QUANTITATIVE SECTION: EXPERIMENTAL DESIGN

I. Overview
A good research question stems from a well-conceived theory. Further, a good
research question dictates the method most appropriate to answer that question. In this
paper I will propose several research questions/hypotheses and provide a rationale for the
chosen methodology. First, I will give a brief overview of the topic under examination,
stereotype threat. Second, I will outline a quantitative study and then a qualitative study
aimed at exploring this topic including details about the participants, data collection,
procedures/measurement, and proposed analyses of the data. Lastly, I will conclude with
possible limitations of the chosen methods and potential contributions to the existing
literature. When specific terminology is used, it will subsequently be identified and
defined in parentheses.
Rationale for examining stereotype threat
As explained in my response to comprehensive exam question one, stereotype
threat is a burgeoning field of research. Because of its deleterious impact, social scientists
have focused on identifying the causes and consequences of this effect. However, is
stereotype threat exclusively negative? How could researchers explore possible positive
outcomes? How might the current investigation demonstrate more ecological validity?
The studies presented in this paper were designed after reviewing the literature and
noting areas not yet exploredspecifically, positive stereotype research and interpersonal
interactions.
Not all stereotypes are negative, thus, one area worth investigating is the
influence of positive stereotypes. While a number of studies have attempted to attenuate
the ill-effects of stereotype threat (e.g., Cohen et al., 2006; Martens, Johns, Greenberg &
Schimel, 2006), few have directly primed participants of a relevant positive stereotype.
One such study which did (Shih et al., 1999) asked Asian-American female participants
to take a math test. They found that the participants primed to recall their gender (and,
consequently, the stereotype that women are not good at math), scored significantly
worse than the non-primed participants. Howeverand most relevant to the proposed
studythe participants who were prompted to remember their Asian heritage (and the
stereotype that Asians are good at math), scored the highest of all groups. As stated by
McGlone et al., Because students belong to many social groups, one means of mitigating
the threat is to remind them of their membership in groups for which there are positive
performance expectations (p. 1, McGlone). Thus, one important goal of these studies is to
address the potentially encouraging outcomes that may result from having individuals
reflect on their positive stereotypes.
A second goal of these studies is to compare the oft-used pen-and-paper
measurement style to that of a face-to-face measurement technique. In their efforts to
uncover stereotype threat anxiety, most researchers utilize self-report scales, with mixed
results (p. 247, Bosson). Only one study has used this interaction approach to my
knowledge (see Bosson et al. 2004). After eliciting stereotype threat the researchers
compared the participants self-reported anxiety levels to their observed anxiety levels

and found significant differences. The researchers concluded that using an indirect
measure to capture stereotype threat anxieties is ideal because respondents cannot easily
control their responses to them, and they may capture processes of which respondents are
unaware (need reference, see also Dovidio & Fazio, 1992). These proposed studies will
compare the effectiveness of measuring peoples nonverbal behaviors with that of selfreport scales to see which is more accurate at detecting anxiety. It may be that people say
one thing, but feel another; thus, the existing stereotype threat research may be greatly
under-reporting its effects.
II. Quantitative section
Underlying all research projects are assumptions about what constitutes
knowledge. These epistemological assumptions are prerequisites for developing criteria
that can assess the importance and worth of the data gathered. Many communication
scholars have increasingly embraced epistemological assumptions consistent with a
scientific approach in the last fifty years. Adopting a scientific perspective inevitably
leads to quantitative thinking when defining and measuring concepts, describing
relationships between variables in precise mathematical terms, and assessing the merit of
hypotheses.
This first quantitative study will employ a 2 x 2 x 2 between-participant factorial
design experiment. It will look to test the aforementioned constructs, namely, positive
stereotype threat and the use of face-to-face measures. Before detailing the hypotheses,
proposed measurements and data analysis, a description of participant recruitment and
procedures will be provided.
Participants
Students enrolled in an introductory public speaking course at the University of
Texas will be recruited through their classes and offered extra credit for participation in
this study. They will sign up for a thirty minute appointment that is convenient for their
schedule but organized by the researchers. On their set date, the research team will
explain that participation is strictly voluntary; those that do not wish to complete the
surveys and interviews will have other opportunities to earn extra-credit points in the
semester. Once consent forms have been signed and collected, participants will be given
one of four possible written descriptions to read, based on their gender (for a total of
eight cells). Stratified random assignment will be used to ensure that equal numbers of
men and women are distributed into each of the eight treatment cells, four designed
exclusively for women, four intended only for men. Ideally, there will be 20 men and 20
women in each of the four condition sets (each set having two cells), thus the total
number of participants will be 160. Their approximate age will likely be between 18 and
22, but no attempt will be made to sample a particular type of student in terms of race,
health, religion, or attitude.
Procedure
Each of these eight cells will represent different manipulations of the independent
variables. The first independent variable will be the gender of the participant (either male

or female). Each of four conditions will represent different manipulations of the other two
independent variables. The second independent variable will be the explicitness of the
stereotype (either explicit or implicit).the type of test (either one aimed at assessing
Social Intelligence or Logical Intelligence), , and the third independent variable will be
the type of test (either assessing Social Intelligence or Logical Intelligence) the
explicitness of the stereotype (either explicit or implicit). Therefore, the four conditions
are as follows: (1) Explicit, SI, (2) Explicit LI, (3) Implicit, SI, and (4) Implicit, LI. To
provide more clarity, here is a brief description of each cell:
Condition 1: Explicit, Social Intelligence
Cell 1: A female participant will read a paragraph which reminds her that women, in
general, do not perform well on tests of Logical Intelligence. She will then take a verbal
Logical Intelligence test. This description will be quite explicit and negative (thus
mirroring how previous stereotype threat researchers have operationalized the construct).
Cell 2: A male participant will read a paragraph which reminds him that men, in general,
do not perform well on tests of Social Intelligence. He will then take a verbal Social
Intelligence test. This description will be quite explicit and negative, and, just as in
Condition 1, will be similar to how other scholars have elicited stereotype threat.
This first condition represents the way stereotype threat research has been
conducted in the past. They are necessary to include in this study for the sake of
comparison to the next three conditions.
Condition 2: E
Cell 3: A female participant will read a paragraph which reminds her that women, in
general, do perform well on tests of Social Intelligence. She will then take a verbal Social
Intelligence test. This description will be explicit, but unlike the aforementioned
condition, it will be a positive stereotype prime.
Cell 4: A male participant will read a paragraph which reminds him that men, in general,
do perform well on tests of Logical Intelligence. He will then take a verbal Logical
Intelligence test. This description will also be an explicit positive stereotype prime.
While these two conditions represent the extremestwo cells explicitly priming
negative stereotypes and two cells explicitly priming positive stereotypes, there is still a
need to test the influence of more subtle stereotype threat priming effects. These last four
cells manipulate the same paragraph used in the previous conditions, but attempt to
remind participants of their negative/positive stereotype through implicit descriptions.
The descriptions will emphasize the strengths/weaknesses of their opposite gender, not
their own. The goal of these last two conditions is to test whether stereotype threat
priming is effective, even when it is indirectly described.
Cell 5: A female participant will read a paragraph which reminds her that men, in general,
do perform well on tests of Logical Intelligence. She will then take a verbal Logical

Intelligence test. This description is implicitly negative because, while never mentioning
anything about women, it does imply a comparison.
Cell 6: A male participant will read a paragraph which reminds him that women, in
general, do perform well on tests of Social Intelligence. He will then take a verbal Social
Intelligence test. This description is also implicitly negative.
This condition is quite similar to the first condition in that they are both negative
reminders of stereotypes; however, this condition is indirect. The participant has to make
the mental comparison about his/her own gender group, it is not explicitly stated. This
final condition also presents implicit descriptions, but, similar to the second condition, it
is aimed at prompting positive stereotypes.
Cell 7: A female participant will read a paragraph which reminds her that men, in general,
do not perform well on tests of Social Intelligence. She will then take a verbal Social
Intelligence test. This description is intended to prime a positive stereotype in women but
in a more indirect, implicit manner.
Cell 8: A male participant will read a paragraph which reminds him that women, in
general, do not perform well on tests of Logical Intelligence. This final cell is also
implicit and positive.
Thus, this eight-cell factorial design suggests distinct hypotheses:
Hypotheses
H1: Participants in cells 1 and 2 (who will be explicitly primed of a negative stereotype)
will perform worse than the participants in all other cells.
H2: Participants in cells 3 and 4 (who will be explicitly primed of a positive stereotype)
will perform better than the participants in all other cells.
H3: Participants in cells 5 and 6 (who will be implicitly primed of a negative stereotype)
will perform worse than the participants in all other cells, except those in cells 1 and 2.
H4: Participants in cells 7 and 8 (who will be implicitly primed of a positive stereotype)
will perform better than the participants in all other cells, except those in cells 3 and 4.
H5: Self-report measures will be less effective in capturing stereotype threat anxiety than
observed measures.
Measurements
This study will conceptualize Logical Intelligence as a typical male characteristic
and Social Intelligence as a typically female characteristic. Drawing on previous
research, it does appear that these assumptions are stereotypes of which the general
populous is aware (Cadinu et al, 2006). In the cells which explain that a verbal test of
Logical Intelligence will follow, a definition will be provided which will read: Logical
Intelligence in a persons ability to handle information in a rational way in order to find
the best solution in everyday life situations. People with high Logical Intelligence (a)
possess good logic and mathematical reasoning abilities, (b) are able to employ adequate

strategies in problem-solving, and (c) show high levels of accuracy and rationality in
information processing.
For those cells describing a Social Intelligence verbal test, the definition will read:
Social Intelligence is a persons ability to connect with the external environment to
adapt to the different circumstances of life. People with high Social Intelligence (a) are
able to capture and organize the cues from the environment in order to produce the most
adequate behavior for each situation, (b) are able to predict other peoples behavior, and
(c) are able to understand the motivations underlying other peoples behavior. Both of
these definitions were taken from the recently published (2006) Cadinu et al. study which
also sought to assess Social and Logical Intelligence.
After reading one of the eight scenarios (four appropriate for females, four
appropriate for males), participants will then meet the experimenter in an adjacent room
where their interactions will be videotaped. This experimenter will be same for all
participants. She will be blind to the participants condition and will proceed to
administer the verbal Social/Logical Intelligence test. This test will be identical for all
participants regardless of the description they read. The goal of this verbal test will be to
detect (and record) the participants level of anxiety, not necessarily to measure the
accuracy of their responses. In fact, there will be no right or wrong answers to the
verbal test. This experimenter-participant interaction will serve as the mechanism which
prompts the participants to experience and record the dependent variable, the participants
level of anxiety.
This verbal test will simulate an interview. The experimenter will give
information regarding a social/relationship problem or a logical/practical dilemma and
ask the participant what he/she would advise to do. For example, one possible question
will be: Recent studies have found conflicting evidence regarding the relationship
between the number of children in a family and the happiness of the parents. If a friend,
who was considering starting a family, wanted advice on this issue, what would you
suggest he/she do? There will be ten questions in total. The experimenter will only read
aloud the questions (and not provide additional probes) such that all participants will
have the same amount of interaction with the interviewer.
Finally, after completing the verbal Social/Logical test, participants will be asked
to fill out two final scales: the first will be a self-report measure of anxiety, the second
will be a manipulation check test. The anxiety level self-report scales will aim to measure
the degree to which the participant felt comfortable responding to the verbal test after
being explicitly/implicitly primed of a relevant positive/negative stereotype. Using scales
ranging from 1 (not at all) to 7 (extremely), participants will indicate how much they
experienced five different anxiety-related emotions during the task: anxious, worried,
calm, pleasant, and relaxed. Additionally, participants will use the same scale to rate the
extent to which they felt self-conscious, were concerned about how they appear to others,
were able to be themselves, and were able to act natural during the interaction (see
Bosson et al., 2004).
The manipulation check test will be necessarily to ensure that the scenarios
captured their appropriate construct. Participants will be asked a series of questions about
the nature of their verbal test (e.g., The test I have just performed was aimed at
measuring my a) Social Intelligence or b) Logical Intelligence). After completing these
final scales the participant will be free to leave.

Data Analysis
First, participants who did not correctly identify their prescribed test type (either
Social or Logical) will be excluded from the data analysis. Second, participants selfreport anxiety scales will be reverse-coded (where appropriate) and averaged so that an
index of evaluation apprehension can be recorded. Third, coders will watch the
videotaped interactions and identify anxious behaviors. This step will require two judges
blind to all procedures and hypothesesto independently watch each videotaped
interaction and rate each participants nonverbal anxieties (e.g., fidgeting, chewing on lip,
playing with hair, nervous smiling, stiff posture, etc.). A rating will be made every thirty
seconds and a composite score will be given to each participant from both coders. A test
for intercoder reliability will be conducted to ensure high levels of accuracy in this
interpretation.
Next, a 2 x 2 x 2 ANOVA will be performed where main effects and interaction
effects will be assessed which will either support or fail to support the hypotheses. Lastly,
a regression analysis will be conducted to test the final hypotheses concerning the
accuracy of capturing stereotype threat anxiety via self-report or observed measures.