Finding new gaps in a well established theory can be very challenging, traditionally we rely on well-established leaders in the theory to do an review, and point out the directions of future. Recent development in Meta-analysis provided new possibilities to mathematically examine existing "theoretical system", therefore finding the bridges, this paper provided an example of how-to

1. Department of Hospitality Management
College of Human Ecology
Using The Theory Of Planned Behavior To Predict Food Safety Behavioral
Intention: A Meta-analysis
Naiqing Lin Ph.D.; Kevin R. Roberts, Ph.D.
Kansas State University
SSRN: https://ssrn.com/abstract=3104751 or http
://dx.doi.org/10.2139/ssrn.3104751

2. Department of Hospitality Management
College of Human Ecology
Introduction
 In 2016, restaurants were still the most commonly reported
location of foodborne diseases outbreaks (469, 52%) and
associated illnesses (4,757, 31%)1.
 It has been suggested that 97% of outbreaks traced to non-
manufacturing food businesses involved food handler error or
malpractice2.
 To effectively reduce foodborne illnesses, foodservice
managers and food handlers must perform essential food
safety behaviors3.
1. Angelo, Nisler, Hall, Brown, & Gould, 2017; CDC, 2017
2. Howes et al., 1996; Griffith, Livesey, & Clayton, 2010
3. Debess et al., 2009; Green et al., 2007; Green & Selman, 2005

3. Department of Hospitality Management
College of Human Ecology
Problem
 Health inspections within the public sector1 and audit reports from
the private sector2 have identified significant degrees of non-
compliance food safety behavior.
 Behavioral resistance and failure to comply with proper food safety
practices are both widespread and problematic3.
 Using best available antecedents like attitudes, norms, self-
efficacy, perceptions of risk and severity, and personality factors
only explained about 14% to 24% of the variance in behaviors4.
1. CDC, 2016; 2. Egan et al., 2007; Valerie et al., 2008
3. Roberts et al., 2008, 2009; Roberts & Barrett, 2009, 2011)
4. Godin & Kok, 1996; McEachan et al., 2011; Sheeran & Webb, 2016

4. Department of Hospitality Management
College of Human Ecology
Problem
 High-quality systematic syntheses are rarely conducted with food
safety behaviors1.
 A significant proportion of variance within behavioral intention is
unexplained2, suggesting room for improvement.
 Previous review articles focus on summarizing the educational
effectiveness (i.e., improved knowledge) not on food safety
behaviors, thus have considerable limitations2,3.
1. Egan et al., 2007; Ong, Frewer, and Chan, 2017
2. Sheeran, 2002; Sheeran & Webb, 2016
3. Pilling et al., 2008; Sniehotta et al., 2014; York et al., 2009

5. Department of Hospitality Management
College of Human Ecology
Purpose
The purpose of this study is to summarize and evaluate the ability of the
Theory of Planned Behavior (TPB) to predict food safety behavioral
intentions.

6. Department of Hospitality Management
College of Human Ecology
Literature Review
Theoretical Assumption - the Theory of Planned Behavior
(TPB)
Attitude
Subjective Norm
Perceived
Control
Behavioral
Intention
Self-Reported
Behavior
Note. Adapted from Fishbein, M., & Ajzen, I. (2011). Predicting and changing behavior: The reasoned
action approach. Taylor & Francis.
Experience
Age
Gender

7. Department of Hospitality Management
College of Human Ecology
Literature Review General Validity and Utility of the TPB
 The TPB remains one of the most influential theories of cognitive behavioral change1.
 The TPB asserts that it is vital to change intention to perform a behavior in order to
change existing behavior or initiate a new behavior2.
 Various correlational studies3 and meta-analysis of meta-analyses3 has supported
the general validity and utility of the TPB4.
 Intention offers a superior prediction of behavior in correlational tests compared to
other cognitions including explicit and implicit attitudes, norms, self-efficacy,
perceptions of risk and severity and personality factors4.
1. Fishbein & Ajzen, 2011; Sheeran et al., 2016; 2. Rhodes & Dickau, 2012; Webb &
Sheeran, 2006; 3. Armitage & Conner, 2001; McEachan et al., 2011; Sheeran & Webb,
2016; 4. Chiaburu et al., 2011; Sheeran, Harris, & Epton, 2014

8. Department of Hospitality Management
College of Human Ecology
Proposed Research Questions
RQ1: Does attitude, subjective norm, perceived control have a significant random-effect on food
safety behavioral intention among food service workers?
RQ2: What are the unobserved heterogeneity ratios (I2) in observed effects between food safety
attitude, subjective norms, and perceived control to food safety behavioral intention?

9. Department of Hospitality Management
College of Human Ecology
Methods
 Boolean phase searches were used in databases and the search
strategy were peer reviewed by two indexing Librarians.
 Manual reviews were conducted with the journals with the most food
safety behavior publications from June 1982 to October 2017.
 The researchers contacted the corresponding authors of the
selected studies and request any unpublished data or studies.
 MIX 2.0 (Englewood NJ: Biostat) were used to manage data.
Meta-Analysis Procedures

10. Department of Hospitality Management
College of Human Ecology
Methods
 Studies reporting consumer intentions to adhere to food safety
behaviors were excluded.
 Studies using only college students or high school students were
excluded1.
 Studies must have at least one direct measure of attitudes,
subjective norm or perceived behavioral control2.
 A flow diagram were used to visualize and provide rationale for
exclusion at the screening stage.
Meta-Analysis - Inclusion and Exclusion Criteria
1. Landers & Behrend, 2015
2. Ball, Wilcock, & Aung, 2010

11. Department of Hospitality Management
College of Human Ecology
Methods
 Study coding were confirmed to the APA Meta-Analytic Reporting
Standards and PRISMA 2009 standard1.
 Aspect of research design were coded as control variables2.
 Publication biases were visualized using the Trim and Fill method
(T&F)3 and the significance level (p = 0.05) were tested using
Egger’s Regression of the Intercept Test (ERI)3.
Meta-Analysis – Coding, Reporting and Bias Correction
1. APA, 2011; Liberati et al., 2009
2. Valentine & Cooper, 2008
3. Duval & Tweedie, 2000; Egger et al., 1997

12. Department of Hospitality Management
College of Human Ecology
Methods
• Fisher’s Z were used to calculate
effect sizes, which allows
collective aggregation between
study effects for overall effect size
(𝐸𝑆) based on a random effects
model using weighted average
intra-study correlation1. Forest
plots were used for visualization1.
• The level of data heterogeneity
among the included studies were
calculated using Qz
2 and I2 3.
Meta-Analysis – Statistics
1. Rosenthal, Cooper, & Hedges, 1994; Liberati et al., 2009
2. Borenstein et al., 2009
3. Deeks, Altman, & Bradbum, 2008

13. Department of Hospitality Management
College of Human Ecology
Results

14. Department of Hospitality Management
College of Human Ecology
Study Retrieval

15. Department of Hospitality Management
College of Human Ecology
Study Retrieval

16. Department of Hospitality Management
College of Human Ecology
Study Characteristics
Survey Type
Telephone Survey Paper Survey
Place
US UK
Language
English Spanish Chinese

17. Department of Hospitality Management
College of Human Ecology
Sample and Variables
Sample
Food handlers (School/ Care)
Restaurant employees
Food industry workers
Dependent Variable
Hand hygiene
Avoid contamination
Food safety behavior
Training Behavior

18. Department of Hospitality Management
College of Human Ecology
Study Quality
Pilot Study
Yes No
Measurement
7-point Likert type
Bi-polar scale
5-point Likert type scale
TACT
Yes No

19. Department of Hospitality Management
College of Human Ecology
Summary of Study Effects – Main Results
23
• A total of 46 study records
• 19 Attitude to intention correlations:
– 0.271 (95% [Cl] = 0.149 to 0.385, p < 0.01), with a small to medium effect size
• 13 Subjective norms to intention correlations:
– 0.370 (95% [Cl]= 0.190 to 0.455, p < 0.01), with a medium to large effect size
• 14 Perceived behavioral control to intention correlations:
– 0.247 (95% [Cl]= 0.096 to 0.386, p < 0.01), with a small to medium effect size

20. Department of Hospitality Management
College of Human Ecology
Heterogeneity and Explained Variances– Main Results
25
• Results indicate the level of between-study heterogeneity is not
significant (Q = 1.851, df =2, p=0.396).
– a satisfactory degree of homogeneity is achieved.
• The total variance of all TPB studies included in this particular meta-
analysis explained of 22% in true effect total variances (R2 = 0.22).
• A considerable amount of variance is still unexplained (78.4%).

21. Department of Hospitality Management
College of Human Ecology
Quality Control – Publication Bias
26
Egger’s regression
(t44 = 0.676,
p = 0.502).

22. Department of Hospitality Management
College of Human Ecology
Discussion and Limitations

23. Department of Hospitality Management
College of Human Ecology
Theoretical Implication
 Examined prediction and observed variance within food safety behavioral intention1
 The weighted average correlation from each independent variable ranged from 0.247 to
0.370, with medium to large effect sizes.
 Help solve theoretical conflicts regarding perceived control 2
 0.247 (95% [Cl]=0.096 to 0.386, p<0.01).
 Theory enhancement by segmentation3, low level of heterogeneity ratio
 the TPB model is surprisingly robust in terms of predicting food safety intention with
medium effect sizes indifferent from languages used.
1. Jedidi et al., 1997; Van de Ven, 2007
2. Cooke, Sniehotta, & Schuez, 2007; Lin & Roberts, 2017
3. Creswell 2005; Tashakkori & Teddlie 2003

24. Department of Hospitality Management
College of Human Ecology
Theoretical Implication – Measurement
 The results noted that less than half of the studies included in this
study utilized an elicitation study.
 Less studies used the TACT elements of the behavioral definition to
improve accessible behavioral outcomes
 Bipolar measurement
 Future food safety researchers are encouraged to clearly define the
food safety action using the TACT elements
1. Jedidi et al., 1997; Van de Ven, 2007
2. Cooke, Sniehotta, & Schuez, 2007; Lin & Roberts, 2017
3. Creswell 2005; Tashakkori & Teddlie 2003

25. Department of Hospitality Management
College of Human Ecology
Practical Implication
 The results indicated that subjective norms tend to be the most
influential construct in TPB model to predicate food safety intention.
 Government authorities rely on high-quality systematic
reviews to update policies and best-practice guidelines.
 Studies with a large sample tend to have significantly stronger
individual subjective norms.
 Correspondent to the group’s culture and identities
 Academic researchers use quality synthesis to identify
research gaps within existing literature.
 Help identify cost-effective risk prevention strategies and revise
epidemiology practices.

26. Department of Hospitality Management
College of Human Ecology
Limitations
 Only included studies published in English
 Did not include longitudinal studies
 Publication bias
 Low combined variance of true effect.

27. Department of Hospitality Management
College of Human Ecology
Questions?
nlin@ksu.edu
SSRN: 2867180