses were combined for the Education variable because
of low response for the primary school option. Relevant Haz-

Food Protection Trends January/February

ard Experience queried the respondent's previous personal
experience with foodborne illnesses. One item pertained to
personal experience and whether the illness was resolved
by medical care or was self-limiting. An additional item
queried the hazard experience of close friends or relatives of
the respondent and the resolution of illnesses. The Relevant
Hazard Experience items were nominal choice, with responses for "don't know" and "don't remember" combined into
one named "Don't know." Political Philosophy was nominally queried as the best choice describing the respondent's
personal assessment; choices ranged from very liberal to very
conservative. Another item asked if respondents considered
themselves Libertarian.
Perceived hazard characteristics
A respondent's personal control over contracting foodborne illness from consuming unpasteurized milk products
was measured in two Likert-scaled items. One item (Personal
Control, self-protection) addressed the belief that the immune system will protect the respondent from foodborne illness, and the other (Personal Control, personal choice) asked
about the degree to which the respondents believed they
have a choice about consuming unpasteurized milk products.
Responses to the two items were not summed because of low
internal consistency (α = 0.58). Risk Judgment was measured
by two Likert-scaled items that queried the respondent's
perceived likelihood of contracting serious foodborne illness
if unpasteurized milk products were consumed and that measured perceived seriousness of illnesses caused by consuming
unpasteurized milk items. Responses to the two items were
summed with high internal consistency (α = 0.82). Four
Likert-scaled items measured aspects of Institutional Trust.
Summed responses had an internal consistency of α = 0.93.
Affect
The emotional response to the consumption of unpasteurized milk products was assessed in items that measured
the degree of affect the respondent felt. Two items were
constructed as Likert-scaled items measuring the independent concepts of Worry and Anger. A single variable was
constructed and named Affect (α = 0.863).
Informational subjective norm
The influence of family and friends on the expectation
that the respondent stays informed about the safety
of milk products was measured in two Likert-scaled
items. One item measured the verbal and non-verbal
expectations that family and friends exerted over the
respondent, and the second item measured the motivation
to stay informed that the respondent felt from significant
others. The variable was constructed as the product of the
responses to the two items and was named Informational
Subjective Norm.

http://www.SelectSurvey.NET

Table of Contents for the Digital Edition of Food Protection Trends - January/February 2018