Thursday, November 30, 2006

Presentation: Alex and Amanda

It was on deception in dating profile pictures. Please leave us your feedback on our presentation.

As Far as I Could Throw Him: The Factors that Contribute to Suspicion -Nikki

Please leave comments or suggestions! Thanks for everything this semester. I had a great time with all of you...trust me ;).

-Nikki

Lies in a Conversation - Barrett & Josh

Where do lies occur in a conversation?

Comments, suggestions, and criticisms are all welcome.

Thanks!
Barrett & Josh

Wednesday, November 29, 2006

Digital Deception



Nice to Meet You?
How do you behave in initial interactions?
Comments and questions about my presentation are welcome!
~Erica

Lying to Your Facebook



How honest is your Facebook profile?
Any suggestions for the paper are welcome. Thanks.

- Corey, Robin, & Jenna

Tuesday, November 28, 2006

Presentation - Lauren, Cameron, Brad

Beliefs about Online Deception. Please tell us what you thought. You may download the presentation at http://drbeagle.com/445/baod.pdf

Wednesday, November 22, 2006

I know Something You Don’t Know

Presentation-wise, I thought you did a good job. You didn’t rush through your points, and you were very clear and logical in your progression. I liked your use of an audience member for analysis, since it made the presentation very engaging. I thought there was a lot of text in the presentation, so it could’ve benefited from some alternate visuals like more pictures, illustrations or graphics; especially since you dealt with an idea that could have been displayed easily through visual examples. I thought you could’ve used an overview of your points at times (or how they relate to deception/Nyberg/etc…) and that might’ve been reading from your slides a bit too, but otherwise, I thoroughly enjoyed the presentation.

You deal with a very interesting idea that I felt engaged in immediately. Your content was clear, and I liked how you explained the premise of asymmetric information, your own hypotheses, and then the strategies used in practice and the results for your hypotheses. It would’ve been exciting (though totally unnecessary) to see how people get around the problem of asymmetric information instead of just how they use it. It also would’ve been interesting to see what people think about in terms of “what’s important for deception” (as discussed in class, ie. liking a person). In terms of a confederate,I think this is a great idea. It should be a few confederates (consisting of various characteristics to create more genrealizable results), but controlling for the information known is a good idea because it eliminates different reactions to profile length, profile details (in terms of type of details like personal info categories and basic info categories and specificity of info, like “I like dogs” and “I enjoy household animals”), and controls for reactions by confederates (of course).

Deception in Photos

You did a great job with your presentation. In terms of the non-content aspect, I really liked that you had a brief overview of your points and explained how they fit together logically to start the presentation. I liked that you used varying media in the presentation, as well as numerous slides to demonstrate each point. You also did not read off your slides, using them only for the audience’s reference, and this helped make your speech engaging. Furthermore, I felt that you did an excellent job speaking clearly, and (changing gears here) connecting your points logically and in an easy way to follow. Lastly, I loved your use of the graphic Jeff pointed out in class. Using a graphic as a means to explain a theoretical point was a great idea that my group will definitely use as well, thanks in part to your group’s pioneering ways!

In terms of content, I thought your presentation was also good. You did a fine job of explaining your main idea. You then did a great job of expanding upon this idea by relating it to Nyberg in interesting ways. You examples were concise and well-illustrated. I thought someone brought up a great point in class that you could have discussed the truth bias and expectations of deception with relation to your project, though this is an extra step for sure. Also, I liked how you critically analyzed the theories you used and your own conclusions. You did a rather effective job at this I might add, though I think you could’ve shown the inherent problems in comparing text-based theory to graphics.

Tuesday, November 21, 2006

Comments for our presentation

Hi guys,


Just in case you forgot, our presentation was about applying Nyberg's framework to pictures. We discussed how the different ways of showing and hiding can be applied to pictures.


Thanks for your comments!

Nicole and Kaitlin

Comments for Kaitlin and Nicole

All good things to say about you two.

My Presentation

Here's a little post so that you all can comment about my presentation.

A few things I'm thinking about and would love feedback on:
-Should I use a confederate as person B in the experiment? I have many hesitations, but I know there would be benefits too.
-Should I use some kind of task for their conversation? Would an introduction/biographical task work?
Or is there anything else important I didn't cover?

Thanks!

Wednesday, November 15, 2006

Alex and Amanda's Paper Foundation

Here is a concise summary of our work.

Research Question
1) What individual factors predict lies in the main profile picture?
a. Computer Experience
b. Self-perceived physical attractiveness
c. Gender
d. Age
e. Anticipation of Future Interaction
2) From here we will be considering how each of these five factors effects:
a. Frequency of Deception (ie. Do they lie or not?)
b. Level of Deception (We’re deciding now in our methodologies what scale to use. Originally we were considering if the level of deception was subtle, exaggerated, or outright.)

Hypotheses
H1: Individuals that have more computer experience will be more likely to deceive in their photo than individuals with less computer experience.
Sources:
-Ellison
-Dutton
-Mahar, Henderson, and Keane
-Wade

H2: Individuals with lower self-perceived attractiveness will be more likely to deceive in their photo than individuals with higher self-perceived attractiveness.
Sources:
-Higgins
-Elliott
-Doherty and Schlenker

H3: Women will be more likely to deceive in photo than men.
Sources:
-Polce-Lynch
-McAuley, Bane, Mihalko
-Martin, Leary, Rejeski
-Adamson, Doud Cralli

H4: Older people will be more likely than younger people to deceive in their photo.
Sources:
-Martin, Leary, Rejeski
-Coupland
-Sousa Campos

H5: Individuals that anticipate a future FtF interaction will be less likely to deceive in their online profile picture.
Sources:
-Walther
-Kellerman

Our biggest dilemma in the first assignment was that we took what Jeff calls, a lot of “garden paths.” We spent last week truly going through all of the literature and honing our question and finding theory to point us in the direction of these hypotheses.
This week we will be working on our methodologies for our proposed study.

Tuesday, November 14, 2006

Two Women in a Lab

Assignment 9 Option 1

I chose to assess the grainy black and white photo of two women in a laboratory for ten minutes. I wrote 409 words mostly about the testing of a new drug intended to help women achieve more pleasurable sex.
My LIWQ scores are as follows:

LIWC dimension
Need for Achievement 7.58 (mine) 5.8 (male average) 5.6 (female average)
Need for Affiliation 0.73- 1.1- 1.3
Need for power 2.93- 1.7- 1.8
Self-references (I, me, my)0.00- 0.5- 0.8
Social words 13.45- 11.4- 12.0
Positive emotions 1.22- 1.8- 2.1
Negative emotions 1.22- 1.5- 1.6
Big words (> 6 letters) 26.16- 18.7- 17.7

Need for Achievement. My results show that I wrote significantly more about achievement-related themes than "the typical person."

Need for Affiliation. Apparently, I don't pay much attention to human relationships. I mentioned in the story that the two women were student and teacher, but didn't provide any insights to their intimate encounters. I also mentioned the student having a sexual encounter with her boyfriend, but I guess this simulation categorized that as more physical than affiliation.

Need for Power. I am a power-hungry beast. Even in a ten minute creative story, I needed someone to clearly be in control. I scored well about the average in this output.

Self-references: While people who use a high rate of self-references tend to be more insecure, nervous, and possibly depressed, and honest, I think it's unlikely to have self-references in a made-up story about two women in a photo. I also think it would be considered self-centered and egotistical to put yourself as a character in the story. I'm happy(?) my score was very low, indicating that I'm not a basket-case.

Social words: I was writing a story about two people. OF COURSE I was going to use social words. Duh. Also, of course I'm outgoing and well-connected to others-->I am a Facebook whore.

Positive emotion words: I think I may have used a few happy/good words in my brilliant piece of literature, but I don't think my creativity indicates anything about my optimism or mood.

Negative emotion words: Similarly, I don't attribute my use of sad/bad words to my pessimism or anxiety.

Big words (words with more than 6 letters): Essentially, I am a genius. I chose to use big words because they are more fun to use than smaller words. I am so so smart and got a perfect score on my SATs and tomorrow I'm being inducted into MENSA. Unfortunately, I'm not very happy about that because I'm distant and have no emotions.

While this assignment was fun and entertaining, it honestly wasn't very indicative of anything. I wrote whatever came off the top of my head and I even had to be reminded to keep typing because I got a little bored with my own made up story.
According to Keila and Skillicorn, my few (zero actually) self-references is supposed to show my "dissasociative" nature. This is dubious to me because the story didn't lend itself to any opportunity to say "me, my, or I." I counted my exclusive words and I had a few, and my story was "less cognitively complex" (it was mostly gibberish), but my story was COMPLETELY fabricated. According to previous research, in 409 words, I shouldn't have had very many (if any at all). Additionally, my task scores don't match up with research regarding negative emotion words; I was well below average, even though I made the story up out of thin air.

A9 Option 1: The Return of Jeff

This title has nothing to do with anything.




I decided to do Part I, and took the Perceptual Style Experiment (“You Are What You See”). I wrote for five minutes about a picture of a water bottle, though for me at least, it wasn’t quite as boring as it sounds. I got to 175 words, which was reliable enough for the data analysis. But I found the analysis to be unreliable in major ways. I’m sure that’s one of the reasons why it says at the top of the page: “Take what is said with a grain of salt.” When I was ready to write and the timer began, I described the features of the water bottle. I talked about the cap, the color and opacity of the bottle, described the label in the middle, how much water was in it … things like that.



I was shocked and horrified by the results of the analysis. It explained that I was completely normal in every category. Not what I was hoping for. I was glad to see, however, that a PHP page with major coding flaws could tell me what it did. Here were the results:

































Visual DimensionYour dataThe average response
Words on the label: Verbal thinking1.711.74
Colors and text: Visual sensitivity4.573.74
Bottle contents: Functional thinking1.711.67
The bottle itself: Tactile sensitivity2.292.91
Light and shadow: Contextual thinking1.710.79


So for one, I found these results to be wrong in almost every way. I spent little time describing the words on the label, just stating that they exist. Maybe this is average, but I’m not too certain. I spent almost the entire time describing the color and opacity of the bottle, label, text, etc… and it still said I was average in every way. Considering this was a percentage of my total text, I highly doubt this was average. I spent almost no words on function, but somehow, I got the average score anyway. Same for tactile sensitivity: I never even described the feel of the bottle. And the most hilarious and telling result to me as the last one: contextual thinking. I never once mentioned the context, but because it picked up the word surrounding, lighting, or something (I had been talking about the logo on the label), it assumed this is what I was doing. Brilliant language analysis.



There are entire CS classes devoted language parsing and analysis, so I was pretty skeptical coming in, considering PHP is not exactly well-equipped for proper analysis techniques. Sure enough, the script does something very simple, and very flawed: it looks for key words. My buddy Killer Cam (aka Cameron Hall) is right: we could make this page and it might even be better. The whole time I was writing, I wanted to analyze the visual aspect of the bottle, but my score ended up being average for everything.



So this relates to deception in a couple of ways. For one, time to construct thoughts and write played a crucial role in the experiment and in all analysis. If you write for an extended period of time, it’s possible to start acting out of character. This is why there was a cap on time for the experiment. More to the point, having more time would allow you to have more time to tailor your message and deceive others. So, it seems to me there is a balance that needs to be struck between thee two factors when it comes to detecting deception through literary (or any) analysis. Moreover, message length (somewhat related to time) needs to reach a certain point to be sufficiently analyszed. Many people barely reach 50 words (the site’s stated minimum word count) in their text communications. Another deceptive thing people can do is selectively choose what we write about in our analysis, which of course, would effect what someone reading our description would think the bottle looks like. Someone below “stole” my point that it’s even possible to describe things in such a way that it might seem like a totally different object.



Finally, there is a fair amount of research about writing styles between various subgroups of people (ie. genders), but not as much about deception. It is believed that linguistics can be used to catch a liar because by investing in lying, your writing style, which is subconscious, changes because you can’t control it. People practicing deception generally use fewer first person pronouns and exclusive words, and more negative emotions and action verbs. Deceptive messages are also more informal and expressive, and are less complex in their syntax and verbiage. All of these things can be used in language analysis to find deceptive messages. However, until we learn to better analyze these things with more advanced tools, analysis will remain fairly imprecise and inaccurate.

Ruth's Not Pregnant

Assignment 1: The TAT

I took it upon myself to participate in one of Mr. Pennebaker’s research studies. It was actually quite an enjoyable experience—as I had the opportunity to chronicle the anxiety of Ruth as she awaited the results of her pregnancy test. As I was writing I couldn’t help but wonder what this might say about my personality, but what the hell, it was all in good fun. Upon completion of the assignment I more seriously pondered the merits of such a research study.

The TAT (the specific study I participated in) analyzes words you use in your story about an ambiguous picture using the LIWC (Linguistic Inquiry and Word Count). The idea is that through analysis of your story there might be gained some incite as to your personality. The analysis was similar to both the Zhou and Keila readings. Each developed a list of specific types of trigger words, the frequency of which would indicate something about the person sending the message. The LIWC focused on themes of the story and also self references, social words, positive emotions, negative emotions and big words. Based on your score you could infer something about your personality and compare yourself with the average score of the appropriate gender.

The linguistic analysis of the message, for means of identifying deception, raises some interesting issues and patterns. Zhou’s findings were consistent with most previous similar research. Deceivers used more modal verbs and fewer individual references, deceivers disassociated themselves from the message, resorted to less complex language etc. She also made a point about deception that I think proves especially important and plays a significant role in an exercise like The TAT.

Most previous literature focuses on statements of facts or recollections. As such, it is proposed that those stories that include very little spatial and sensory details are deceptive, because they aren’t describing an actual event. If they were describing a real event their recollection would be rife with spatial and sensory details. In Zhou’s present study, deceivers are asked to lie about a hypothetical situation. As such deceivers may be intent on being persuasive and Zhou believes as a result, those who were deceiving packed their stories with spatial and sensory details—effectively reversing the previously studies pattern. Whether or not her conclusion for this effect is correct, it still indicates that the context in which deception takes place plays a significant role in being able to draw consistent linguistic patterns involving deception.

The TAT is a situation in which someone is not limited in how they choose to describe a scene. In my case, I told a story that included first person dialogue between the two characters. As a result my scores were probably skewed towards using more self references. On the other hand, that may not make a difference, but I think in order for a test like TAT to be more predictive the manner in which a picture is described needs to be more controlled. Like Zhou’s study, where the context of deception is a key, the context of description here plays a role.

Normal is so boring

The TAT test basically flies in the face of everything I’ve ever been told, or believe, about myself. I wrote a 387 word story about a woman in science…which may or may not have been a take-off on the life story of Madame Curie. Anyway, perhaps because the story was not actually about me, the analysis was entirely out of whack.

On the Need for Achievement scale, the higher the number, the more you felt a need to achieve. It is undeniable that I have a serious need for achievement; it reaches unhealthy levels periodically. But according to the LIWC test, I have relatively little drive. While I did get cut-off before I got to the part where my fictional character wins her second Nobel Prize, I am surprised that the mentioning of the first did not send my score into the higher regions of the scale. Further, my entire story was about women breaking down science barriers and taking a lead in the field. I think something must have been off in the analysis.

I’m also low in a need for affiliation and power (both clearly untrue), and it’s kind of a toss-up whether I’m a positive or negative person (mid-scale on each).

Although it seems to have failed in the case of my own, personal, analysis, I think the LIWC test does have some predictive validity for deception. As Keila and Skillicorn point out, deception theory suggests that deceptive writings include a reduced frequency of first-person pronouns and exclusive words. Two of the LIWC dimensions are Self-references and Social Words.

Using fewer first-person pronouns enables individuals to dissociate themselves from their lie. Interestingly, according to the fine folks at LIWC, the use of big words has the same effect. While I did use a fair amount of big words, I made no first person references. Thing is, the directions for this exercise did not include that the story had to be about yourself. I was writing about one of the most influential women in the history of chemistry and physics, clearly not an undergraduate Communication major. Had I been given instructions to write a story about me and not an imaginative story with relation the provided picture, I might have lied. And research suggests that in that case, I would have used fewer first-person pronouns and bigger words.

Exclusive words are also indications of deception. Words such as “without,” and “but,” indicate a cognitively complex story, while lies are frequently not as complex. I used a lot of inclusive, or social, words in my story. Pennebaker et al. would say that this means I am outgoing and socially connected with other people; Keila and Zhou would each say this means I am a liar.

Assignment 9, Option 1: “I’m Normal.”

For this assignment, I described every aspect of my personality in only fifteen minutes of time, with no preparation or planning. Needless to say, that was not exactly the case; however, I did attempt this extremely difficult task. After visiting Jamie Pennebaker’s website of online research studies, I chose to participate in the personality study and was hoping to learn a good deal about myself, as well as, online methods of personality profiling.

After typing continuously for fifteen minutes and clicking “finished”, a new window with one’s personality analysis automatically appears. The study divides one’s personality into seven categories, including physical appearance, family orientation, social connections, achievement striving, religion and spirituality, optimistic orientation, and negative concerns. My self-description produced “normal” response levels for each category. The website states that twenty percent of people who participate in this experiment score in the normal range for each category, and “normal” means that the computer did not find anything particularly important along any of the seven dimensions.

I was a little surprised with my results, especially in the categories of self-appearance and family orientation. After reading through my response, I thought that I included a great deal of information, which the study considers to be associated with physical appearance. I found this category easy and simple to type about for a long period of time, and I included information such as, “I have blue eyes,” and “I am 5’5’’ tall.” Additionally, shortly before completing the study, I had spoken on the phone with my mom, and I believe that conversation led me to spend a significant amount of my time typing about my familial relationships. These two aspects of my results seemed a little odd to me, but it is hard to judge whether I believe them to be accurate because I do not know how exactly the study evaluates responses. More information about what words trigger activation of the computer would be interesting to know and would help in analyzing the usefulness of this activity.

I think that this method has the possibility of detecting deception in a text based format, and may even turn out to be more accurate in detecting deception than in identifying characteristics of one’s personality. I did not feel that my ratings reflected a true picture of what I related in my description, but this could have simply been because I did not include typical triggers that the computer uses in order to rate the individual categories. However, the program does have some attributes, which make it promising in deception detection. According to Keila and Skillicorn, deception leaves a “linguistic signature” in messages that are produced because language production is fundamentally subconscious and the cognitive load present when deception occurs decreases the deceiver’s performance in other areas. Zhou et al. also focused on language behavior as a subconscious process, and thus felt that deceptive clues would “leak out” in email messages. These studies analyzed deceptive messages and scanned them for first person pronouns, exclusive words, negative emotion words, and action verbs. It is feasible that this methodology could be applied to written statements and be quickly analyzed by computer programs. The accuracy of detecting the signal words would be very high, if a sophisticated computer program could be developed, and since the writer is typing continuously for a full fifteen minutes, the content of the statement is mostly unplanned and does not provide opportunity for editing. Zhou et al. purport that when conducting a linguistic analysis the cues known to be relevant to deception should be studied and other signals should not be included. If this is a true key to deception detection, this method would work extremely well because it would operate on a program designed to track only certain key words and phrases. For these reasons, I believe Jamie Pennebaker’s experiments would be more useful in deception detection than they currently are in personality profiling.

Assignment 9- Not What I Was Expecting

I was interested in doing the personality test, but my computer wouldn’t analyze the language; my data was blank every time. Although I was tempted to give up on Jamie Pennebaker’s website, I decided to give it one more shot, and my results were correctly analyzed for the perceptual style task. After reading the results, I admit that I am greatly surprised. Although I did not anticipate writing about an empty water bottle on a gray background for five minutes, I talked about how the water bottle is a genius idea that made a lot of people rich. I went on to describe how the water bottle is somewhat a symbol of our modern, fast-paced society. We turn to items that are most convenient for our lifestyle, although we pay a high price for bottled water. After this analysis, I went on to describe how advertisers market water bottles, which fonts and colors they use, and why these might be pleasing to the eye. 267 words later, I read my results, and I continued to lose faith in Jamie Pennebaker.

For the verbal thinking portion of the test, I scored far below average with a score of 0.37. This means that I did not talk about the exact letters or words on the sides of the water bottle, such as “OZARKA.” This result is unsurprising, because I did not think it would be interesting to describe the text. The next portion of the analysis was the visual sensitivity portion, for which I also scored below average with a score of 1.12. This indicates that I was not interested in describing how the text was displayed. However, this result left me greatly confused. I would say that at least one third to one half of my five minute analysis was spent discussing the colors, fonts, and appearance of the water bottle. In addition, I talked about ways to market or advertise this water bottle to a modern society. Thus, I was very surprised that this program did not pick up on my visual and descriptive language. Yet, as I thought about this further, I realized that this reveals some fundamental truths regarding language. Although a computer algorithm can attempt to measure language, it is only analyzing and categorizing words in a systematic way. For this test, words might be placed into categories which would contribute towards scores in particular areas, such as visual sensitivity. However, no computer program can measure intention, thought processes, or successfully translate language into what the writer meant. Further, I question the ability of this program to pick up on sophisticated language, such as metaphors or complex terms.

Back to the results, I scored a 2.25 for the functional thinking portion of the test, which means that I spent considerable time discussing the fact that this item is designed to hold water. Again, this result left me feeling very confused, even though I did mention that we trust that the contents inside the bottle are nothing more than pure H20. However, this demonstrates that this algorithm picks up on certain words that may not necessarily belong in a certain category because they are “typical” to descriptive language from that category.

For tactile sensitivity, I scored low, for which the computer program determined that I do not appreciate the dimensionality of objects. According to this program, I can never be a sculptor, artist or graphic designer- all activities which I spent the majority of my free time pursuing at Cornell. On the last portion of the test, contextual thinking, I scored normal, which means that I have some ability to step back and see things from a broader perspective.

Overall, I would argue that this type of assessment is highly inaccurate for categorizing language. I argue this not because I found the results to be surprising in terms of my activities, but for the fact that they almost completely contradict what I had in mind when writing for the period of five minutes. It was my intention to focus on the visual aspects of the photograph, and this is the topic of most of the writing that I produced. Yet, I was rated lowest for this portion of the test. After re-reading the text I wrote, I am still perplexed as to how this can be true. Of course, I know that language is very complex and really needs human thought for analysis. No computer program can read into the words that are written; they can only form simple analyses.

It is difficult to find relationships between deception and this type of language analysis, because I feel that the results are so inaccurate. As we have discussed in class, it is very difficult to detect lies based on verbal cues alone, which is why many individuals are very successful at deceiving others in text-based, asynchronous media such as the Internet. However, there are some subtle connections. For one, we can relate a task such as this to the “showing and hiding” form of deception described by Vrij. If we are to describe a photograph to someone who has never seen the photograph before, it would be easy to leave out essential details, or to focus on small details which are irrelevant. By using this method, we could trick someone who is relying on only verbal cues to assess a situation or imagine a particular object. With language, we might even convince them that a completely different object was shown in the photograph.

An interesting twist to this type of test would be to measure someone’s verbal descriptions instead of text generated from a keyboard. For one, we know that the words we use in casual conversation or to verbally describe something vary significantly from the words we write. In conversation, we often use conjoined phrases, rarely speaking in fully sentences. Spoken language is also a lot more informal. In relation to deception, we also know that deceivers have less time to make communication plans in synchronous media, where they do not have time to tailor the message to a particular audience or to go back and erase what was previously said. In my opinion, Jamie should spend her time looking into the differences between spoken and written language instead of creating computer programs which fail to make accurate assessments.

Assignment 9: What's your personality like?

Option 1

I went to Jamie Pennebaker’s website and participated “Personality: Describing yourself.” For this experiment, you are required to write about yourself for at least fifteen minutes, doing your best to describe your personality. You are then given an evaluation that describes your personality in terms of seven different dimensions based on the words you used to describe yourself. The seven dimensions are physical appearance, family orientation, social connections, achievement striving, religion and spirituality, optimistic orientation, and negative concerns.
So, how accurate is this kind of approach to language? I personally believe that this kind of approach is relatively inaccurate. First of all, my score was “normal” for each category which I certainly don’t think is true of myself. Second of all, the nature of the experiment brings up some interesting issues. For example, the results of the experiment are prefaced with “Keep in mind that the more words you wrote, the more trustworthy these analyses.” So what about slow typers? Or people that fumble with their thoughts or organizing their thoughts? Let’s think about the Channel Expansion Theory maybe not in a deceptive context. This theory implies that the more experience a sender or receiver has with the channel, the richer they find that media. Someone more experienced with computers may have the simple advantage of typing faster. Should this really have an indication of someone’s personality? I think the method of analysis is also important to determining the accuracy of this kind of approach to language. The method of tallying the words I used in my personality test to indicate certain personality traits is similar to that of Zhou et al. and Keila et al.. Essentially they use a measurement system and statistical analysis to search for and tally up cues they think are relevant to deception and test to see if they were accurate in comparison to the actual deception level present. But Zhou et al. makes a really good point that “observing deceptive behavior over the course of an interaction rather than from individual messages may yield the best predictive models.” I suppose then, the personality test may be more accurate or believable if I took it several times and continually got the same results.
Another obvious similarity in the methods of observation is the use of linguistic cues as an indicator. Both of these studies observe language features. For example, the personality test described someone focused on physical appearance as someone that used many physical descriptions of faces, bodies, physical health or manner of dress. Deception, on the other hand, was based on features such as informality, expressivity, complexity (Zhou, et al.), or word counts on things such as first-person pronouns and negative emotion words (Keila et al.). I think this kind of approach is more appropriate for deception. Zhou points out the success in capturing relationships in the deception data and the overall generalizability in their results. I think this is a lot more legitimate because there’s sometime more tangible to tie is back to. That is, we know when deception is present for their study so we’re really just testing in the classification method is accurate. In the case of personality testing, the results are a lot more subjective. Not that personality is a lot more ambiguous and the method of measuring frequency of words isn’t a viable technique. Before beginning to write, I was prompted with questions of possible things to write about. Doesn’t this make the whole process bias anyway?
To sum it all up, I don’t this approach is reliable, but maybe sometimes valid. This type of analysis, however, may be more appropriate for deception.

Assignment 9.2: I know how you like my style

I would like to begin by saying that Jamie Pennebaker’s site is awesome, but I could make it. However, he uses a very extensive and impressive algorithm that I could not build on my own. I will get back to this in more detail later. I decided to go with the ‘perceptual style’ because I don’t enjoy writing for extended periods. I had to write about a water bottle for 5 minutes. They say that a picture is worth 1000 words, but I could only think of 168. I found it pretty difficult to write for the entire time about something so basic. The time goal is a double edge sword. I was forced to talk about things that I might not have if there wasn’t a clock. However, if I wrote a few sentences, the data would be very inaccurate. I scored right around average for all of the topics. The two dimensions with the greatest difference from the average were tactile sensitivity (low) and contextual thinking (high). Most of my contextual thinking came towards the end when I was running low on ideas. If I was asked to only write for 3 or 4 minutes, I probably would have skipped this category all together.

I click the link to find out how they calculated the table and learned that they used a program called the Linguistic Inquiry and Word Count (LIWC). As with most computer calculated statistics, it works best under conditions with high volumes of data. This is similar to many of the issues address in the Burgoon, Nunamaker, and Twitchell paper. In other words, the calculations would be more accurate had I provided twice as many words. However, as I addressed above, giving a participant too much time can cause them to act out of character.

How can this apply to deception? As I stated a couple times, there is a critical balance between amount of time and the ability to remain normal (like the self). Once this limit is exceeded, the subject may act sporadically. If you were to use this same test to analyze the deception level of a subject you would have to perform three tests. The first would be to make up a story based on a picture or drawing (the 3rd option I believe). Next, they would be asked to describe an event from memory. Finally, they would be asked to describe the event under investigation. This of course assumes that the statistical data (clusters for LIWC) is known for each test. Next the results from the table would be compared to examine changes in performance. The results should be similar to those from the memory story. I would expect this test to be inaccurate if possible, because it is unrealistic to assume that the statistical clusters for each story are known.

Assignment 9 - Not what I expected...

For Assignment 9, I decided to go for the first option which challenged me to do one of the research studies, particularly the one called Perceptual Style: You are What You See. This task asked me to look at a picture of an everyday object and write about it for 5 minutes as if I was describing the object to someone I couldn't see. The object I was asked to describe was a normal-sized plastic bottle of spring water with a red label on it. So, I described...for 5 minutes.

The LIWC results were kind of counterintuitive and suggest that I should be in Materials Science or another engineering field rather than Communication -- though I beg to differ. In my description I opted to offer a lot of details regarding the actual feeling and shape of the bottle itself. I wanted to familiarize the person I was describing it to with what the bottle would actually feel like if it was in their hand and I thought these were important descriptors. In doing this, I discussed the actual function of the water bottle, with a lot of details that supported and explained the exact function of the plastic bottle. These two dimensions, Functional thinking and Tactile sensitivity as they were coined, were the most used characterstics in my description. On the other hand, I used no descriptors relevant to the actual words on the label or the colors on the label that might be helpful, according to the summary I received. But, I thought, these are obviously not helpful to a person who can't see the label, and that's why I decided to discuss the important factors to someone without sight: the feeling of the object that should be recognizable. The summary mentioned that because I didn't mention anything about the words on the label specifically, then I'm not someone who is particularly interested. Wrong! That's very contrary to the truth and I think it's very ironic, and very funny! It also said I'm not someone who doesn't need to discuss the obvious, which, to be honest, is sometimes the truth though sometimes it's not -- it varies, I guess. And then, because I didn't mention anything about the colors on the label, I apparently won't be a graphic designer. That was a former goal of mine, but that no longer rings true, so I guess I'm doing okay on that.

Overall, I thought this was an interesting task. Like I said, the irony of it is that I described what I thought was important to the potential receiver of my message and was very sensitive to their needs, but that apparently is contrary to what I normally would do -- which is pretty obvious as a Communication major who LOVES words and verbal characterstics of language. On the other hand, I apparently was good at describing the physical characteristics of the bottle, though it really bears no weight on my future or what I really care about, honestly. I think tasks like these are interesting but don't really work or mean anything. It's a simple task that anyone can do and that anyone can fabricate. Therefore, I don't think there are too many implications with regards to deception because it's something that can be done over and over again, have a different result, and not be accurate of the actual person, even if he/she is telling the truth or lying -- basically it's based on a case-by-case basis. It was fun, though, realizing that being sensitive to my subject turned out to be contrary to my typical writing style or the goals of my typical writing focus.

Monday, November 13, 2006

I am how I feel...according to LICW

Seconds after completing the Perceptual Style Experiment: “You are what you see,” I was immediately informed that I am a person concerned with achievement and success. Based on my above average in the categories: “need for achievement,” “positive emotions” and big words; below average scores categories: “need for affiliation,” “need for power,” “self references” “social words,” “negative emotions,” the LICW test has made an inference on my perceptual style and my personality. While, yes, especially in times like these when I am bombarded with tests and papers, I am concerned with achievement and success, I believe this analysis has amazingly diagnosed my current state of mind, while it has failed to assess my true personality.
The LICW approach could in fact be a very useful and accurate tool for text-based deception detection. There are several aspects of the categories used to determine my personality traits that could indicate deception within text-based communication. Keila states that “deception leaves a linguistic signature, both because language production is a fundamentally subconscious process and so affected by emotional states associated with deceiving and because the cognitive demands of deception cause performance deficits in other areas.” It has been found that individuals who are trying to deceive generally use less first person pronouns (indicating attempt of the author to dissociate themselves from words), and exclusive words (such as except, without etc… indicate that a less cognitive complex story), and more negative emotions (indicating feelings of guilt) and action verbs (because the deception communication is less cognitively complex, the text should contain more action words). First person pronouns would be detected in the category for self references, and negative emotions would be detected in the category “negative emotions.” Similarly, the Zhou article contends that “deceptive messages have been found to include higher informality and expressivity, and lower lexical diversity and complexity.” Several of these features including the big word assessment could indicate deceptive messages as well according to Zhou’s assumption regarding the diction and of deception messages..
The method used to discover my perceptual style could also be very useful in detecting deception in corporate emails as well in that the “need for affiliation” and “need for power” can establish a motive for successful deception. Therefore, not only can the method detect deception in the language, the method can also detect an inherent meaning within the content.
I feel that this method is not flawless. If this method were used to detect deception in emails, or other CMC correspondences, I feel it would be difficult for the LICW to detect deception based on very few words. Although emails may be long, the LICW detected personality traits that were not completely accurate based on a story that took me 10 minutes plus a few seconds. Based on our participation in the diary study, very few people spent more than 10 minutes constructing emails. When applied to the Enron emails, individuals are constantly under tremendous time pressures. I’d like to infer that very few emails took more than 10 minutes to complete as well.
In addition, because the Enron emails analyzed in the Keila study were of a business nature, it is unlikely that many self references and emotions were evident. Perhaps emails indicated social words, considering there are often team efforts in business settings and some emotions regarding projects. These aspects are important to consider as well.
The LICW is a very interesting tool and I feel it could be extremely useful. At this point, I feel it will be most useful in lab settings when text is controlled. At this point, it would be hard to judge several different kinds of text-communication regarding deceptive nature based on general indicators.

Assignment 9: I shouldn't be a graphics designer.

For this assignment I did the perceptual analysis test. I was shown an image of a plastic water bottle, and asked to describe it to someone who couldn’t see the image, to try to give them the best understanding I could of what the image looked like. My description was then analyzed using LIWC, and it was used to make a (semi-accurate) prediction of my personality.

This process is known as Meaning Extraction. Before this study was done, a large number of people were given this task, and their text was analyzed. The researchers found 5 different clusters of words, and these 5 clusters are what they look for when someone now performs this task. The 5 dimensions were: Words on the label: Verbal thinking, Colors and text: Visual sensitivity, Bottle contents: Functional thinking, The bottle itself: Tactile sensitivity, and Light and shadow: Contextual thinking. So when someone now performs this test, they search through the text to find words that fit into each of these 5 categories.

Your results are then compared the average responses in each category, and then they gave a brief “prediction” of your personality based on how you compared to the average. Because of a high score on contextual thinking, they predicted “This is healthy in the sense that it suggests you can stand back and look at objects in a broader perspective.” From a low score on tactile sensitivity they predicted “Extending this reasoning a bit, you don't automatically imagine how an object feels. Touch is not your dominant sense. I just can't see you as a sculptor, a painter of still lifes, or a postmodern architect.” As the researchers said themselves, this part of the analysis was kind of “made up.”

The importance of this analysis, however, isn’t in predicting that “Graphic design may not be a wise career choice,” but in being able to analyze the text of a message. The researchers were able to accurately categorize my text into different categories, and correctly noted which categories I did tend to use more frequently than others. Now, none of these characteristics (to my knowledge) are related to deception detection, so this specific analysis probably would not be helpful for someone trying to find deceptive messages.

This method of analysis, however, is done in the style of many successful deception detection techniques. Zhou et al notes that when doing a linguistic analysis, accuracy rates increase greatly when only known important cues are studied, and the irrelevant cues are disregarded. I believe that the researchers followed this in their analysis, by looking only at words in certain groupings or styles to figure out which category the text belonged in. Also, they only looked for the relevant words that they noticed in their pilot study, and therefore in the current analysis the results are not muddled by other words which may or may not be relevant to the analysis. I believe that the researchers would also agree with Keila et al that different types of situations (like deception) leave a linguistic signature. Deception obviously is not studied in this analysis, but the researchers do imply that through linguistic analysis you are able to tell certain things about the person or the situation. I believe that this type of analysis is very effective, and could be adapted to deception if specific words that are identified as being related to deception are searched for.

Assignment 9: Option 1: now TAT’s just silly!

So I did the TAT test and I wrote 362 words about a girl in high school who decided to spend a summer in a laboratory with a professor instead of going to the beach or to camp with her friends. (Ok, so I drew from my own dorky life. Sue me.) Anyway, they did a rather fast automated linguistic analysis on my masterpiece and came out with this:


LIWC dimension My data Male avg Female avg

Achievement 1.93 5.8 5.6
Affiliation 1.38 1.1 1.3
Need for power 2.21 1.7 1.8
Self-references 0.83 0.5 0.8
Social words 14.64 11.4 12.0
Positive emotions 3.31 1.8 2.1
Negative emotions 0.83 1.5 1.6
Big words 18.78 18.7 17.7


Then, they gave me this nice little description of what the numbers all mean. There were a couple of interesting results. Apparently, since I used a lot of self-references, I tend to be more insecure, nervous, and possibly depressed. Also, since I used a lot of social words, I am more outgoing and more socially connected with others. So, basically, because I wrote so many of these words, I am a depressed, insecure socialite. Well, that seems likely. Another conflicting little tidbit: even though I’m depressed, I have a pretty positive outlook on life. Isn’t that nice?


As though they had known my assignment, the brilliant people at the LIWC place (wherever that may be) also gave me a description of the depressed people that included their tendency to be honest. It seems consistent with what we’ve done in the past since we know that liars like to distance themselves from their lies. However, for this particular task, I think that there needs to be more accurate classification/coding approaches as described in the Zhou article. I may have used self-references in the story but I wrote it in the third-person format. I only used the words “I” in dialogue of the other characters. In that sense, it wasn’t really a self-reference so much as it was a character reference.


As for the TAT’s ability to assess my personality, I’d say there’s a lot of work left to be done on this system. Not only didn’t the results all make sense, some were just plain impossible. Obviously I cannot be both introverted and extroverted at the same time. I may have had above average on both kinds of terms, but both may not have been above average for me. They could have done something with the z-scores or used a more logical method of classifying me. I do think, however, that the kinds of words we use can give insight as to who we are as people. I just don’t think this does it.

Option 1- TAT Test

For this assignment, I chose to take the TAT test. I was shown a picture and asked to study it for several seconds. I was asked to then write an imaginative story about the people in the picture, including their backgrounds, thoughts, and feelings. For ten minutes I made up a story about the two female scientists shown in the picture. I pretended that one of them was Madame Curie and the other was her jealous friend, Sara, who was eagerly watching her perform an experiment, hoping she would not succeed. I described how both women were the first to ever be admitted into the physics program of some European university, and were now competing to be the most famous female scientist that ever lived. After my ten minutes were up, I was shown a write-up regarding the meaning of my responses. This included a short analysis of my word use. Overall, I wrote 369 words and this is how the LIWC (Linguistic Inquiry and Word Count) analyzed them.

LIWC dimension My Data Female average
Need for Achievement 8.67 5.6
Need for Affiliation 1.90 1.3
Need for Power 0.54 1.8
Self-references (I, me, my) 0.00 0.8
Social words 12.20 12.0
Positive emotions 2.71 2.1
Negative emotions 2.44 1.6
Big words (> 6 letters) 21.41 17.7

For every LIWC dimension, excluding “need for power” and “self-references,” I exceeded the female average. The idea of this analysis is that your writing is supposed to reflect who you really are. This notion confused me, as the story was not about me but about an imagined event. For instance, my score on the dimension of “need for achievement” was above average. The explanation given by the analysis says that the higher my number, the more I wrote about achievement-related themes. My story was indeed about achievement. I wrote about two women who were competing for fame. I am still not sure, however, how this applies to me. Is this assignment supposed to show that we are what we write? I can, however, understand how dimensions such as “big words” might say something about one’s personality. I agree with the analysis given that people who use big words might be less emotional (not in my case, however). I was really surprised that I scored below average on the “need for power” dimension, considering that my story was about a power struggle between two female scientists.

I was not at all surprised to see that I had not made any self-references in my writing sample. This is because I was narrating a story that did not involve me. Keila and Skillicorn also explored the use of first-person pronouns. They hypothesized that deceptive writing generally contains fewer self-references because the author wants to dissociate themselves from what is being said. They, however, faced a similar problem with this dimension. Because the emails they analyzed were written in a business context, they found that the use of first-person pronouns was extremely rare. Because business emails tend to be more formal and to the point, the use of these words are somewhat inappropriate. In my situation, because of the imaginative nature of my assignment, the use of first-person pronouns would be somewhat inappropriate as well. Keila and Skillicorn also discuss the frequency of negative emotion words, stating that individuals who are trying to deceive generally use more negative emotions words. In my case, I used far more negative emotion words than the average female.

For all of these reasons, I think that the TAT test might be limited when applying it to deception. The test specifically asks you to be imaginative and get into the thoughts and feelings of the characters in the picture. I was trying to be as creative and vivid as possible in my writing based on the task at hand. I associated imagination with exaggeration and truth-bending. In other words, I believe that an imaginative story is by nature somewhat untrue, which would make it very hard to accurately analyze these stories in terms of deception. I think it would be far more applicable if the assignment was to tell a story about a personal experience. While SVA is a truth-verifying method and Reality Monitoring is more of a lie-detection method, the TAT test is really just a personality assessment tool. Furthermore, as my experience with the test has shown, it is not a very reliable personality assessment tool.

TAT test

For this assignment, I decided to do the projective test (The TAT). This test showed me a picture from the Thematic Apperception Test, and asked me to describe it. The purpose is to see how individuals reveal parts of their own personalities while looking at an ambiguous picture. The words that I typed were analyzed using the LIWC (Linguistic Inquiry and Word Count) program developed at the University of Texas and University of Auckland in New Zealand.


These are my results:

Need for Achievement 6.96

Need for Affiliation 0.28

Need for power 0.56

Self-references (I, me, my) 0.00

Social words 6.69

Positive emotions 1.95

Negative emotions 0.84

Big words (> 6 letters) 18.11


The results of my story showed that I was just about average with the other females who have taken this test. I ranked higher on the dimensions for “need for achievement” and “big words”, but everything else was either average, or slightly below average. Based on my results, I am not really sure how accurate this kind of approach to language really is. I received below average scores on certain categories that I thought I had included in my story. For example, I received a low score for “need for affiliation”, which is about the attention to human relationships in the story. I got a 0.28, and the average was a 1.3. I thought this was strange, though, because I did talk about the relationship between the women a lot. I do not think that this approach can pick up subtle ideas from the text, and that it can only detect explicit use of language.

This approach could be used to detect deception in the ways that Keila and Skillicorn did their study on the Enron emails. They used the factors of fewer first person pronouns, exclusive words, and more negative emotion words and action verbs to detect a lie. The TAT test’s categories look at negative emotions, positive emotions, social words, self-references, and big words, which could be used to detect deception. This method could also backfire because of the way in which they ask people to tell a story. Since I was telling a story about these two women, I did not reference myself at all; therefore I did not use any first-person pronouns. Keila and Skillicorn also ran into this problem since their emails were mostly business related, and using first-person pronouns in a business context is usually not appropriate.

The Interpersonal deception theory (IDT) attempts to explain deception from an interpersonal and conversational perspective, rather than an individual and psychological perspective. It says that deceivers will display strategic modifications of behavior in response to a receiver’s suspicions, but may also display inadvertent behavior, or leakage cues, indicating that deception is occurring. I think that the projective test using the TAT would be good at detecting the leakage cues that are talked about in the IDT theory. When people write about an event, they use certain words without thinking about it, and this test would be able to find which of those words, such as first-person pronouns, are signs of deception or not. The projective test could also look at the use of negative and positive words to detect deception, since those are usually used by people without them realizing it.

Overall, I think that since I was asked to make up a story, it did not reveal much about what a lie would look like. I was being creative, and believed in the story I was writing, as opposed to trying to deceive someone about something that did not actually happen. I think if I tried to lie about an actual experience, my results might have appeared to be more deceptive.

Describe Yourself in 15 Minutes

Assignment #9
Test: Personality: Describing Yourself


For the assignment, I attempted to describe myself in a textbox given a mere fifteen minutes. The test at Pennebaker's site asks users to simply type for fifteen minutes straight, describing any and all aspects of who they are and what defines their sense of self. Although this may sound easy, it's actually more difficult than it sounds. The fifteen minutes went by pretty quickly, as I simply typed whatever popped into my mind without going back and editing. I did have to pause frequently though to think of what to say next. I found that my description of myself was fairly basic, with very short, simple statements. Most sentences started with "I am...", "I like...", "I dislike...", "I do...", "I enjoy...", etc. My final word count was 533.

The analysis concluded that I was... pretty normal. Of the seven categories that people tend to type about (physical appearance, family orientation, social connections, achievement striving, religion and spirituality, optimistic orientation, negative concerns), I was in the "normal" range for 6 out of 7. The only category that was abnormal was "social connections," in which I scored "very high." According to the website, "high scores [in this category] reflect an awareness about how individuals interact with other people."

It's difficult for me to really judge the accuracy of these results. I've never really thought of myself as being especially attuned to "social connections" moreso than the other attributes that the test evaluated. I actually felt my response came across as especially negative, so I expected that category to be higher. And I barely mentioned any sort of "physical appearance" attributes, but it said I was "normal" for that as well. I suppose it's tough to judge how my self-evaluation was in comparison to others because I don't know how others would evaluate themselves.

I think these sort of text-based analyses definitely have potential, especially for deception detection as discussed in Zhou et al. and Keila & Skillicorn articles. They might even work better at detecting deception than making any sort of personal analysis. With a number of linguistic cues occurring more often in deception messages than not, a system of evaluating and ranking messages seems the most logical and effective, as Keila & Skillicorn did with the Enron emails. Creating a continuum that easily separates out linguistically "abnormal" messages can at least help point people in the right direction or raise suspicion. Perfect accuracy is unlikely to even happen, but such text analysis is bound to help identify many lies that would potentially go undetected.

Sunday, November 12, 2006

TAT- Option 1

For ten minutes I fabricated a story about two female scientists, who had been presented to me in a picture. I focused on the idea of achievement, as do many, discussing how these two women, despite gender prejudices in 1922, managed to discover the “jumping gene” through hard work and dedication. Although the two scientists had developed a rather strong friendship Debbie took advantage of Mary’s submissive nature. Debbie posed her belief that the two of them would not gain any real success if they both took credit as it would reflect that women never can achieve things independently. Mary had initially agreed that more attention would be given to their work if only one of them took the credit for the work. Debbie, being the better public speaker, elected herself to announce the discovery. She hoped Mary would not object and would remain a silent partner in the mean time. However, Mary had internal doubts about this unsaid agreement and found it very unfair that Debbie would receive all of the public praise. I then went on to elaborate how fast Mary’s hatred was growing for Debbie.

My Linguistic Inquiry and Word Count (LIWC) was:
Need for Achievement 4.52
Need for Affiliation 1.51
Need for Power .50
Self-references .000
Social Words 12.56
Positive Emotions 2.01
Negative Emotions .75
Big words 16.08

According to the TAT, I seemed to fall within the average range when discussing the achievement of the two characters. This was an accurate measure because I really didn’t dwell too much on the importance of the scientific discovery. I seemed to pay more attention to the human relationships in the story. I was in fact more concerned with the effect the discovery had on the women’s relationship. In the same respect, I also seemed to fall way below the average when looking at who had the most power in the picture. I personally felt that my story really stressed how Debbie had taken too much power over the situation and I was surprised that I had not scored higher for that matter. This brings up an important point however that this technique is not just looking at my linguistic patterns, but rather comparing my patterns to a host of other people. I should recognize that I may have scored lower because most other people had used even more power indicators than I had.
I used no self-references which could reveal to a stranger that my story was fabricated. I used a great number of social words which would indicate that I am more outgoing and more socially connected with others. However, I think that this sort of result is only accurate in distinct settings. For instance, although I am very outgoing when I am comfortable in my surroundings, in class I am quite the opposite in nature, not very outgoing and rather shy. I also fell right into the average when using positive emotions in my writing, showing that I generally feel good about myself and see the world in a positive way. I seemed to fall below average, however, when using negative emotions, which seemed pretty accurate because this particular Saturday I was having a good day. Finally, I was happy to see that by scoring a rather low score in my use of big words, I would be predicted to be more emotional and psychologically closer to my peers relative to most other individuals. I would say this is a rather accurate description of me as well.
I was given an objective, to make up a story and intentionally lie to my audience. I had to make up events that did not happen and attitudes that did not exist. The idea is that although I may have had ultimate control over the content of my story, my underlying state of mind, to lie, may leak out through the style of language I used to tell the story. Thus I was able to generate a sample of the type of language a deceiver would use which could be used to distinguish real liars from real truth-tellers. Vrij showed through reality monitoring how imagined experiences are qualitatively different from stories based on real experiences. The TAT, however, identifies a fabricator, not only by what they say but how they say it. The numerical technique, according to Keila and Skillicorn, stands on its own because rather that being an independent predictor of deception, the test, “ranks messages by how likely they are deceptive” relative to other people. It is also simple and quick, just a count of types of words. Keila and Skillicorn also point out that, “those who wish to deceive cannot easily avoid creating this linguistic signature even when they are aware that they are being scrutinized.
However, there are limitations to this method of analysis. Pennebaker, in his 2003 study “Lying Words, Predicting Deception from Linguistic Styles,” points out that motivation to lie and emotional involvement are “important moderators.” For instance, as Keila and Skillicorn predict, I did use fewer first person pronouns and exclusive words indicating I wanted to disattach myself from my statements and tell a less cognitively complex story. These are two signs that I was lying. However, to the contrary of their predictions, I did not use very many negative emotion words. I believe this is because I was not scared of the consequences of lying in this context, in fact, I was simply following the command to fabricate a story in 10 minutes. Thus I didn’t feel guilty for lying, and this could have effected whether or not linguistic patterns would leak out in the TAT.
Therefore, I believe it is important that if linguistic inquiry and word count were going to be used in the court room, researchers should closely simulate motivation and emotional involvement to the level it would be at in a highly publicized case. When comparing average counts of linguistic behavior, we are simply ranking a message by how likely it is deceptive. However, it could be possible that an individual under very different emotional contexts may score very differently because of his overall mood that day or how great he perceives the risk of being caught. I don’t believe it is at all accurate to say that most people will use the same linguistic patterns when lying. Although it helps that Pennebaker has achieved an average over many case studies, there will still always be those outliers who would not at all be protected by LIWC averages. I just think it would be to risky for a court to put all their eggs in one basket, and just look to a TAT when deciding the final verdict. There are many more factors involved besides linguistic patterns.

Friday, November 10, 2006

Assignment 9: Worrrd?

Option 1 – Word?

For the option, I did one of the TAT tests which is a Thematic Apperception Test. Basically what I had to do was look at a picture and write for 10 minutes about it. The picture that I was shown was of two women doing some sort of scientific research involving test tubes and things of the such. I went on to write for about 9 solid minutes before I got bored and ran out of ideas and ended up concluding my story by saying:

“Rosalind decides to retire eventually and moves back to the country where she would eventually die in a wagon accident. Debra moves back to New York City where she would live to be 80 years old before she got mugged by a bunch of gang members and shot in an alley. She didn't die immediately though, and while speaking to her children, she tells them that, "Anyone can have success when things are going their way, the real tough people are ones who find ways to thrive when things are going terribly." She eventually dies while trying to jump out a window and one of her kids becomes a cross-dressing bank robber that kills anyone who says the word "banana" in her presence.”

This probably effected my score a little, thought the rest of the story was about how the two women had came together and discovered one of the main causes of cancer after being rejected from working at a research institute because they were women. Basically it was a big success story (until the end), then I got bored and my ADD kicked in.

After the program analyzed my writing, and my personality scores and their translations went as follows:

-Self References: 0
I scored very low. This means that I am not very insecure, self conscious or depressed, also that I may not be as honest as other people who have these traits.

-Social Words: 14.23
I scored very high. This means that I am more outgoing and socially-connected with others than most people.

-Positive Emotions: 0.41
I scored low. It means that I’m not as optimistic as most people and I don’t feel as good about myself as other people.

-Negative Emotions: 0.62
I also scored low (maybe I just don’t have feelings?). This means that I’m not very neurotic and I don’t see the world negatively.

I guess I’m just indifferent about my feelings about stuff. (Ha, that was an indifferent statement. SCORE ONE FOR THE HOME TEAM!)

Big Words: 17.32
I scored slightly below average. Anyone who knows me knows I don’t like using big words. They found that use of big words is loosely correlated with education and intelligence, so I guess I’m average (even though I’m at an Ivy League school).


As I pointed out in many of the scores I received, I found the scoring somewhat accurate (I am very outgoing and not very depressed) though some of the scoring didn’t really make sense (my intelligence score, and if I’m optimistic or pessimistic). So in my experience, I found this to be much better at telling whether I am outgoing or not (though the ending to my story may have ruined this since I got bored).

I think there’s a SLIGHT chance that this could be used for detection deception. As in the Keila and Skillicorn study, the TAT test looks for the use of specific words. In the Keila and Skillicorn study, they looked for the use for first-person pronouns, exclusive words against the use of negative message words and action verbs. They found their method to be good for business-related emails just like I found TAT to seem pretty reliable for detecting if someone is outgoing or not.

The way that I find these methods limited is that people write differently depending on the mood that they are in. I was probably a little less positive today than I normally am because I’m having some problems with things. Also, in the end, I got bored so I just started writing things that I thought were funny so that affected my score. The main limitation with analyzing writing is you aren’t able to tell what kind of mood the people were in when they wrote them. Also, if someone is trying to deceive, we learned that people don’t always use the same nonverbal cues as other and it could be the same with verbal cues. I think they will be able to detect certain things (like the Keila/Skillicorn Enron study) but I would not use writing scores from being analyzed in the court room to decide if someone is innocent or not.

Thursday, November 09, 2006

Corey, Jenna and Robin's Project

We Flushed out our hypothesis and applied them to the theories we are researching.

Hypotheses
According to a sample of facebook users and references to their Facebook profiles, this study will address the hypotheses discussed earlier.

Based on the Berger and Caprese’s Uncertainty Reduction Theory and Walther’s Selective Self Presentation theories we expect that deception will be frequent in Facebook profiles.
H1: Deception will occur frequently in Facebook profiles.
Based on Goffman and Walther’s Selective Self Presentation Theories, we expect that there will be deception in Facebook pictures. In addition, in accordance with the Hyperpersonal Theory, people can no longer idealize each other when pictures are present. Instead, users will choose to manipulate their own pictures as a means to idealize themselves through the most prominent aspect of the profile.
H2: Deception will be most prevalent in picture profiles.
Based on Goffman and Walther’s selective self presentation theories, and Nyberg’s assumption that hiding is a form of deception, we expect that users will present information in their profiles with the intent to appear more likeable. Likewise, we expect that users will conceal information that would be perceived negatively.
H3: Facebook users will present information they feel will be viewed favorably amongst those within their current network. Facebook users are motivated to portray themselves as attractive (social identity) and therefore are likely to use deception.
According to Toma’s research on online dating, we expect similar findings regarding deception differences amongst gender. Deception will occur most frequently amongst female users in their picture profiles, whereas deception amongst male users will occur most likely in areas that may imply future economic success such as work information.
H4: Modalities of deception will differ by gender. Deception amongst female users will occur most frequently regarding physical appearance (pictures). Deception amongst male users will occur most frequently regarding social status (job info).

Assignment 9: Cameron, Brad, Lauren - survey

Please fill out our survey and post comments and suggestions.

Assignment 9: Brad, Lauren, Cameron

Introduction

Many researchers have dedicated their careers to shaping their beliefs and definitions of deception. While research on deception has gone in many directions, the overwhelming theme is that deception occurs with great frequency in almost every environment (DePaulo, 1996). Despite this empirical fact, people still believe they deceive and are deceived less than is actually the case. Thus there is a considerable difference between levels of actual deception and levels of perceived deception. It follows naturally then that people in a functioning society assume they are not being deceived, and further that they are not being deceitful. If a person is to believe that the majority of his or her interactions are laced with deceit, fostering any type of relationship or meaningful conversation would be incredibly difficult. Furthermore, the social stigma attached to deceit would add to the difficulty of believing that day to day interactions are littered with deceit—no one wants to believe they are a party to a stigmatized activity.

Having established that people believe they are less affected by deceit than they actually are, and given the plausible explanations, one might assume that the discrepancy between actuality and perception of deceit would remain constant across communicative mediums. To explore potential differences across mediums, we need a basic understanding of people’s beliefs about deception both FtF and in CMC. In general, people believe lying occurs less than it actually does FtF. This is because people are generally poor lie detectors, and rely on signals, such as nonverbal cues, which have been shown to be unreliable (Vrij, 2002). The logic here is that we fail to comprehend the of extent to which lies are told FtF because we are simply poor at identifying them. In CMC, however, we can not rely on faulty nonverbal cues to detect deception. Without the nonverbal cues we have come to depend on, we feel somewhat handicapped when it comes to detecting deception online. This results in a certain level of mistrust. For this reason, people predominantly believe that lies are much more prevalent on the internet than anywhere else.

Is lying online as prevalent as most people believe? In a 2006 study, Caspi and Gorski tried to answer this question. They hoped to compare beliefs and prevalence by asking people about their own experiences online. They discovered the “prevalence paradox, a phenomenon fundamental to this discussion. The paradox flowed from data that showed that 73% of all participants believed deception is widespread online, while only 29% reported that they themselves lie online. Furthermore, participants reported that few if any attempts were made online to deceit them. If people report that they do not lie and are not lied to very often online, then why do they believe that the internet is such an inherently deceitful environment?

While many studies have explored the prevalence of deception online, few have tried to understand people’s core beliefs about deception online. The aim of our discussion will be to explore people’ beliefs about deception with greater specificity—with regards to the magnitude and location of perceived deceit online. The broad focus of previous studies not only limits findings about the location of lies, but also about the severity of lies. The question of interest in the past has generally been concerned only with whether or not people believe deceitful activity is occurring. We want to discover the type of deceitful behavior people think is most commonly encompassed online, rather than whether or not deceit occurs online in general. Furthermore, we can acquire a more accurate understanding of why people are so mistrusting of the internet.

Another shortcoming of previous studies is that they focus on perceived online deception broadly, rather than differentiating between various modes of online communication (instant message, email, message boards etc.). In their study of deception, Hancock, Thom-Santelli, and Ritchie (2004) explored the frequency of deception across various communication technologies. Our study is similar, but we will be examining beliefs about deception across various modes of computer mediated communication rather than as a dichotomy between online and not online. This will allow us to explore how unique characteristics of online space affect the magnitude and propagation of lies online. For example, different online environments might be better suited for facilitating deceit of greater or lesser magnitude. This analysis might help us gain a better understanding of how the online environment fosters a sense of inflated deception.

Introduction- Kristen, Shane, Amy

Introduction

The possession of accurate information by the public is a necessary facet of our society. Accurate information dispels misunderstanding and provides people with true knowledge about their surroundings and those with whom they interact. False and misleading information, on the other hand, has historically been a source of tragedy and conflict. Wars between nations, within nations, and other cultural conflicts have been the direct result of widely disseminated false information. Many times this information is not simply the result of a misunderstanding or error, but an intentional act meant to deceive receivers of a message. This can be a particularly dangerous and malicious act on the part of the message sender. The exchange of truthful information is critical because people base decisions on the information they gain from the media, and their trust is affected by the spread of deceptive messages.

The consequences of spreading untrue information throughout history have created a need for mass distribution of accurate information. The mass media has assumed this role, and the population looks to various forms of media to provide accurate information about their surroundings. The necessity of media has been known for a long time, as is evident with the introduction of media as early as the nineteenth century. Important media developed over time include the telegraph, newspaper, radio, television, and the Internet, which includes blogs and email. The seekers of information have and continue to look to these sources to acquire information about the world. As these forms of media are so vital to society, receivers bestow a great amount of trust in them.

However, throughout the life of each medium there have been instances of intentional deceit made possible by the trust people have in each medium. These examples of deceit can be devastating to the opinions people hold of the trustworthiness and credibility of the sources they look to for information. In the early days of newspapers, the dramatization and yellow journalism that transpired during the Spanish-American War was extremely deceptive and led America into war with Spain. William Randolph Hearst and Joseph Pulitzer fabricated stories and used them to ignite support for war in America.

In 1937, Orson Welles delivered a radio adaptation of H.G. Wells’ classic novel The War of the Worlds. The messages were disseminated widely, as at the time television did not exist and radio was the main form of media for entertainment and rapid communication. Listeners believed that Welles’ story was real, as it sounded as if it were an actual radio news broadcast. There had never before been a publicized example of deceitful messages being spread over the radio, and as a result, countless listeners truly believed that Martians were invading the earth.

One of the first widely known and believed examples of deception on the Internet occurred in August of 1998 and came to be known as “Our First Time.” The deception involved two actors who purported to be teenagers preparing to lose their virginity on the Internet. This hoax was believed to be real and created a tremendous controversy.

Examples of deception abound in our society, and it is imperative to know when and how they occur. The deception resulting from The War of the Worlds broadcast was devastating, as there was widespread panic; some even committed suicide because they thought the world was coming to an end. By understanding the manner in which deceivers take advantage of the features of media, we can begin to comprehend the effects of deception on the public and the medium itself. The issue of deceitful information being disseminated by the media makes it critical to study the relationship between the variety of media disseminating information and the deception coming from these media. In addition, we will analyze the evolution of each medium based on the process of “institutionalization” described by communication theorist Stober. Using his process, we will identify the stages of evolution for each medium and make assessments regarding the changes in adaptation and exaptation. Based on the severity of the deception and the subsequent loss in trust for that medium by the American public, we will ascertain whether or not the evolutionary process creates a necessity for change in function or a completely new medium.

Wednesday, November 08, 2006

Assignment 9: Lit Review/Hypotheses

Literature Review & Hypotheses:

There are many areas which relate to these ideas of initial conversations and information flow. First we have the concept of asymmetric information – a situation where there is differing amounts/pieces of knowledge held by the conversation partners. In 1969, Ekman and Friesen wrote a paper primarily regarding potential cues to deception. They discussed a range of deceptive situations, one of which being an asymmetrical salience of deception, where the deceiver has more information (particularly that they are lying) while the receiver isn’t aware of this information. The authors predict that this type of asymmetric information will lead to less detection of deception. Hypothesis 1 – Asymmetric information will lead to an increase in deception.

A closely related concept is information control. Multiple studies have been done where participants have engaged in different tasks, and the participants who had the most control over the flow of information reported the most satisfaction at the end of the task (Bavelas, 1950; Leavitt, 1951). This result was shown not only in those who knowingly controlled the information, but even in those participants who were unaware of their central role (Leavitt, 1951). Hypothesis 2 – Those who control or hold more knowledge of the situation will be more satisfied at the completion of the conversation, and will rate their partner more favorably.

Once someone has this “extra” knowledge, there have been many studies looking at how and why people manage or conceal this information. Many people believe they have the right to conceal information from others (Imber-Black, 2003), and in fact most people would agree that lying is often done to save face for yourself or your partner. Withholding some knowledge or truth is how the world often functions, and doing anything differently could create dissonance, a feeling that you are actually doing something “wrong” by telling the truth (Schein, 2004). People have developed many different strategies for hiding or managing information. One woman studied how teens manage information they tell to their parents, and reported strategies including telling information only if the parent specifically asks, telling only partial information to give an acceptable answer but still leave out crucial information, or telling all information when they feel it is necessary for safety, punishment, or other reasons (Marshall, 2005). Hypothesis 3 – People with more knowledge will develop strategies to control this information, in order to avoid dissonance or other uncomfortable feelings.

Susan Fussel has done work regarding peoples’ biases and assumptions about their conversation partners. She has shown that these biases were found to unconsciously affect the way they constructed their verbal messages to their partner (1991, 1992). Hypothesis 4 – People with more knowledge about the situation/their partner will use different strategies for conveying information in their conversation than those less or no knowledge.

Finally, Robert Feldman conducted an experiment where he studied how a person’s self-presentational goals affected their amount of deception. He found that someone with a goal of self-presenting themselves (as likable or competent) told more lies than someone without a goal (2002). Hypothesis 5 – People with a goal of appearing likable will tell more lies, and will also achieve their self-presentation goal and appear more likable than those without a goal.

Overall, I believe these five hypotheses can be summarized in two. Hypothesis A – A person who has more knowledge (asymmetric information) will engage in more deceptive strategies, and will be more liked by their partner. Hypothesis B – A person with a self-presentational goal of being likable will engage in more deceptive strategies and will achieve their goal.

Assignment 9......again

The first known photograph was made by a French physicist named Joseph Nicéphore in 1816. Little did he know that his work, along with that of others’, would result in the creation of a communication tool that was to become as powerful and significant as the printing press. Photographs were considered incontestable proof of an event, experience, or state of being. But what if pictures weren’t what they seemed? What if they were capable of deceiving the untrained eye? By the end of the 20th century, digital imaging and processing and computer-based techniques made it possible to manipulate images in many ways, creating revolutionary changes in photography. The ability to manipulate pictures has major implications for the concept of truth.

A picture may be worth a thousand words, but what good are those words if they’re not true? Social psychological studies have shown that lying is a frequent part of everyday social interactions. In fact, research suggests that as many as one third of these typical daily interactions involve some form of deception (Hancock, Thom-Santelli, Ritchie, 2004). Deception can range anywhere from a slight exaggeration to an outright lie. According to Nyberg a lie can be classified as a subcategory of deception, though it has its own complexity as an adaptive behavior (Nyberg 1993). Photo-manipulation presents an interesting predicament because there are varying degrees of modification. It is possible to slightly alter a picture to better “capture the moment,” but it is also possible to merge photos in order to create documentation for situations that never took place. Nyberg further describes deception as the controlling of what is perceived, assumed, or understood by means of showing or hiding information (Nyberg, 1993). In his book, “The Varnished Truth,” Nyberg explains that an actual lie has four parts: a written or spoken statement, the speaker’s belief that the statement is false, the intention to lie, and the character or rights of the person being lied to. Some photos that have been manipulated may be deceiving in the sense that they don’t necessarily represent truth, but would this equate to a verbal lie? Blurring the background in a basketball action photo to create a sense of movement has different implications of truth than digitally removing blemishes from pictures of furniture meant to be sold on eBay, which may be more associated with a verbal lie.

Nyberg’s description of deception as “the shrewd and somber art of ‘showing and hiding’” brings up some interesting details. “The point is to present a situation in a way that will encourage a person to develop a confident, but mistaken hypothesis, which in some way serves our purpose” (Nyberg, 1993). The various ways of hiding and showing are relevant to deception in photographs because they involve many of the techniques used to manipulate a photo. Nyberg cites various ways to hide (disappear, disguise, distract) and show (mimic, counterfeit, misdirect). This framework was originally designed for verbal communication, but with the introduction of photo manipulation, could this framework also be adapted for visual communication? Nyberg’s framework will help us define when a photograph is being deceptive, or if the information in the photograph is just being managed improperly.

Manipulating a photograph has not always been easy to do. Altering pictures in the darkroom was always feasible, but with the advent of digital technology and resources afforded in an online setting, manipulation has become even easier. Not to mention, the extent of alteration has reached heights that were never thought possible. With the click of a mouse, something or someone can be cropped out of a picture, leaving the viewer to interpret whatever is left at face value. Deception can be established in photographs in several ways; a photograph can be manipulated either before, during, or after it is taken. For example, a photograph can be staged before it is taken, the camera angle can be altered while it is being taken, or Adobe Photoshop can be used to alter the picture after it’s been taken. It is also important to note that it is not only the photographer who can alter a photo, but the subject can be altered as well. That is, the photographer can change the camera settings or modify the physical picture or the person or object being photographed can be transformed. This ease of editability should, in theory, undermine the credibility of photographs; however, it is important to understand that a photograph’s meaning may vary with context. “A picture on a newspaper front page has more assertional force than it will hanging in an art gallery” (Martin, 1987).

The techniques used for photo-manipulation are infinite and photos can be staged or altered after a picture is taken. With digital photography, a photographer takes a picture, and the images are transferred to a computer. At this point, the picture can be edited using a manipulation software package, such as Adobe Photoshop. Photoshop allows for pictures to be cropped, blurred, lightened, darkened, stitched to another photo, and with some training it even allows for adding or erasing something or someone from a picture. Staging a picture may involve camera angle, zoom, or simply setting up the scene (not candid). Photographs were once considered incontestable proof, but maybe, “the camera has always been a liar, especially in the hands of a capable photographer” (Henshall, 1998). Why is it then that we still believe what we see in photographs?