Animation OMB Supporting Statement PART B

Animation in Direct-to-Consumer Advertising
OMB Control No. 0910-NEW
SUPPORTING STATEMENT

B.

Statistical Methods (used for collection of information employing statistical methods)
1. Respondent Universe and Sampling Methods

The survey sample will be drawn from Research Now’s database. Research Now’s opt-in
online survey panel is demographically balanced, including racial and ethnic minorities, a wide
range of different age groups, and individuals with relatively less educational attainment. They
recruit panel members through a combination of e-mail, online marketing, and by invitation, with
over 300 diverse online and offline affiliate partners and targeted website advertising. By using
multiple recruitment methods, Research Now is able to recruit a diverse set of representative
consumers and decision makers to participate in their panels. Panel inclusion is by invitation
only, and Research Now invites only pre-validated individuals with known characteristics to
participate in the consumer panels.
Using Research Now’s database, we will recruit 1,800 individuals for the pretest and
main study combined. (See Appendix B for the Screening Instrument). Table 1 shows the
current sample design and sample sizes.
Table 1.

Sample Design and Sample Sizes
Category

Number of Participants

Pretest

300

Main study

1,500

The sample will be drawn from panel members who report a diagnosis of chronic dry eye
or psoriasis. Although Research Now’s database is always growing and changing, the current
demographic distribution of the panel is presented in Table 2.
Table 2.
Demographic Distribution of Sufferers of Chronic Dry Eye and
Psoriasis in Research Now’s Respondent Database
Percent
Demographic Characteristic

Chronic Dry Eye

1

Psoriasis

Percent
Demographic Characteristic

Chronic Dry Eye

Psoriasis

Female

71%

61%

Male

29%

39%

18–24

2%

4%

25–34

19%

20%

35–44

19%

20%

45-54

21%

20%

55-64

22%

20%

65 or over

16%

15%

High school graduate or
less

8%

9%

Some college

24%

26%

College or technical school
graduate

41%

40%

Graduate school

27%

25%

100%

100%

Gender

Age

Education

Total

2. Procedures for the Collection of Information
Part A of the supporting statement described the rationale for conducting the study.
General Research Questions
1. How does consumer processing of a DTC prescription drug ad differ depending on whether
the ad is live-action, rotoscoped, or animated?
2. Does consumer processing differ depending on whether the sufferer, the disease, or the benefit
is the focus of the animation?
Design Overview
2

To test these research questions, we will conduct two experiments. Both experiments
will be examined in two different medical conditions: chronic dry eye, and psoriasis. The mock
drugs we will create for these conditions mimic currently available medications and were chosen
for their variance in serious side effects, i.e., medications for psoriasis have very long, serious
lists of risks and side effects, whereas chronic dry eye medications have relatively few risks and
side effects.
The first experiment will examine whether animation itself influences consumer
processing, defined as consumer recall of risks and benefits, perceptions of risks and benefits,
and attitudes and emotional responses to the ad, the brand, the product, and the character (Table
3. We will examine two different types of animation in addition to a control ad which will be
shot with live actors: an “in-between” animation technique, rotoscoping, in which live scenes are
drawn to look animated, and full animation with nonhuman characters. The live action and
rotoscoped ad will be identical except for the rotoscope treatment. The animated ad will follow
the theme and message as closely as possible within the limitations of animation itself. The
benefits and risks of the product will be identical, although the ad’s storyline may vary somewhat
to account for a nonhuman protagonist.
Table 3. Experiment 1: Animation design.
Type of Animation
Medical Condition
Non-human sufferer Rotoscoped human
sufferer
Chronic Dry Eye
•
•
Psoriasis
•
•

Human sufferer
•
•

The second experiment will examine whether the object of the animation influences
consumer processing of the ad (Table 4), defined as consumer recall of risks and benefits,
perceptions of risks and benefits, and attitudes and emotional responses to the ad, the brand, the
product, and the character. The animation will focus on the animated character who will
personify either the sufferer of the medical condition, the disease itself, or the benefit from the
drug. In this study, all ads will contain the same kind of full animation and the general theme will
be as similar as possible, accounting for the variations in focus of character. The experiments
will be conducted concurrently, and the same participants in the nonhuman sufferer groups will
be part of both.
Table 4. Experiment 2: Personification design.
Non-human Personification
Medical Condition
Sufferer
Disease
Chronic Dry Eye
•
•
Psoriasis
•
•

3

Benefit
•
•

In both cases, a professional firm will create all ads such that they are indistinguishable
from currently running DTC ads.
Pretesting will take place before the main study to evaluate the procedures and measures
used in the main study. We will recruit adults who have experienced chronic dry eye or psoriasis.
We will exclude individuals who work in healthcare or marketing settings because their
knowledge and experiences may not reflect those of the average consumer. A priori power
analyses revealed that we need 300 participants for the pretest to obtain 80% power to detect a
moderately small effect size. Each experiment will include 30 participants per condition for a
total of 180 participants each, but 60 of those in the nonhuman sufferer conditions will overlap
between the two experiments. We will need 1,500 unique participants for the main study to
obtain 90% power to detect a moderately small effect size. There will be 150 participants per
condition for a total of 900 participants in each experiment, with 300 participants in the
overlapping nonhuman sufferer conditions.
In both experiments, participants who have been diagnosed with either chronic dry eye or
psoriasis will be recruited via opt-in Internet panel to watch one ad for a prescription drug that
treats their medical condition. In experiment 1, participants will be randomly assigned to view
either a live-action, rotoscoped, or fully animated ad. All themes in experiment 1 will focus on
the main character as the sufferer of the condition. In experiment 2, participants will be randomly
assigned to a personification condition: sufferer, disease, or benefit. All ads in experiment 2 will
be fully animated. Participants will watch the ad once and then answer an online survey with
questions addressing recall of risks and benefits, perceptions of risks and benefits, and attitudes
and emotional responses to the ad, the brand, the product, and the character. The questionnaire is
available upon request. Participation is estimated to take approximately 25 minutes.
Specific Research Questions and Hypotheses
We will focus on answering the following research questions in Experiment 1:
Research Question 1A.

Does animation in DTC television advertisements influence the
processing of prescription drug information?

Research Question 1B.

Does the impact of animation on processing of prescription
drug information in DTC television advertisements vary by
medical condition?

Additionally, we will test hypotheses related to several of our dependent variables. Although we
incorporate two medical conditions into the study design, we view analyses involving the
medical-condition factor as essentially exploratory. Characteristics of the medication profiles and

4

ads will not be controlled across medical conditions. As a result, significant effects by medical
condition — especially interaction effects — would be open to multiple interpretations. With this
in mind, our theoretical rationale for Experiment 1 focuses on advertising effects by different
types of animation.
Experiment 1 hypotheses stem from theory and previously observed relationships between
animation, attitudes, and information processing. Animated characters are often used to grab
attention, increase ad memorability, and enhance persuasion to ultimately drive behavior1.
However, there is reason to expect that different types of animation techniques will elicit
different reactions from an audience. Following Clayton and Leshner2, we expect that an ad
featuring a rotoscoped human character will activate an avoidance response, leading participants
to develop an unfavorable attitude toward the character and give less attention to the ad. The
mechanism underlying these effects draws from Uncanny Valley Theory3, which argues that
characters that closely resemble human beings, but are eerily unnatural in movement or
appearance, evoke discord in the viewer and a sense of revulsion. In turn, the unpleasant
emotional response evoked by the rotoscoped character is expected to activate the aversive
motivational system4, leading to withdrawal from the ad, lower attention, and reduced memory
of message content5. An animated nonhuman character, on the other hand, is not likely to evoke
the eerie feelings expected of a rotoscoped human character.
Lastly, research on the affect heuristic provides a basis for Experiment 1 hypotheses concerning
the influence of participant attitudes on perceptions of risk and benefit. Here, affect refers to “the
specific quality of goodness or badness (a) experienced as a feeling state (with or without
consciousness) and (b) demarcating a positive or negative quality of a stimulus”6. Notably, by
this definition, affect is conceptually indistinct from attitude, or “a psychological tendency

1

Bell JA. Creativity, TV commercial popularity, and advertising expenditures. International J Adv.
1992;11(2):165–172; Diao F, Sundar, SS. Orienting response and memory for Web advertisements: Exploring
effects of pop-up window and animation. Communication Res. 2004;31:537–567; Fox J, Lang A, Chung Y, Lee S,
Schwartz N, Potter D. Picture this: Effects of graphics on the processing of television news. J Broadcasting
Electronic Media. 2004;48:646–674; Garettson JA, Neidrich RW. Spokes-characters: Creating character trust and
positive brand attitudes. J Adv. 2004;33(2):25-36; Heiser RS, Sierra JJ, Torres IM. Creativity via cartoon
spokespeople in print ads. J Adv. 2008; 37(4):75-84; Leiner M, Handal G, Williams D. Patient communication: a
multidisciplinary approach using animated cartoons. Health Educ Res. 2004;19(5):591-595; Luo JT, McGoldrick P,
Beatty S, Keeling K A. (2006). On-screen characters: their design and influence on consumer trust. J Services
Market. 2006;20(2):112-124.  
2
Clayton RB, Leshner G. (2015). The uncanny valley: The effects of rotoscope animation on motivational
processing of depression drug messages. J Broadcasting Electronic Media. 2015;59(1):57-75.
3
Mori M. (1970/2012). The uncanny valley (K. F. MacDorman & Norri Kageki, Trans.). IEEE Robotics and
Automation. 1970/2012;19:98–100. doi:10.1109/MRA.2012.2192811.
4
Cacioppo J T, Gardner WL, Berntson GG. The affect system has parallel and integrative processing components:
Form follows function. J Personality Soc Psychol. 1999;76:839-855.
5
Lang A. The limited capacity model of mediated message processing. J Communication. 2000;50:46–70.
6
Slovic P, Peters E, Finucane ML, MacGregor DG. (2005). Affect, risk, and decision making. Health Psychol.
2005;24:S35–S40, p S35.

5

expressed by evaluating a particular entity with some degree of favor or disfavor”7. The affect
heuristic refers to a mental short-cut whereby judgments about an object (e.g., perceived risk and
benefit) are based on a readily available affective impression of it, rather than retrieval and
integration of information about the object’s relevant attributes and features. In short, lacking
sufficient motivation or ability, people who feel favorably about a stimulus will make judgments
and decisions aligned with that positive affect (e.g., greater perceived benefits, fewer risks);
people who feel unfavorably about it will make judgments and decisions aligned with that
negative affect (e.g., lower benefits, greater risks)8. Indeed, the affect heuristic has been cited as
an explanation for the documented inverse relationship between perceived risks and benefits9. In
this study, the affect heuristic also has implications for the transference of evaluative judgments
from one object (e.g., advertising character) to another object (e.g., advertisement, project).
Benefit Recall and Recognition
Hypothesis 1.1

Participants who see the rotoscoped ad will show lower recall and
recognition of benefit information than those who see the nonhuman
sufferer ad or the live-action ad (i.e., no animation).

Specific RQ 1.1

Will participants who see the nonhuman sufferer ad show greater recall
and recognition of benefit information than those who see the live-action
human sufferer ad (i.e., no animation)?

Risk Recall and Recognition
Hypothesis 1.2

7
8

Participants who see the rotoscoped ad will show lower recall and
recognition of risk information than those who see the nonhuman
sufferer ad or the live-action ad (i.e., no animation).

Eagly AH, Chaiken S. The psychology of attitudes. 1993. Ft. Worth, TX: Harcourt, Brace, & Janovich.
An important assumption underlying this logic is that basing these risk and benefit judgments directly on the risk
and benefit information given in the ad would require more effort than participants are able or willing to
expend. The affect heuristic is less likely to come into play when, for example, risk and benefit information
is readily available and people have sufficient ability and motivation to integrate that information when
making a judgment about risk/benefit (e.g., Sloman SA. (2002). Two systems of reasoning In Gilovich T,
Griffin D, Kahneman D (Eds.), Heuristics and biases: The psychology of intuitive judgment. 2002.
Cambridge, UK: Cambridge University Press.).

9

Alhakami AS, Slovic P. A psychological study of the inverse relationship between perceived risk and perceived
benefit. Risk Analysis. 1994;14(6):1085-1096; Finucane ML, Alhakami A, Slovic P, Johnson SM. The affect
heuristic in judgments of risks and benefits. J Behav Decision Making. 2000;13(1): 1-17.

6

Specific RQ 1.2

Will participants who see the nonhuman sufferer ad show greater recall
and recognition of risk information than those who see the live-action
human sufferer ad?

Overall ad comprehension
Hypothesis 1.3

Participants who see the nonhuman animated ad will show greater
overall ad comprehension than those who see the live-action ad.
Participants who see the rotoscoped ad will show lower comprehension
than those who see the live-action ad.

Perceived Benefits
Hypothesis 1.4

Participants who see the nonhuman animated ad will have greater
perceived benefit than those who see the live-action ad. Participants
who see the rotoscoped ad will have lower perceived benefit than those
who see the live-action ad.

Perceived Risks
Hypothesis 1.5

Participants who see the nonhuman animated ad will have lower
perceived risk than those who see the live-action ad. Participants who
see the rotoscoped ad will have greater perceived risk than those who
see the live-action ad.

Attitudes
Hypothesis 1.6.

Participants who see the nonhuman animated ad will show more
positive attitudes toward the character, the ad, and the product than
those who see the live-action ad. Participants who see the rotoscoped ad
will show more negative attitudes toward these objects than those who
see the live-action ad.

Intentions
Hypothesis 1.7.

Participants who see the nonhuman animated ad will show greater
product-related behavioral intentions than participants who see the liveaction ad. Participants who see the rotoscoped ad will have lower
intentions than those who see the live-action ad.

7

Experiment 2 Research Questions and Hypotheses
In Experiment 2 we will test the effects of different types of nonhuman personification in DTC
advertisements on key message processing outcomes. Here nonhuman personification refers to
the use of nonhuman animated characters to personify constructs related to prescription
medication (i.e., sufferer, disease, benefit). These analyses will be designed in response to the
following research questions:
Research Question 2A.

Does nonhuman personification in DTC television
advertisements influence processing of prescription drug
information?

Research Question 2B.

Does the impact of nonhuman personification on the
processing of prescription drug information in DTC television
advertisements vary by medical condition?

Our theoretical rationale for the Experiment 2 hypotheses focuses on advertising effects by
different types of nonhuman personification. As in Experiment 1, analyses involving the
medical-condition factor will be exploratory. We have no reason to expect that medical condition
will moderate the effects of nonhuman personification on information processing outcomes.
In addition to psychological mechanisms outlined under Experiment 1, our hypotheses for
Experiment 2 draw on outcomes related to identification with animated characters in an
advertisement. Identification is a cognitive and emotional process whereby an audience member
adopts a stance of empathy toward a character, takes on the character’s perspective and goals,
and experiences a temporary loss of self-awareness10. Identification with a character leads people
to become deeply absorbed in a media text and take a less critical stance toward it11. One
antecedent of identification noted by Cohen (2001) is the similarity of audience members to the
character. For example, feelings of similarity may be brought about when an audience relates to
a character by virtue of a common experience or situation, like suffering from the same medical
condition. With this in mind, we expect identification to be higher in the sufferer condition than
either the benefit or disease conditions. In turn, we would expect people who experience stronger
identification to adopt a stance toward perceived drug risk and benefit that aligns with the
character’s point of view. In the broader context of direct-to-consumer advertising for
prescription drugs, characters act in pursuit of finding relief from the signs, symptoms or
consequences of a medical condition. Thus, we would expect stronger identification with a
character in a prescription drug ad (e.g., adopting the character’s goals and perspective) to orient
the audience toward drug benefits and away from risks. Further, because stronger identification

10

Cohen J. Defining identification: A theoretical look at the identification of audiences with media characters. Mass
Communication Society. 2001;4(3):245-264.
11
Fiske J. Television culture. 1989. London: Routledge.

8

is associated with greater action, we expect participants in the sufferer condition to have stronger
drug-related behavioral intentions12.
Identification with the Character
Hypothesis 2.1

Participants in the sufferer condition will show greater identification
with the character than those in the benefit-personification or disease
conditions.

Benefit Recall and Recognition
Hypothesis 2.2

Participants in the benefit-personification condition will show greater
recall and recognition of benefit information than those in the sufferer or
disease conditions.

Risk Recall and Recognition
Specific RQ 2.1Will recall and recognition of risk information differ by nonhuman
personification condition?
Overall Ad Comprehension
Specific RQ 2.2

Will overall ad comprehension differ by nonhuman personification
condition?

Perceived Benefits
Hypothesis 2.3

Participants in the sufferer and benefit-personification conditions will
show greater perceived benefit than those in the disease-personification
condition.

Perceived Risks
Hypothesis 2.4

12

Participants in the sufferer and benefit-personification conditions will
show lower perceived risk than those in the disease-personification
condition.

Basil MD. Identification as a mediator of celebrity effects. J Broadcasting Electronic Media. 1996;40: 478-495.

9

Attitudes
Hypothesis 2.5

Participants in the sufferer and benefit-personification conditions will
show more positive attitudes toward the character, the ad, and the
product than those in the disease-personification condition.

Intentions
Hypothesis 2.6

Participants in the sufferer condition will show greater product-related
behavioral intentions than participants in the benefit-personification and
disease conditions.

Analysis Plan
This analysis plan describes our approach to answering the study’s research questions and
exploring relationships between variables. The proposed analysis will consist of three steps: (1)
descriptive data analysis, (2) hypothesis testing, and (3) multivariate modeling.
Descriptive Analysis
During descriptive analysis, we will calculate frequency distributions and check the
apparent validity of the data (i.e., range checks, frequency of missing responses, or response
distribution). For continuous/ordinal variables, statistical output will include means, medians,
standard deviations, ranges, and counts. For categorical variables, output will include counts and
percentages.
In addition to frequency distributions, we will conduct three other types of analyses
during this step. First, we will calculate reliability of composite variables and multi-item scales
to determine if the individual items hang together as composite measures. Specifically, we will
calculate Cronbach’s alpha for each composite variable. If alpha for a composite measure or
scale does not meet our pre-established threshold of 0.75, we will discuss whether to use singleitem measures rather than the composite or to consider such composites as indices (because of a
theoretical reason to consider an aggregate measure regardless of item correspondence) in
hypothesis testing.
Second, we will conduct a content analysis of responses to the open-ended risk and
benefit recall questions. We will develop a codebook to guide classification of responses based
on their match with risk and benefit claims made in the chronic dry eye and psoriasis ads. To
ensure consistent and reliable coding of open-ended data, we will develop and implement an
inter-rater reliability protocol before proceeding to code the full content.

10

Finally, we will conduct a non-response analysis to determine if individuals who do not
respond to the study’s invitation differ from those who complete the study. We will compare
responding individuals to invited but non-responsive individuals on key demographics—such as
age, sex, race, and education—to see if significant differences exist. Specifically, we will
conduct t-tests comparing the proportions of respondents and non-respondents using a standard
significance threshold of p=0.05.
Hypothesis Testing
We will test hypothesized relationships implied by our central research questions by
conducting one of several statistical tests as outlined below. In most cases, we plan to conduct an
overall test of the relationship between the independent and dependent variables and then
conduct hypothesis-specific planned comparisons to assess whether the data support predicted
differences among experimental groups. We do not have specific hypotheses concerning
interaction effects by medical condition with type of animation or nonhuman personification. If a
significant interaction is observed, we will conduct follow-up analyses to describe the
interaction. Foremost, we will test for the predicted pattern of means across manipulated
experimental conditions (i.e., type of animation in Experiment 1; nonhuman personification in
Experiment 2) in each medical condition. We will develop planned contrast equations for this
purpose corresponding to each of our research hypotheses.
For hypotheses examining continuous or scale outcomes (e.g., perceived benefit,
behavioral intentions), we will conduct two-way ANOVAs to detect significant relationships.
For Experiment 1, we will test for effects by type of animation (animated nonhuman sufferer,
rotoscoped human sufferer, live-action human sufferer), medical condition (CDE, psoriasis), and
their interaction. Experiment 2 will test for effects by nonhuman personification (sufferer,
benefit, disease), medical condition (CDE, psoriasis), and their interaction. Statistical output will
include F statistics, degrees of freedom, p values, mean differences, and standardized effect sizes
(e.g., Cohen’s d) for the main effects of each independent variable as well as any interaction
effects. We will conduct planned comparisons based on hypothesized relationships to identify
significant differences between specific experimental groups. An example template for ANOVA
output is shown in Exhibit 4.
Multivariate Modeling
Our descriptive analyses may reveal opportunities for exploring whether the effects of
type of animation (Experiment 1) and nonhuman personification (Experiment 2) are influenced
by additional variables. Specific plans for multivariate models that control for additional
variables, test for complex moderation, or test for mediation will be discussed with FDA based
on the hypothesis testing results. The plans will include a rationale for selecting potential
covariates, mediators and/or moderators, procedures for verifying statistical assumptions (e.g.,

11

normal distribution, homogeneity of variance, parallel regression lines), and a description of the
proposed modeling approach.

Power
The following assumptions were made in deriving the sample size for the main studies:
(1) 0.90 power, (2) 0.05 alpha level for main effects and interactions or 0.008 for post-hoc
pairwise comparisons, and (3) a small-to-medium effect size. We use Cohen’s conventional
thresholds for interpreting the magnitude of effect sizes13. Effects with f values in the order of
0.40 and greater are large, from 0.25 to but not including 0.40 are medium, and from 0.10 to but
not including 0.25 are small. Corresponding effects measured with d statistics equal to or greater
than 0.80 are large, from 0.50 to but not including 0.80 are medium, and from 0.20 to but not
including 0.50 are small. For continuous dependent variables in the main study experiments, we
will conduct a set of two-way analyses of variance (ANOVAs) and planned contrasts to test for
significant differences among the six experimental groups in a 2 × 3 factorial design. Results
from a sensitivity analysis suggests that, with 150 participants in each experimental group (N =
900 per experiment), omnibus F tests will be able to detect moderately small differences among
groups (f = 0.14). The main study design is also sensitive to detect moderately small differences
for up to six non-orthogonal planned contrasts (f = 0.13) or six post-hoc pairwise comparisons (d
= .46), assuming a Bonferroni-adjusted alpha of 0.0083.
For analyses involving discrete outcome variables (e.g., correctly understood risks vs.
incorrectly understood risks), the proposed main-study sample size will allow detection of an
absolute difference of 18 percentage points in group-to-group comparisons (e.g., a difference
between 68% in one experimental group versus 50% in another) with a power of 0.90. In this
calculation, we assumed equal-sized samples in each arm (n = 150), a design effect equal to 1, an
alpha level of 0.05, a two-sided Fisher’s exact test, and an underlying proportion of study
participants in a particular response category equal to 0.50. An underlying proportion of 0.50 is
the most conservative estimate and overestimates the sample size relative to alternate
proportions.
We will conduct the pretests with a smaller sample size than the main study. The objective
of the pretest is to confirm that the entire survey process runs smoothly and that the stimulus will
be effective for the study design, not to test the hypotheses. The sample size will be large enough
to pretest the stimuli and data collection process thoroughly. The pretest experiments – assuming
the same power and alpha levels as the main study – omnibus F tests and up to six nonorthogonal planned contrasts will be sensitive to detect medium-to-large effects (f = 0.31 and f =
0.30, respectively).
3. Methods to Maximize Response Rates and Deal with Non-response
13

Cohen J. A power primer. Psychol Bull. 1992;112:155-9.

12

Both the pretests and main survey will use an existing Internet panel to draw a sample.
The panel (described in B.1) comprises individuals who share their opinions via the Internet
regularly. To help ensure that the participation rate is as high as possible, FDA and the contractor
will:
• Design a protocol that minimizes burden (short in length, clearly written, and with
appealing graphics);
• Administer the survey over the Internet, allowing respondents to answer questions at
a time and location of their choosing;
• Email a reminder to the respondents who do not complete the protocol within 24
hours of the original invitation to participate.
In the absence of additional information, response rates are often used alone as a proxy
measure for survey quality, with lower response rates indicating poorer quality. However, lower
response rates are not always associated with greater nonresponse bias14. Total survey error is a
function of many factors, including nonsampling errors that may arise from both responders and
nonresponders15. A nonresponse bias analysis can be used to determine the potential for
nonresponse bias in the survey estimates from the main data collection.
There are several approaches to address the potential for nonresponse bias analysis in this
study, such as comparing response rates by subgroups, comparing respondents and
nonrespondents on frame variables, and conducting a nonresponse follow-up study16. For the
proposed project, we will perform two steps: comparing response rates on subgroups and
comparing responders and nonresponders on frame variables.
We will first identify the subgroups of interest, such as age and gender. At the end of the
data collection, we will calculate response rates by subgroup. If the response rates are the same
within subgroups, then nonresponse bias should not affect the results related to those group
categories. For example, if the response rate for males and females is the same, then there will
not be a large nonresponse bias in the survey estimates for gender.
To the extent that information is available about all sample cases on the frame and that
information is associated with the key survey estimates, this approach can provide additional
information about the potential for nonresponse bias. At the end of data collection, we will
review the sampling frame to determine if any variables are associated with the key survey
estimates, such as age. We will then compare the frame information for the full sample compared

14

 Groves R. Nonresponse rates and nonresponse bias in households. Public Opinion Quarterly. 2006;70(5): 646–
675. 
15
 Biemer P, Lyberg, L. Introduction to survey quality. 2003. New York: Wiley. 
16
Office of Management and Budget, Standards and Guidelines for Statistical Surveys, September, 2006.
www.whitehouse.gov/sites/default/files/omb/inforeg/statpc. Last accessed April 18, 2013.

13

with respondents only. Differences between the full sample and the respondents are an indicator
of potential bias.
4. Test of Procedures or Methods to be Undertaken
Eighteen (2 waves of 9 each) cognitive interviews will have been conducted to assess
questionnaire flow and wording. After this round of cognitive testing, we plan to conduct
pretests on a larger scale to ensure the main study will run smoothly. We propose to test 300
individuals in the pretest.
5. Individuals Consulted on Statistical Aspects and Individuals Collecting and/or Analyzing
Data
The contractor, RTI, will collect and analyze the data on behalf of FDA as a task order
under Contract HHSF223201510002B. Bridget Kelly, Ph.D., 202-728-2098, is the Project
Director for this project. Data analysis will be overseen by the Research Team, Office of
Prescription Drug Promotion (OPDP), Office of Medical Policy, CDER, FDA, and coordinated
by Amie C. O’Donoghue, Ph.D., 301-796-0574, and Kevin R. Betts, Ph.D., 240-402-5090.

14