Validation of an online questionnaire
measure of the relative reinforcing value of food
Karen K.
1
Saules ,
1
Reslan ,
Summar
and Mark K.
2
Greenwald
Eastern Michigan University, Department of Psychology 1
Wayne State University, Department of Psychiatry and Behavioral Neurosciences 2
Typically, the relative reinforcing value of food (RRV-F) is assessed
using a laboratory-based computerized choice task wherein
participants work for units of food. Despite yielding important
information, measuring the RRV-F in the laboratory is time consuming
and generally allows for the examination of only small sample sizes.
Although a paper-and-pencil questionnaire measure of the relative
reinforcing value of two food reinforcers has been established and
validated (see Goldfield, Epstein, Davidson, & Saad, 2005), this too
was designed for laboratory administration. The aim of the current
investigation was thus to validate a questionnaire measure of the
RRV-F deployed using an online survey methodology. The
advantages of online data collection include reduced administration
time, lower cost, greater anonymity, and increased ability to recruit
from larger samples (Granello & Wheaton, 2004).
Methods
The full sample included 281 undergraduate college students, 72% of
whom were female, with a self-reported mean BMI of 25.67 (SD =
5.66). Students were invited to complete a 20-30 min online survey.
The online survey assessed demographic variables, liking for
chocolate and cheese, self-reported height and weight, and the RRVF. Eligible survey participants (N = 21; female, non-pregnant, nonsmokers, and without carpal tunnel syndrome) were also invited to
take part in an experimental laboratory food choice task lasting 30
min. Participants were asked to refrain from eating 2 hr prior to the
experiment. They were told that unit food amount would remain
constant (1 piece per trial), but response requirement (i.e., number of
mouse button presses) would increase on successive trials. Food
was consumed post-session.
Food type
Mass
(grams)
4.55
4.54
Chocolate kiss
Cheese cube
Energy
(calories)
16.7
15.0
Fat
(grams)
1.33
1.38
Sugar
(grams)
2.55
0
80
70
Data Analysis
There was a significant positive skew for responses to the online
questionnaire. Therefore, non-parametric correlations were used to assess
the association between online and laboratory food choice responding and
their relationships with height, age, gender, and BMI. Consistent with
studies that have assessed the association between food reinforcement
and BMI (e.g., Giesen et al., 2010), participants were classified as normal
weight (BMI < 25 kg/m2) or overweight (BMI ≥ 25 kg/m2). A binary version
of the online measure of the RRV-F was created and chi-square analyses
were conducted to examine whether food choice responding on the online
questionnaire differed between overweight and non-overweight
respondents.
Results
The online RRV-F instrument showed strong predictive validity for
laboratory food choice behavior, convergent validity with overweight status,
and discriminant validity with unrelated constructs (age, gender, height).
Correlation coefficients between the online questionnaire measure of the
RRV of chocolate and variables supporting predictive, convergent and
discriminant validity
1
---
2
1.RRV
original
.86*** --2. RRV
binary
.44* .18
3. Lab food
Choice-Choc
-.40 -.05
4. Lab food
Choice-Cheese
5. Age
6. Gender
7. Height
8. BMI
a Value cannot
3
4
5
6
7
---.40
---
-.07
-.07
-.32
---
.03
.00
a
a
.11
---
-.04
-.05
.25
.28
-.12*
.68***
---
.12*
.13*
.06
-.01
.22**
.07
-.07
-.11
8
be computed because gender is constant
---
No Chocolate Choices
Any Chocolate Choices
60
Percentage
Introduction
Contact: ksaules@emich.edu
50
40
30
20
10
0
Not Overweight
Overweight
►Overweight participants were more likely to select chocolate at least
once on the online questionnaire measure of the RRV-F, while their nonoverweight counterparts were more likely to never select chocolate on
the online questionnaire, X2(1) = 4.57, p = .033.
Discussion
•This was the first study to validate an online questionnaire
measure of the RRV-F
 The online RRV instrument showed strong predictive validity for
laboratory food choice behavior and convergent validity with
overweight status. Discriminant validity was supported by a lack of
association between data from the online measure of the RRV-F
and theoretically unrelated constructs (age, gender, and height).
•Advantages of online assessment of the RRV-F
 The relative privacy of online assessment may be preferable for
assessing eating behavior, given the secretive nature of various
forms of disordered eating (APA, 2000).
Relative to completing a food choice task in the laboratory, online
data collection also requires less time, is more cost-efficient, and
affords greater candor in responding.
More respondents were recruited utilizing this online survey
methodology relative to studies assessing food choice behavior in
the lab (Epstein & Dearing, 2010; Reslan et al., 2011).
•Future Research
Focus on validating this questionnaire with a clinical sample.