We chose this topic for our IAT because it is a more subtle variation of the race IAT. As we discussed in class, there is evidence which shows that when athletes are described in sports articles, there is a tendency to associate black athletes with being more physically strong and white athletes with being more strategic. Because the IAT is a measure designed to indicate the existence of these subtleties, we figured it would be interesting to see if other members of our class were affected by this inclination as well.

After running a one-sample t-test on our IAT we found some interesting data. With a total of eight participants we did not have a significant IAT effect and had a fairly large standard deviation (p = .160, SD = 187.7). With this said it is important to note we did find a mean IAT effect of 104.1 ms. When looking at more individual scores we find a potential outlier of -191.6 ms which would have lowered the average and increased the standard deviation a fair amount resulting in a less significant result. Even though the results of the t-test were not significant, both of these findings show that there could potentially still be an effect which could be revealed with further testing. This effect was be in the direction we would have expected, where individuals spent more time pairing the incongruent condition (Black/Intellectual words and White/athletic words) than the congruent condition.

Our Explicit measure was the most consistent in measuring what we hoped to understand- whether people thought of black athletes as more athletic and white athletes as more intellectual. This test was very close to signifigance, and the small sample size leads us to wonder how much more significant this measure would be were there more people measured (t(7)=2.303, p=.055, N=8). Interestingly, the explicit measure and the IAT had the highest correlation (r=.479, p=.230) although the finding was not signifigant. This seems to say that our explicit measure and IAT (implicit) measure were more closely measuring what we had intended then the criterion measure, when generally one would NOT expect the explicit and the implicit to measure the same thing.

When analyzed with a one-sample t-test our criterion measure was not signifigant (t(7) = -1.426, p =.197, N = 8). Our criterion measure was designed to test whether when asked to describe two male athletes with the same qualifications, people would tend to describe a white male with more intellectual terms and a black male with more athletic terms. Although our findings were not significant, we would have to wonder whether or not this is because of a confound with the list of adjectives chosen. When typed on the computer the athletic words were placed in a different color to keep them separate. When printed off of the original printer both columns were printed in black. However, when participating in the experiment most of the print-outs had distinct color differences with the words. Although not all of the subjects were affected, it was clear that most may have been influenced by the color differences picking up that they were from two different categories. This methodological problem may have resulted in non-significant values for both the one-sample t-test of the criterion data and the correlations between the criterion and the IAT and the criterion and the explicit measure. This unfortunate methodological issue may be the reason neither the IAT nor the explicit measure was able to predict the criterion or behavior measure. So, implicit and explicit attitudes were more simmilar than implicit attitudes and behavior or explicit attitudes and behavior.

Perhaps this IAT effect was strong because the words chosen were very easily categorized into one of two categories, and the pictures were not ambiguous at all, and the IAT is actualy measuring what we intended. Perhaps people do have implicit attitudes that Black athletes are more easily associated with strong and brawny and white athletes with strategic and bright. What does everyone else think?

-Dave, Andy, Meghan, Lucia

Sexuality IAT

Our experiment dealt with individuals' attitudes toward sexuality (heterosexuality and homosexuality). We hypothesized that people would have a bias toward heterosexuality. The study had three different variables: an explicit measure, implicit measure, and a criterion variable. The implicit measure was in the form of an IAT, in which subjects were asked to associate heterosexual images with pleasant words such as "sunny" and homosexual images with unpleasant words such as "vomit" (congruous condition). Then, subjects were asked to associate homosexual images with pleasant words and heterosexual images with unpleasant words (incongruous condition). We hypothesized that people would be more readily able to and faster at associating the conguous conditions, which would show an implicit bias toward heterosexuality.

The explicit measure was in the form of a survey, in which people were explicitly asked about their views of homosexuality as compared to heterosexuality. Their responses were on a scale of 1 - 7 (ranged from very uncomfortable - very comfortable).

The criterion measure was comprised of 2 short stories which dealt with homosexual situations. Subjects were asked to measure their heart rate before and after reading the stories. A drastic increase in heart rate would signify a high level of discomfort with issues pretaining to homosexuality.

Our IAT effect was -8.7688, with a standard deviation of 71.002, p = .737 (no significance).

There was a high, positive correlation (.583) between the IAT and the criterion...as one variable increased, so did the other. This shows that the two variables are good at predicting eachother. If someone has a bias toward homosexuals in the IAT, they're probably going to show a bias in the criterion measure, as well. The significance of the relationship between the IAT and the criterion measure was .130, which approaches significance. This was the most significant relationship between any of our variables.

What was interesting was that there was a negative correlation between the IAT and the explicit measure (-.188, p = .655), which shows that as one variable increased, the other decreased. It is posisble that this could be attributed to the fact that homophobia is considered to be taboo, and perhaps people explicitly tried to appear less homophobic than they really are. The IAT may have showed higher levels of a bias toward heterosexuality than the explicit measure. This is shown in the Poehlman and Uhlmann experiment, in that is it possible that "social deiriablilty concerns are the source of low correspondance between IAT and self-report measures". Another reason could be that "implicit and explicit measures tap into distinct attitudes which sometimes correspond and sometimes do not".
In the Poehlman and Uhlmann experiment, the IAT measures outperform explicit measures in predicting steotyping and prejudice, which corresponds with our findings.

Limitations of our study could have been related to the small sample size of our study. Also our critereon stories may not have been graphic enough and were too short, which made it difficult to monitor the responses of heart rate. It is possible that not having legitimate equipment to measure heart rate could have affected the results of the criterion measure. Also, some subjects noted that the pictures in the IAT were ambiguous with regards to gender and that perhaps even the level of attractiveness of the people in the pictures (both heterosexual and homosexual images) affected their responses.

- Amanda, Sarah, Allen, and Suhani

Our IAT examined the difference between comfort levels of students with male versus female faculty. We hypothesized that students would feel more comfortable with female faculty. We began by asking participants to approach three faculty members that they were not close to and address them by their first names, and then rate their level of comfort on a scale of 0-100 with 0 being very uncomfortable and 100 being very comfortable. These scores became our criterion measure. Our explicit measure was a survey asking participants their attitudes about male and female faculty.
Our IAT data revealed that there was a difference in the comfort levels of participants with male versus female faculty (p=.116, M=67.46). This was an impressive finding considering the fact that there were only seven participants, one being an extreme outlier (-102). Our correlation between the IAT and the criterion measure was .519 and our p-value was approaching significance (p=.233). The correlation between the explicit and criterion measures was .08, but our p-value was very high (p=.865). These results indicate that the IAT was better at predicting behavior than the explicit measure.
The outlier we had was extreme in both the IAT and the criterion (but not the explicit), which may have changed the outcome of our statistics, but did remain consistent with our IAT better predicting behavior.
According to the Poehlman and Uhlmann article, the IAT was better at predicting behavior regarding culturally sensitive issues than explicit measures. Gender differences, especially in a professional setting, are included in this group of issues. Our findings support this idea.
We did feel that there were some methodological problems during this experiment. One participant did not complete the criterion measure, two participants only chose one gender, and one participant only chose two faculty members. This made the criterion difficult to score and may have affected our results.
Katy Bissell
Laura Serafin
Jessie Harpel
Ritwik Niyogi

If any of you are reading this sometime now: WHERE ARE YOU?

This paper introduces an interesting twist to our discussions of the IAT and the adaptive unconscious. I was personally intrigued by three points in particular - the effectiveness of the IAT in predicting criterion measures in situations where social desirability concerns are involved ( compared to other explicit measures), the fact that “high implicit-explicit correspondence (i.e high correlations between IAT and explicit measures) was associated with both higher ICCs and ECCs”, and the idea that implicit attitudes/ motivations may play a larger role in determining behavior which involves deliberative responses than previously believed.
This research backs up Wilson’s claims in more ways than one. It supports his idea that the unconscious plays a very significant role in influencing behavior. It also corroborates his view that there may be discrepancies between our unconscious motivations and consciously stated ideas/values ( e.g. in areas concerned with race, stereotypes, prejudices.) However this paper does not really back up his claim about the reason for this discrepancy. The analyses could not sufficiently reveal whether the discrepancy was because of “naïve self-deception” (which is in line with Wilson’s ideas) or “strategic deception of others” (which implies that the subjects were aware that they were prejudiced, and chose to conceal their true opinions because of social desirability concerns). The fact that the IAT can override the effect of social desirability concerns, however, makes it a particularly useful tool in the ascertainment of true attitudes/opinions. On the other hand, explicit measures would be more useful in predicting behavior in situations where “social desirability concerns [are] weak”. (such as choosing between insects and flowers)

The paper provides a possible explanation for the “high implicit-explicit correspondence = higher predictive validity” results – “When implicit-explicit correspondence is low, the associations measured by IAT measures may correlate more weakly with criterion measures because the person is attempting to intentionally override an unwanted automatic response. Additionally, self-reported views may relate less strongly to criterion measures when automatic processes are pulling the person in the opposite direction.” Wilson did not dwell too much on the interaction of the conscious and unconscious although he did mention that people whose unconscious motivations were in harmony with their conscious motivations are likely to be happier (or something to that effect, I can’t find that paragraph!) The idea of this tug-of-war b/w the conscious and unconscious seems appealing and fits well with his other ideas. It makes allowances for the possibility that interactions b/w the two entities can be bilateral and divergent, instead of just plainly unilateral (determined by the unconscious).


Suhani

Katy Bissell

Something that was particularly interesting to me about this paper was the authors’ claim that the IAT can predict explicit as well as implicit responses. The authors state that “IAT measures were similarly effective at predicting responses that are difficult to control and those that are easy to control” (25). I found this interesting because it almost seems as if the authors are working to support the idea of the adaptive unconscious controlling conscious actions and thoughts. The authors even mention “the results of this meta-analysis suggest automatic associations may contribute similarly to both automatic and deliberative responses” (Rudman, 2004).
This is a very important discovery with the IAT, because if people consciously act a certain way toward someone of a different race or background etc., but are not aware of why they act this way, the IAT would be able to clue them in. Going along with that, the authors found that “whereas explicit measures were impaired in their predictive validity when social desirability concerns were high, IAT measures were relatively unaffected” (23). This seems to prove that even when people know they shouldn’t harbor certain feelings and/or attitudes because they are socially unacceptable, it can be revealed that they may be harboring such attitudes unconsciously, which means they would not be aware of these feelings, even if they acted on them.
If the IAT can connect implicit and explicit measures like this paper shows, does it mean that the adaptive unconscious does, in fact, exist AND control our conscious actions without our being naturally aware of it, unless we all take the IAT? This is an essential test if that is the case.

Laura Serafin

One of the things about this study I found particularly interesting is the idea of implicit-explicit correspondence lead to greater predictive validity, “Discrepancies between automatic and deliberative reactions reduce the extent to which both responses are reflected in behavior” (pg. 28) This to me suggests that the best way to predict ANY kind of behavior is when the conscious and the unconscious are in agreement. There isn’t any kind of disagreement between unconscious and conscious, so there is no fight for which to follow in behavior. One thing this finding suggests to me is a connection between conscious and unconscious. The Gladwell idea of the “locked door” is clearly not applicable here – information is getting back and forth to some degree.

However, what was even more interesting is that a lack of implicit-explicit correspondence is more influential in validity of explicit measures. What this says to me is that the unconscious is more influential on the conscious than vice versa. The unconscious is able to influence seemingly controllable behavior, which is a pretty impressive finding.

- by Andy

A lot of the other posts have addressed many of my same thoughts about the article. One thing that did stick with me, however, was the paper acknowledging one possible area of research in the future. The authors point out that a lot of the IATs were conducted with college students, and that the results may be different if a more "normal" population was used.

I would find it interesting if this was tested, and how much of a difference there would be. Although many colleges are still liberally-minded, I've encountered some people on our "liberally biased" campus who have had some surprisingly prejudiced views. Amanda brought up the influence of culture; if a college student made stereotypical associations, does this mean that this is a result of the individual, the hometown culture of the individual, or perhaps something with the campus itself? I find this question to be pretty interesting.

The other point which I found intriguing in the statement on p. 27 "(Although of course longitudinal research is needed to determine whether automatic associations determine explicit attitudes or the reverse.)" I'm curious about this myself. This goes back to a question we've asked before: is it our conscious which influences the unconscious or vice versa? It'd be interesting to see the results.

-Megan Rosenbluth-
I would have to agree with Sarah that most of the statements that were made by the researcher were slightly obvious. Even from the beginning in the abstract, the conclusion that was drawn was a conclusion that most people I feel would assume was true, "When IAT and explicit measures were strongly correlated, both predicted criterion measures more effectively than when implicit-explicit coorespondence was low" (2). I felt that this was a rather obvious statement, and felt that most people would assume that if you showed a behavior it would be revealed through an IAT more readily than an implicit feeling.

I still have trouble accepting the IAT as a valid measure of behavior, however, I did like how the researcher stated that the correlation was higher as an explicit behavior and the IAT than an implicit-explicit relationship. I felt that in most cases people are more willing to say that the IAT should predict behavior, rather in their research I felt as though they were saying that the IAT would confirm that the behavior being demonstrated was that of an implicit feeling.

A statement that I particurarly agreed with was, "A difference in overall speed between the two blocks is taken to indicate the direction and magnitude of association strengths among the categories and attributes" (4), becauseI personally feel that it is difficult for certain stimuli to be categorized into a specific category and rather might fall into a grey area. I felt that this quote showed that in fact a slower response time would show that maybe the participant felt that it might not necessarily belong into the group, making it less of a correlation. I would wonder, if all associations were made slowly what would their results look like? What would this say about their implicit feelings towards the specific stimuli?

-Lucia Stancioff-

In the discussion, the Benaji article purposes that "automatic associations may contribute similarly to both automatic and deliberative responses" and " the present findings may suggest a meaningful revision to dual process models" (25). The implications (if the IAT indeed is measuring unconscious associations or hidden associations) of this change to the dual process model speak directly to what we have been blogging about in the past, namely the role of the unconscious on conscious thought. I find this section of the meta-analysis the most intriguing and think it should be examined more deeply because it is implying that something we can't control or even notice on our own like the time difference between associating "fat" and "good" or "fat" and "bad" is influencing decisions which we feel and think we are controlling! Going back to the Blink readings, this is further evidence that all of the confidence we have that we are in control of our conscious behavior is higher than it should be. Amanda posted "I think that societal influences affect the IAT. This is evident in the fact that the results that most favored the validity of the IAT were for prejudice and stereotyping, which are both influenced by society." The changes proposed to be made to the dual process model reflect these societal influences on our unconscious because those are the first influences to our implicit attitudes, then those implicit attitudes influence our external decisions and predict (to some extent) even our controllable behavior! I like that this meta-analysis is reaching outside the box of "implicit measure predict uncontrollable behavior and explicit measures predict controllable behavior" to say that implicit measure can predict controllable behavior, but I also wonder if it works the other way. Do explicit measures predict uncontrollable behavior? Benaji et. al. state that "the more controllable the response in question, the better explicit measures were able to predict it" (17) but does not (as far as I can tell) speak to whether explicit measures have any predictive validity over less controllable behavior. I wonder how a study would fare testing this relationship and if the results would really turn the dual process model around!

When reading this paper, I feel that all it did was really enforce my belief that the IAT results are different based completely on the type of IAT you take. There are societal pressures which influnce your IAT results, which is something we have discussed in depth. I feel that the information presented here is at times obvious. For example, on page 24 Banji and colleagues state that "although self-report measures were excellent predictors of controllable acts like voting, they were found to be weaker predictors of spontaneous acts such as eye contact with an interaction partner and anxiety during a stressful speech." I feel that this is quite an obvious statement. Spontaneous acts are, in my opinion, much more likely to be harder to qualify in a self report statement.

The point that was most interesting to me, yet only briefly touched upon, was the fact that "previous work suggests that IAT effects get smaller as the number of IATs taken increases(Greenwald, Nosek & Banji, 2003; Greenwald & Nosek, 2001). I think that this is an issue that actually would go against the belief that the IAT is an effective way of measuring one's implicit attitudes. I understand that with multiple IATs you would become better because of practice, but what I don't understand is if this is your implicit attitude, which is located within your unconscious, why should your results change with the number of IATs taken? It doesn't makes sense to me that your "implicit" attitude could be so easily changed by repetition of IAT participation. Any thoughts on this?

Sarah M.

When reading this paper, I feel that all it did was really enforce my belief that the IAT results are different based completely on the type of IAT you take. There are societal pressures which influnce your IAT results, which is something we have discussed in depth. I feel that the information presented here is at times obvious. For example, on page 24 Banji and colleagues state that "although self-report measures were excellent predictors of controllable acts like voting, they were found to be weaker predictors of spontaneous acts such as eye contact with an interaction partner and anxiety during a stressful speech." I feel that this is quite an obvious statement. Spontaneous acts are, in my opinion, much more likely to be harder to qualify in a self report statement.

The point that was most interesting to me, yet only briefly touched upon, was the fact that "previous work suggests that IAT effects get smaller as the number of IATs taken increases(Greenwald, Nosek & Banji, 2003; Greenwald & Nosek, 2001). I think that this is an issue that actually would go against the belief that the IAT is an effective way of measuring one's implicit attitudes. I understand that with multiple IATs you would become better because of practice, but what I don't understand is if this is your implicit attitude, which is located within your unconscious, why should your results change with the number of IATs taken? It doesn't makes sense to me that your "implicit" attitude could be so easily changed by repetition of IAT participation. Any thoughts on this?

Sarah M.

Amanda Eickman

Banaji et al (date was not listed) presented statistical proof that the IAT measures are vetter at predicting stereotyping and prejudice than the explicit measures (self-report). I feel as though this is an obvious statement and something that doesn't need to be proven with statistics, but rather, logic. The results with regard to stereotyping and prejudice could simply be due to the culture we live in. Coming from New England and New York, prejudice is not accepted culturally. Stereotyping and prejudice is considered to be socially unacceptable, which could affect the explicit measures of this experiment. The average person, despite what prejudices he/she may have, is generally reluctant/unwilling to express racist (etc.) views. Perhaps if this experiment was conducted in a society where stereotyping and prejudice was commonplace and not considered to be a negative thing, the results would show that the implicit and explicit measures were the same (or at least closer). I think that societal influences affect the IAT. This is evident in the fact that the results that most favored the validity of the IAT were for prejudice and stereotyping, which are both influenced by society.

Similarly, Banaji et al (_date?_) also found that "self-report measures outperformed IAT measures in predicting brand-related choices...and political preferences". This is also probably due to societal influence. Society tells us which brands are "cool" and that expensive = desireable. Society affects these choices, as well. With regards to brand preferences, perhaps because it is a relatively unimportant aspect of cognition, our unconscious does not even attend to these preferences as much.

Irregardless, society affects our attitudes a great deal. Becuase of this powerful influence, I feel as though the IAT results are slightly less valid. If society can influence our attitudes/behaviors, then IAT results could vary from day to day under the right circumstances.

Here begins the second round of blogging for the blogging week that ends Wednesday, 4/17.

I thought that the clinical-research IAT was very intriguing, although I'm not sure if the results definitely reveal what they claim to reveal. I, for one, found myself placing both clinical and research related terms on the "women" side. I thought that the terms that were related to clinical psychology were fairly general and I had a hard time associating them with clinical psychology specifically. I did think that it was interesting that I placed positive words such as "support" on the "women" pile because that, to me, points to a different kind of bias - that women are more "supportive" or "caring" than men. I didn't have the same problem with the research related terms (I tended to place them slightly more often on the "women" pile. )

It think that these results point to the fact that society assigns a set of definite roles to men and women. Society sends us a number of direct and indirect messages that so deeply ingrain in us the importance of performing certain activites that correspond to these roles, to the point where we stop asking basic questions. I think the IAT taps into these phenomena, which in part explains why the results are often so shocking.

In my English class, we were discussing how the English literary canon largely consists of "male" works by male authors. As a result, both men and women alike are taught to think and read like men because of continued exposure to the "classics" that, in turn, shape their views on literature, and the world in general.

Elizabeth Showalter's critique of the literary curriculum illustrates this idea:
" The first result of my reading was a feeling that male characters were at the least more interesting than women to the authors who invented them. Thus if, reading their books as it seemed their authors intended them, I naively identified with a character, I repeatedly chose men; I would rather have been Hamlet than Ophelia, Tom Jones instead of Sophia Western, and, perhaps, despite Dostoevsky's intention, Raskolnikov not Sonia."
She goes on to write " I quickly learned power was unfeminine and powerful women were, quite literally, monstrous..Bitches all, they must be eliminated, reformed or at the very least condemned..."
Unfortunately, all readers are not as perceptive as Showalter. We often take the "classics" for granted and almost "swallow" them whole, often without as much as a critical thought. Showalter restricts herself to classic works, but I am tempted to include the staple diet of fairy tales and childhood stories that we all grow up on( A case in point would be "Beauty and the Beast")

These are all examples of subtle yet intensely powerful messages that society conveys to us in a variety of ways. We are mostly unaware of these influences, yet the fact is that they undeniably exist and often influence our behaviour throughout the course of our lives. In that light, I think that the IAT does a good job of revealing the (often negative) effects of such societal conditioning.

Suhani

When it comes to the IAT test, I feel that the most interesting topic we discussed on Wednesday was in regards to the IAT results and one's outward behaviors. When I think about this topic without analyzing it, I immediately think that one's IAT results would be outwardly reflected in their behavior. I think that is my initial reaction, because it makes the most sense. I feel that even if you have implicit biases, you must be exhibiting them in some way, be it small or significant. When I think more critically about this question, and consider examples such as my group's question where Black civil rights activists show IAT effects of bias against blacks, I think that the issue needs to be more carefully looked at.

In my opinion, the largest influence on IAT test results is societal pressures. In the example of civil rights activists showing bias against blacks, we are bred in a "caucasian based" society. There is so much influence from the media, and entertainment industry categorizing black and white people into certain stereotypes that eventually everyone is influenced by this way of thinking. You certainly would not see a black civil rights activist acting in a negative way towards another black person, so that seems to be the only explanation that makes sense to me. Race is only one of the IATs, and I feel that all of the tests should be considered differently. For example, an IAT looking for a bias against handicapped people is different to me than one looking for racism. Like we talked about in class, people may think that it is a negative thing to be handicapped, but not think about the handicapped person in a negative manner. For this reason, I feel that society does influence all behavior, especially the IAT results.

Sarah M.

In terms of the experiments I felt they were very good and showed solid results despite the small sample that we had. Especially with the handicap IAT it was good to see such prominent results. The only complaints I had about the test was that they were not on the computer. It was difficult to separate the papers and I feel my scores would have been significantly faster on the computer. However, this did affect everyone and it also affected both of my scores. I don’t know how I feel about the VW and Ford IAT, because I feel like this assumes that everyone has a bias for the VW Company since they were paired with the pleasant words initially, I might not be analyzing this properly, but I feel this isn’t a good group of things to test on because they don’t have any culturally affected stigma to them. The biggest point of discussion is that the whole bias towards handicapped people since this was our only data that was significant. I thought Lucia brought up an interesting point that maybe our bias is towards the fact that the people are handicapped and not the people themselves. I would like to believe that this is true, but I don't know if I can. I feel that the disability is a apart of them, something that defines them, makes them who they are. That if you have a bias towards the disability then you have a bias towards them because they are a representation of the bias and it is also intrinsically a part of them. So, to feel a bias towards the handicap, I feel, is a bias towards the person.
-Allen

-Megan Rosenbluth-
Something that Lucia said during class really stuck with me. She mentioned that people may not have an unpleasant feeling towards someone that is handicapped but rather that being handicapped is unpleasant. I found this to be a very important distinction that I do not feel is thoroughly taken into consideration on tests such as an IAT. Although I myself have always been skeptical that they were good predictors of behavior, after hearing Lucia's comment I am even more convinced that in fact they do not show a bias.

I have always personally felt that it is not a bias that is revealed but rather a preference. Although the two words are similar, I feel that in an instance like the IAT it is a very valid, although small, difference. If someone would prefer to not be handicapped, that is much better than being biased towards the handicapped.

I also believe that if someone over emphasizes a reaction towards someone, even in a good way, that is just as much a bias as someone who would behave negatively. I feel that someone who would treat someone with a handicapped badly because of their disability, is just as bad as someone who goes out of their way to do something nice just for the sake of being nice. In neither situation is that person being genuine, and therefor I strongly believe that in both situations there is a bias present. I feel that the word bias has such a negative connotation, that people tend to forget that being insincere is the internal version of behaving badly.

The IAT results show the role of familiarity in unconscious processing. The fact that we had a greater latency for the incongruent condition of disabled and pleasant shows that we have stronger associtions with the more familiar normal person. We can extend this by saying that we are more inclined towards the familiar than the unfamiliar in our behavior. While making snap/unconscious decisions, we would be going towards the more familiar method or path. The latency in the incongruent condition seems to show that we prefer and are more comfortable with what we know than the unknown...and tend to classify the unknown as unpleasant more often.

However, there are 2 things which have made me think even more about the IAT's implications. The first concerns the effect of practice categorizing something into a category (social influence) over the years against practical, more immediate practice. I had a +12 on the female-clinical male-research IAT. That is I my difference between the latencies of the congruent conditon of female-clinical, male-research and the opposite was 12 seconds...much more than anybody else in the class. This, I could explain is because of my association for research being a domain of scientific exploration, and over the years I have associated males and scientists (because of the overwhelming number of male scientists over females). Clinical seems like it would need more feelings and once again because of social influnce, I can associate it more with females. However, the +12 is highly contradicting the fact that the only research group I have been a part of has more females than males. This would mean that my regular practice is not able to overcome my established association (and based on what I suggested in class, a +12 IAT effect suggests a very strong association and could be interpreted as a bias). This suggests that our theoretical erroneous associations and biases are, due to a huge amount of practice,very difficult to overturn even with personal and practical experience. I had a strong association with Caucasians and good even though I have colored roommates (one of them is an African American) and an African as one of my closest friends.

The 2nd is the exact opposite. If one has practice with a clinical psychologist being a male (his/her clinical psychologist was a male) then he/she registers the reverse of the effect I showed. Do associations then have to do more with personal practice? My results can then be interpreted as that I want to take up research as a career and being a male associate it more with males.
Any thoughts on this ?

Ritwik

I was thinking about how the times for the IAT would be if we took into account that we have preferences for things on the right side. Would it be easier if for the congruent groups to be on the right side as opposed to the left side? Would this make enough of a difference, or would practice with putting the items into both categories overcome what small difference this might have to begin with.

To continue with the debate over cultural associations vs. biases on the IAT, I think it would be important to look at what parts of the brain become active when doing the IAT. If we were to see what areas of the brain light up as people are choosing we could compare those parts with what people normally think about using those parts. If it is more of a memory part used in thinking about history it might be linked more too cultural references. Hopefully a different area would be around personal experiences. There could be a different area people use when referencing their opinions and that would be able to light up if it was more based on biases. I’m not too familiar with the different areas of the brain so I’m not completely sure if this would work at all. Does anyone have an opinion on this one?

-Dave

I think that, for the time being, I stated all of what I wanted to regarding the IAT in class, or someone else had expressed the same thoughts. So I guess I'll just address three things which occurred to me while we were doing the experiment.

The first is more of a methodological concern. I'm not sure if this happened to anyone else, but I know that despite the practice rounds of sorting, I still made mistakes during the "abled-positive/disabled-negative" condition. If I recall correctly, I think I got confused by the positive/negative switch. I think I'm so used to doing the positive/negative associations on the right and left respectively that while I was sorting, I realized I screwed up and that made me take longer.

I apologize if the above doesn't make too much sense; I remember I definitely had mistakes going on during one of the conditions, but can't remember which one it was.

Regardless, the second observation I wanted to make is much more relevant. While we were doing the conditions with the pictures of the Psych professors, I found myself making another error when we were doing the "female-research/male-clinical" condition. Not only did I notice that I was putting some of the clinical words with the female pictures and the research words with the male pictures (despite how we were supposed to do the opposite), at one point a thought occurred to me along the lines of "Wait, this isn't right." when I think I was doing it right.

This really surprised me, because not only was I aware of an association, but I was momentarily conscious of it before going back to the task. Amazingly enough, the latency period still wasn't that as big and definitely not as long as the able/disabled task I mentioned above (which was 9 seconds for me). Regardless, I found it really interesting that I had that momentary thought right in the middle of the condition.

The final observation is fairly amusing to me. I really don't care too much about cars (with some noteable exceptions), so I didn't think I had a bias either way towards Ford or VW (though perhaps a slight inclination more towards the latter). Interestingly enough, my latency time was 0 seconds. Needless to say, I found it kind of funny that this may confirm my neutrality on the issue.

-- Andy

Amanda Eickman

I felt as though a few of the pictures used in the IAT were not ideal, and could have possibly affected our results: for the able-disabled pictures, there was a picture of someone sitting in an armchair, and every time, I started to place it in the disabled pile, and then had to check myself and place it in the able pile (I'm not sure if maybe this was intentionally supposed to make us think?). Also, a couple of the VW-Ford pictures were a little difficult to see which kind of car it was...What was interesting, is that I know what VW cars look like, but because I had to do the task so quickly, I found myself thinking about it more than I would have in normal circumstances. Usually, I would easily be able to determine that the car was a Passat, Jetta, etc., but I found myself examining the picture for a moment to determine if I could see the tiny VW insignia or not...even though I knew that I knew what kind it was. I know this was not feasible for this particular experiment, but I thinkk that if the pictures had been in color, it would have been a lot easier to determine the difference between conditions.

I'm still not entirely convinced that the IAT is as powerfuly of an indicator of implicit attitudes as people claim it is, simply based upon my own results. Looking at our results from class, I know that statistically it was effective, but because of my own results, I'm still skeptical.

Amanda Eickman

Our hypothesis was that we would be better at associating hte congruent conditions than the incongruent conditions (thus, we would be faster at categorizing the congruent conditions). Yet we only got strong results for one of the congruent conditions. Furthermore, my own results were not consistant with other people's results.

I've been trying to figure out how to account for my skewed results, and I'm still leaning towards it being an issue of practice. My first time (for the congruent condition of the 1st task) was 40 seconds, and my final time (for the incongruent condition of the 4th task) was 26 seconds. I improved more and more with each task. So for me, I guess perhaps it was just that I kept getting faster with practice. I did encounter trouble at the begining, however, because I was having difficulty separating the sheets from eachother (especially the big pieces of paper with the pictures on them) and felt they were rather hard to turn over at times. Then I started wrinkling the pages a little and it made it a bit easier to turn them over, so this could also partially account for my improvement.

What was interesing with my individual results was that my IAT scores (in order) were: -12, -2, 6, -12. The socres that most closely corresponded with our hypothesis were in the middle, surrounded by two definitive outliers. I'm not sure what exactly this would indicate (if anything), but I found it interesting.