Thursday, March 19, 2009

Return to the IAT (Native/White American)

I know we covered this a few weeks ago, but i've had some time over break to get back to some questions and ideas I had for posts.
Here are some examples of images used in this IAT for reference(they can be downloaded from project implicit's website):



I had a few questions about the IAT I took, the native american vs white american test, the one i really was curious about was how they chose the photographs. There were photographs of native americans from the late 1800's to the mid 1900's, i'd guess no later than the 40's, and they were all sepia or b&w. The photos of the white americans were equally old, i'd say from the late 1800's to the early 1900's, maybe 1910. Then, in contrast, to show locations both in america and around the world, they used recent vivid color photographs that would be used for travel brochures or postcards. The goal of the test was to see which group of Americans I felt more comfortable with or automatically associated American identity with, and because of the age of the photographs, I felt equally alien to both groups. The result of my test was a preference for neither, and I think it may be because i felt equally distant from both. The photographs didn't remind me of real life, i.e. neither group made me think "america", instead i thought of old horror movies where they show spooky old photos talking or something. Because of that reaction, I really wanted to
know why they chose those images. I was checking out the project implicit site, and found a section where they share pdfs of published and unpublished research and articles written on the tests. I am going to assume that the sources were valid, because I don't think the IAT creators would put up unreliable information on their site, it would reflect badly on the acceptance of the IAT's accuracy. I found the only article that was written specifically about the IAT i took. I figure that the article, approved of and posted by Project Implicit, could help explain their choices, because obviously they reviewed it before posting, and would have corrected any information about their test or test choices if it was inaccurate.

The article, "Aliens in their Own Land? Implicit and Explicit Ascriptions of National Identity to Native Americans and White Americans", written by
Thierry Devos (San Diego State University), Brian A. Nosek(University of Virginia), and Mahzarin R. Banaji (Harvard University), was available for pdf download at: http://www.projectimplicit.net/articles.php, is a research/lab text written about the results of 2 year's worth of White/Native American IAT tests. The parts that helped to answer my questions I will post in a comment. I focused on the introduction, abstract, and methods:procedure for how the images were chosen. The simplified explanation to explain the choices were that in America, there is a stereotype that american is thought of as white, which is supported by a majority of American representations in pop culture as white, and the connection of american symbols, like a flag, with white people. When choosing the images, they chose photos that did not look like regular americans, because they wanted to see how strongly images that are of "white" people would be associated with American, and because of that they chose the weird old photos of, well, white people. Basically, they wanted to pick photos of white people that did not really look like modern, everyday americans, and instead chose pictures that would look strange and unfamiliar, so that they could tell our implicit association of white skin tone as American. Then, partly to match the age/time of the white photos, they chose the Native American images. They discuss the problem with stereotypical representations of native americans, as simplified images, or in full costume, and they wanted to show Native Americans in an easy to identify way, almost in a way to emphasize the stereotype in order to get a more noticeable reaction. The images of the locations were chosen to show American locations that had more significance to Native Americans, and that would be associated with Native Americans, and there is no mention as to why they were in color or modern.

Now the face image choices make much more sense to me. Knowing the reasoning behind the choices, and the goals of the experimentation helps to make sense of the IAT. I think that the oversimplification of the information given at the end of the test, with the results, is confusing and that they should put more effort into a rewrite, that helps to explain what the results mean and what or why they are using them for. I would be curious to see a different version of a white vs native american test, showing modern people in modern dress, maybe even in color, so as to see what kind of implicit associations someone would have in real life, so that the results could have some relevance to a real-life situation, and maybe even a study of the modern white/native IAT with how people react to white or native americans in everyday life. After digging through the IAT test site to find the link to project implicit main site, i was able to find alot of information and resources, and once I figured out where everything was I was able to answer my own question, I am impressed with how much information they give, and how easily accessible it is.

4 comments:

Lily Probst said...

Here is the text i referenced in my post, if you are interested it can be downloaded on the project implicit site, it is listed in the first catagorey (unpublished manuscripts). I did not include the results/conclusion because i felt it did not have anything to do with the questions I had on procedure. There is lengthy breakdown of results, data charts, graphs, tables, and conclusions in the full text.
Aliens in their Own Land? Implicit and Explicit Ascriptions of National Identity to Native Americans and White Americans
by Thierry Devos, Brian A. Nosek, Mahzarin R. Banaji

Abstract
Legally, ethnic groups in the United States do not differ in the extent to which they are granted
citizenship: all American citizens are equally American. But do ethnic groups differ in the extent
to which they are psychologically granted a national identity? Prior work on this issue has
focused mostly on Asian Americans. Here we look at the national identity as it pertains to the
original Americans, Native Americans. A sample of Internet users (N = 44,878) completed an
Implicit Association Test (IAT) assessing the direction and strength of associations between
Native Americans and White Americans and the attribute American relative to foreign. In
contrast to explicit ascriptions of national identity, Native Americans were less strongly
associated to the concept American than were White Americans. This effect was displayed by
all ethnic groups, except by American Indian or Alaskan Native respondents. In addition, the
American = White effect was more pronounced among self-identified conservatives than among
self-identified liberals. These results reveal the propensity to implicitly equate American with
White American.

Introduction:
National identities provide individuals with rights and responsibilities and offer
protections that are not granted to foreigners. Legal manifestations of national identity exist in
the form of citizenship and formal documents that support it, but psychological forms of identity
can vary in systematic ways that can support or contradict the legal facts of citizenship. Devos
and Banaji (2005) provided evidence that the quality American was “given” to a lesser extent to
Asian Americans than to White Americans on an implicit measure that assessed the associations
indirectly (see also Cheryan & Monin, 2005). This result could stem from a fact of immigration
timing: as a group Asian Americans arrived after European Americans on the continent and
therefore might be seen to be less strongly American. If that is the case, the result should be
reversed when testing for relative national identity among White and Native Americans, the
latter predating Americans of European descent by being the original Americans. On the other
hand, if implicit national identity continues to favor White Americans, then timing of arrival on
the continent is not the sole basis of psychological granting of national identity. Instead, greater
social or numerical status might produce differences in ascriptions of national identity.
The present research examined to what extent Native Americans and White Americans
were implicitly and explicitly differentiated in terms of their association with the concept
American.1 Partially because of the small numbers of living Native Americans and their
marginalized status in American society, the group tends to be lumped with immigrant
minorities in spite of its historically distinct status (Marker, 2000). Because of their historical
and current positions in American culture, Native Americans are perhaps the most
psychologically interesting social group to pose a test of the propensity of modern Americans to
bestow them with status of American. The present study tests whether American Indians are
indeed American on par with White Americans or “aliens in their own land” (LaFromboise,
1998).
A majority of Americans subscribe to the idea that individuals should not be treated
differentially based on the color of their skin, their origin, or their cultural heritage (Schuman,
Steeh, Bobo, & Krysan, 1997). In line with an explicit commitment to equality, most Americans
hold an inclusive and pluralist definition of the American identity (Citrin, Wong, & Duff, 2001).
From this perspective, and considering the fact that Native Americans were recognized U.S.
citizenship in 1924 with the passage of the Indian Citizenship Act (Deloria, 1988), the American
identity should be equally ascribed to Native Americans and White Americans.
This being said, other considerations may undermine the extent to which Native
Americans are seen as an integral part of America. First, American Indians and Alaskan Natives
are a relatively small ethnic minority (approximately 2.8 million people, less than 1% of the
total population) in the U.S. (U.S. Department of Commerce, Census Bureau, 2004). Numerical
status or distinctiveness is known to play a role in the extent to which groups serve as cultural
defaults (Hegarty & Pratto, 2001; Smith & Zárate, 1992; Stroessner, 1996; Zárate & Smith,
1990). Accordingly, Native Americans could be seen as less representative of the category
American than White Americans or even other ethnic minorities (African, Asian, and Latino
Americans).
American Indians have also experienced long oppressions by other arriving in North
America. By the end of the 18th century, warfare and diseases had reduced the population of
native people to 10% of its original size (Oswalt, 1988). Forced relocation to Indian reservations
and removal of children from their families to boarding schools had a profound impact on this
group (Kawamoto, 2001). Moreover, American Indians continue to face limited educational and
economic opportunities (Belcourt-Dittloff & Stewart, 2000). Recent research lends support to
the idea that the relative status of these groups shape intergroup images (Alexander, Brewer, &
Livingston, 2005): White college students hold an image of Natives as dependent, whereas
college students from a Native American nation hold an imperialist and barbarian image of
Whites. In addition, reminding White Americans of violence perpetrated against Native
Americans intensified the extent to which they infra-humanized Native Americans (Castano &
Giner-Sorolla, 2006). Taking into account the hierarchical nature of ethnic relations (Sidanius,
Feshbach, Levin, & Pratto, 1997; Sidanius & Petrocik, 2001), leads to the prediction that White
Americans are more likely to be thought of as being prototypical of the category American than
are Native Americans.
Finally, depictions of American Indians frequently reinforce common misconceptions
(Bell, Esses, & Maio, 1996; Corenblum & Stephan, 2001; Mihesuah, 1996; Vorauer, Main, &
O’Connell, 1998). Stereotypes rely heavily on the notion that Native Americans are “exotic
others.” In U.S. History textbooks, Native Americans are often portrayed as if they were extinct
or as a different culture that typically only includes unusual or strange components (Hawkins,
2005). Convergent observations could be made based on the use of inaccurate or simplified
representations of Native Americans in the tourism industry, news media, advertisements, and
sports (Magelssen, 2002; Merskin, 2001; Miller, 1999; Pewewardy, 2004; Sanchez, 2003). By
caricaturing cultural differences, these representations reduce the likelihood that Native
Americans can be construed as typical, modern Americans.
In sum, people may hold multiple and sometimes discordant views about the extent to
which the American identity is associated with these groups. Depending on the determinants of
the psychological ascription of national identity, Native Americans could be construed as more,
less, or equally American than White Americans. These alternative conceptions may reveal
discrepancies between more or less intentional thoughts. Over the past two decades, a
substantial body of research has established that thoughts or feelings about social groups can
operate automatically and deliberately. More precisely, a theoretical distinction is made between
responses based on controlled or deliberate processes and responses that are not available to
introspection or cannot be consciously controlled (Devos, in press; Fazio & Olson, 2003;
Greenwald & Banaji, 1995). An appealing feature of techniques developed to assess implicit
social cognition is the possibility of revealing associations that are unknown or unwanted to the
possessor (Greenwald & Banaji, 1995).
In the present research, we examined the extent to which an implicit measure reveals
knowledge that is distinct from parallel explicit assessments regarding national identity. Explicit
responses typically reflect processes of propositional reasoning concerned with the validation of
beliefs or attitudes (Gawronski & Bodenhausen, 2006). The explicit endorsement of a statement
is based on its subjective validity (or “truth value”) and on the extent to which it is consistent
with other statements held valid. Length of immersion of a group in America and egalitarian
principles are likely to come into play when individuals deliberatively assess the extent to which
Native Americans and White Americans are American. Based on the fact that American Indianswere the first inhabitants in Northern America, they should be granted the attribute American to
a greater extent than immigrants from European countries. If individuals subscribe to the idea
that every American citizen should enjoy the same legal and political rights, they should not
differentiate groups or individuals based on their ethnicity or cultural background and should
view Native Americans and White Americans as being equally American.
In contrast, performances on an implicit measure are typically driven by processes of
pattern activation that depend on a pre-existing structure of associations in memory and the
particular set of input stimuli encountered in a given context (Gawronski & Bodenhausen,
2006). Everyday images of prominent Americans – all the U.S. presidents, most political and
social leaders, celebrities, and a majority of the population – reinforce associations of America
as White. As a result, the American identity might be associated with the ethnic identity of being
White – so much so, that even deliberate, conscious rejection cannot alter this association. In
sum, we anticipated that the American identity would be more strongly associated with Native
Americans than with White Americans on measures tapping deliberative processes, whereas the
reverse pattern would emerge on a measure of associations that cannot be consciously controlled
(American = White effect).
A second goal of the present research was to examine the extent to which implicit and
explicit associations were sensitive to group or individual differences. More precisely, we
sought to document the fact that indirect measures, as well as self-reports, reveal meaningful
differences based on respondents’ ethnicity and political orientation. Competing hypotheses can
be stated for the impact of ethnicity on ethnic-American associations. Previous research showed
that the American = White effect was not restricted to White American respondents. In
particular, Asian American participants viewed their own group as being less American than
Whites on implicit and explicit measures (Cheryan & Monin, 2005; Devos & Banaji, 2005).
These findings are consistent with the notion that members of minority or disadvantaged groups
sometimes display beliefs or attitudes that do not serve the interest of their group (Jost & Banaji,
1994) and not claiming national identity for one’s own group would be evidence of such a state
of affairs. If the perception of Native Americans as marginal to modern American is internalized
by Native Americans themselves, and this is not inconceivable given the trauma that has
resulted from the massive loss of lives and removal of their dominance (Brave-Heart &
DeBruyn, 1998), they and White Americans should equally show weakened association of
national identity for Native Americans.
However, in contrast to other ethnic minorities, it could be argued that most Native
Americans have an identity that is profoundly connected to a sense of place (Marker, 2000). For
example, when questioned by an anthropologist on what the Indians called America before
European immigrants came, an American Indian said simply: “Ours” (Deloria, 1988). From this
perspective, one would predict a more symmetric pattern of ethnic-American association: White
American and Native American respondents would both construe the ingroup as being more
American than the outgroup. This hypothesis is consistent with the ingroup projection model in
that both groups would view ingroup members as being more prototypical of a superordinate
category than outgroup members (Mummendey & Wenzel, 1999; Waldzus, Mummendey,
Wenzel, & Weber, 2003; Wenzel, Mummendey, Weber, & Waldzus, 2003). The present dataset
offered us the opportunity, for the first time, to test these competing hypotheses for a
comparison between Native Americans and White Americans.
If the extent to which groups are assigned the American identity partially reflects groupbased
hierarchies, one would expect ethnic-American associations to covary with ideological or
political placement. Recent models building on classic and contemporary work on the
authoritarian personality (Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950; Altemeyer,
1998) posit that conservatives are less concerned with equality and more likely to display
intergroup biases than are liberals (Jost, 2006; Jost, Glaser, Kruglanski, & Sulloway, 2003).
More precisely, people who identify themselves as conservatives tend to show implicit and
explicit preferences for higher status groups over lower status groups, disadvantaged groups, or
groups seeking social changes or egalitarian reforms to a greater extent than people who identify
themselves as liberals (Conover & Feldman, 1981; Cunningham, Nezlek, & Banaji, 2004;
Nosek, Jost, & Banaji, 2007; Nosek, Smyth et al., 2007; Sidanius, Pratto, & Bobo, 1996;
Whitley, 1999). The American = White effect should not be equated with an evaluative bias
favoring White Americans over ethnic minorities, but it is inconsistent with egalitarian
principles and is expressing a greater inclusion of White Americans than other ethnic groups in
the national identity. If implicit and explicit distinctions reflect the relative status of White
Americans and Native Americans in the U.S., one would expect self-identified conservatives to
be less likely to ascribe the American identity to Native Americans than self-identified liberals.

Methods:
Stimuli
To represent the target groups, sixteen black-and-white pictures of Native Americans and
White Americans were used. For each group, four men and four women were selected. Pictures
were cropped at the torso. We explicitly choose to represent the concept American via
Are Native Americans American? 6
exemplars that were not indicative of White America. For example, we did not use the American
flag or monuments that are explicitly the creation or representation of European Americans.
Given the potential influence of stimuli properties on IAT effects (Bluemke & Friese,
2006; Devos & Banaji, 2005; Govan & Williams, 2004; Mitchell, Nosek, & Banaji, 2003), two
versions of this IAT were created. One version (Landmark-IAT) used pictures of natural scenes
from the American landscape that can be easily associated with Native Americans: Old Faithful
(Wyoming), Niagara Falls (New York), Grand Canyon (Arizona), Devil’s Tower (Wyoming),
and Redwood Forest (California). These natural landmarks were contrasted with recognizable
foreign natural landmarks: Amazon River (Peru), Antarctica, Mt. Everest (Nepal), Rainforest
(Costa Rica), and Sahara Desert (Africa).
The second version (Location-IAT) used five names of cities or states with names of
Native American origin: Ohio, Miami, Missouri, Seattle, and Utah. These stimuli were
contrasted to five names of European cities or countries: France, Italy, Moscow, Oslo, and
Warsaw. Both stimulus sets highlight the connection of Native Americans to America and
therefore creates tests that if anything, should provoke an easier association of America with
Native Americans. The test can be sampled at https://implicit.harvard.edu/. Stimuli used for
these tasks can be obtained at http://projectimplicit.net/stimuli.php.

Procedure
Visitors found the Web sites through media coverage, course assignments, blog or
chatroom discussions, personal recommendation, search engines, Web sites that provided a link,
or accidentally. They were presented with background information about implicit attitudes and
stereotypes and invited to participate and receive feedback about their performance. Visitors
who decided to participate could select from a list of tasks. Study sessions lasted approximately
10 minutes and consisted of administration of the IAT and a brief questionnaire. For most
sessions, the order of the IAT and questionnaire was randomized.
Implicit association test. In line with previous research (Devos & Banaji, 2005), the
Implicit Association Test (IAT, Greenwald, McGhee, & Schwartz, 1998) was adapted to
measure the direction and strength of the associations between two groups (Native Americans
and White Americans) and two attributes (American and foreign). The technique is based on the
assumption that the strength of associations between two pairs of concepts can be revealed by
the ease with which participants discriminate (or combine) stimuli representing these concepts
under different conditions. The technique does not require introspective access and minimizes
the role of conscious control or intention (Nosek, Greenwald, & Banaji, 2006). Respondents
were assigned to one of two versions of the task (Landmark-IAT vs. Location-IAT). Exemplars
representing each of the categories appear at the center of the computer screen and participants
categorize them into one of the four categories as quickly as possible using two computer keys.
More precisely, the IAT included seven blocks of trials with the following categorization
rules: (B1) 20 trials sorting the group exemplars (e.g., e key for Native Americans, i key for
White Americans); (B2) 20 trials sorting the attribute exemplars words (e.g., e key for American
stimuli, i key for foreign stimuli); (B3) 20 trials sorting all four exemplar types with one group
and one attribute sharing a response key and the other group and attribute sharing the other
response key (e.g., e key for Native Americans and American stimuli, i key for White
Americans and foreign stimuli); (B4) 40 trials using the same sorting rules as B3; (B5) 40 trials
Are Native Americans American? 7
sorting the concept exemplars as in B1, but with the key mappings reversed (e.g., e key for
White Americans, i key for Native Americans); (B6) 20 trials sorting all four exemplar types,
but reflecting the change in key mapping in B5 (e.g., e key for White Americans and American
stimuli, i key for Native Americans and foreign stimuli); (B7) 40 trials using the same sorting
rules as B6. Blocks B3, B4, B5, and B6 comprise the primary data for analysis. The seven
blocks were presented in the order described above, or with the sorting combinations of B1, B3,
and B4 exchanged with B5, B6, and B7.
Group labels were presented in green font and attribute concepts in white, all on a black
background, to emphasize that individuals were to be categorized by their ethnicity, not whether
they were American or foreign. If a participant made an error in sorting during any of the
response trials, a red “X” appeared just below the exemplar and remained there until they
corrected the error.
Self-report measures. A single-item measure was constructed to parallel the relative
nature of the IAT. Participants were asked to indicate which statement best describes their belief
among the following options: -4 = I strongly consider Americans of European descent to be
more American than American Indians, - 3 = I moderately consider Americans of European
descent to be more American than American Indians, -2 = I somewhat consider Americans of
European descent to be more American than American Indians, -1 = I slightly consider
Americans of European descent to be more American than American Indians, 0 = Both are
equally American, 1 = I slightly consider American Indians to be more American than
Americans of European descent, 2 = I somewhat consider American Indians to be more
American than Americans of European descent, 3 = I moderately consider American Indians to
be more American than Americans of European descent, 4 = I strongly consider American
Indians to be more American than Americans of European descent.2
In addition, participants completed separate assessments of the association between
ethnic groups and the concept American: “In your mind, how American are people who belong
to the following groups? That is, how strongly are they identified with America and all things
American?” Responses were provided for four groups (Native Americans, White Americans,
Asian Americans, and Black Americans) using 7-point scales labeled as follows: 1 = Not at all
American, 2 = Barely American, 3 = Slightly American, 4 = Moderately American, 5 = Fairly
American, 6 = Strongly American, 7 = Absolutely American.
Demographic survey. Participants were asked to provide demographic information such
as gender, age, ethnicity, political orientation, education, country of primary citizenship and
country of residence, and current postal code.
At the end of the session, participants were debriefed, provided feedback about their IAT
performance and additional background materials such as answers to frequently asked questions
about the IAT and this test in particular.

Lily Probst said...

if you want i can recopy all of this and post it as a comment to the blog post called IAT, i just had a lot of information and links and images, and figured it would be easier to format as a post.

RWinkel said...

Wow Lily, fantastic post. You asked questions based on material presented in class and you did excellent research which led to your answer. You should consider writing to Banaji with your recommendations. Perhaps you could write a better summary for them. That could be your project...
rw

Lily Probst said...

I actually sent an email to the IAT general email address after i took the test, i only got a response in the past week or so. This was the response:

"I believe that these photos, not being of living individuals, can be
used on the demo site. Perhaps at the time this IAT went online,
there were no suitable pictures of these categories. We dont use
faces of actual people in the demo tests on the site (they are usually
morphs) unless they are celebrities.

The key issue is that the exemplars should be associable with a
category.. These photos fit this role as exemplars. Similarly in the
race-weapons task, images of ancient weapons are used. The standard
IAT is not particularly sensitive to the choice of exemplars, unless
of course the exemplars redefine the putative category label.

We have used names and symbols for other categories.. so in principle
it should be possible to use something along these lines in this case."

I thought that it added to and confirmedwhat I learned from the research I did. I still feel that the test could be more dynamic and informative if they had a modern version, with morphs of current white and native american faces, hopefully they will consider it in the future.