Fall 2009 PSY 202 Examples
From PsychWiki - A Collaborative Psychology Wiki
Dependent Variable
Definition: According to the textbook, a dependent variable is the variable that is measured.
Example: [1]
Application: In the study, the experimenters tested the extent of breast cancer caused by certain foods. As level of breast cancer was the variable of interest that was measured, it was the dependent variable.
Nominal Variable
Definition: According to the textbook, a nominal variable is mutually exclusive and exhaustive categories differing in some qualitative aspect.
Example: Shoe Pain
Application: In the study, the experimenters tested which shoes caused the most pain. The variable "type of shoe" is a nominal variable because it is a qualitative measurement involving different categories.
Ordinal Variable
Definition: According to the textbook, an ordinal variable or scale has the properties of a nominal scale, but in addition the observations may be ranked in order of magnitude (with nothing implied about the difference between adjacent steps on the scale).
Example: [http://www.facebook.com/photo.php?pid=4082887&id=18343191100 Newsweek BackStory about "Can you Cheat Death"
Application: The categories are listed in order, but the differences between each number are not equal.
Correlation does not equal causation
CIN: 111222333
Definition: According to the textbook, there are four possibilities when a correlation exists between two variables: (a), (b), (c), (d)
Example: See if you can spot the "correlation does not equal causation" in the opening monologue of Jimmy Fallon's "Late Night" Show.
Application: Did you find the "correlation does not equal causation". At minute 3:59, Jimmy Fallon says that a recent study from the University of New Hampshire showed that children who are spanked have lower IQs. The implicit implication is that spanking "causes" lower IQ, but is that true? There are four options: (a) spanking (variable X) could cause lower IQs (variable Y), (b) having a lower IQ could result in parents spanking the children more frequently, (c) a third variable like socio-economic status of the family could influence both spanking (because of increased violence associated with socio-economic status) and IQ (because of decreased IQ associated with diminished access to quality education in lower socio-economic status situations), or (d) there could be a complex interrelationship between spanking, parenting style, socio-economic status, IQ, cultural norms, and so forth. Notice, that "IF" the study was an experimental design in which parents were randomly assigned to one of two conditions in which their were forced to spank their child (condition 1) or not spank their child (condition 2), then it would be possible to prove causation. Given the ethical concerns with forcing parents to spank a child, the study was not experimental, but instead surveyed parents and children across time. The result of this study were publicized across every news outlet. See for yourself whether or not these high profile news outlets correctly identified that "correlation does not equal causation" -- Time Magazine, NPR, CBS video of their Morning Show, MSNBC, Los Angeles Times, San Francisco Chronicle.
Guideline 5
Definition: Guideline 5 is Watch out for confounding variables that can invalidate the conclusions of a study.
Example - Change Blindness
Application: In the video, one potential confound is that the two experimenters who switched places were very similar in appearance. It is possible that an attentive subject would not be able to tell the differences between two similar looking white young males both wearing lightly colored long-sleeve shirts.
Guideline 3
Definition: Guideline 3 is Examine the sampling method to decide whether it is likely to produce a representative sample.
Example Change Blindness
Application: In the video, the experimenters only used psychology subject pool subjects. The problem with that sampling method is that the participants may not have been motivated to be aware of their surroundings, and instead only interested in getting the credit for taking part in the experimenter. For example, if you look at the video, the students are concentrating on where to go to take part in the study and what to do next after the instruction by the confederate. In other words, they are self-focused!
Guidelines 5
Definition: Guideline 5 is Watch out for confounding variables that can invalidate the conclusions of a study.
Example: Stephen Colbert's Fancy Feast
Application: In the example, Stephen Colbert discusses an experiment from the American Association of Wine Economists that tested "Can People Distinguish Pate from Dog food?" In the study, the participants were asked to taste test each type of food, and then identify which was the dog food and which was the pate. The results of the study were that 83% of the people could not identify dog food from pate. Were there confounds in the study? Stephen Colbert identified a confound when he said that "These people were either eating really great dog food or really cheap pate". If the dog food was higher quality and/or the pate was lower quality, then participants may be judging whether the food was dog food or pate by the quality of the food. Thus, "quality" is a confound because it is something other than the intended constructs that are impacting the dependent variable. He identified another confound when he asked what type of people would agree to this study after learning they had to eat dog food! That confound is a self-selection bias in that maybe only a particular type of person would submit to this study, thus biasing the results. A final type of confound was shown when Stephen Colbert ate the cat food in the video in a fancy drinking glass. While we don't know how the researchers presented the food to the participants in the study, another way to potentially confound the results is by presenting each type of food in a different way.
Guideline 2
Definition: Guidelines 2 is Consider the source, particularly with regard to whether the researcher may be biased.
Example: Ghostbusters clip
Application: The researcher was biasing the results in two ways: (1) purposefully allowing the beautiful girl to believe she had ESP by telling her that her answers were correct, even though they were not, (2) purposefully lying to the male subjects by telling him that his correct answers were incorrect. What is the source of the bias? At the end of the clip you see the experimenter talking to the beautiful girl in a way that suggests he wants a date.
#2: 'Lie' on purpose
Definition: Given intent to purposefully lie, strategies include... ◘ Data manipulation = Collecting data but changing data points to achieve the results you desire.
Example: Ghostbusters clip
Application: Is it possible the researcher had ulterior motives for faking the results of the study? Notice the various ways in which he engaged in data manipulation... In the video, in the middle of the experiment, and in the middle of data collection, the researcher is purposefully skewing the results so that he can impress the beautiful girl while also removing the other subject from the room. If he tried to publish the results, the data/publication would be invalid.
#3: 'Lie' by misrepresentation
Definition: Damaging misrepresentation occurs by selective reporting: ♦ Presenting only part of the data ♦ Discarding unfavorable data ♦ Conducting multiple studies but only reporting the one(s) that confirm the hypothesis
Example: Newsweek article and again
Application: The last page of each Newsweek article (called "Backstory" is a 1-page graphical display of information/data. In the two links above, the "Backstory" article for Oct 13 is reproduced, showing data about two time periods in Russian history (then, now) and some categories (population, divorces, crime) with the caption "Was Russia Better off Red". While not a true research study, the story is arguing a hypothesis (russia was better off before democracy) and is showing data to test the hypothesis (data about certain categories before and after democracy). How is the data "lying by misrepresentation", first, it is only presenting part of the data, 11 categories of interest. What about other categories? Did the authors purposefully include only the categories that fit their hypothesis? Did they omit the categories that disprove their hypothesis? Its likely that if they had included a category such as "Freedom of Speech" or "Number of people murdered by the State" or "Perceptions of Oppression", the information would have disconfirmed their hypothesis.
Guideline 4
Definition: Look for problems in defining or measuring the variables of interest, which can make it difficult to interpret any reported results.
Example: Newsweek article and again
Application: The two links reproduce a Newsweek article showing data about two time periods in Russian history (then, now) and some categories (population, divorces, crime) with the caption "Was Russia Better off Red". While not a true research study, the story is arguing a hypothesis (russia was better off before democracy) and is showing data to test the hypothesis (data about certain categories before and after democracy). Notice the many difficulties defining or measuring each variable, thus reducing the validity of the information. For example, for the category "People Diagnosed with Disease", how did they measure that variable? and are there more diseases now than decades ago? and are there better hospitals/doctors better able to identify disease now versus in the past? As another example, for the category "Economically Active Population", what do the researchers mean by "economically active", how do you measure that variable? how do you define that variable? As a final example, the data are split into two categories - Then, Now. Are those time periods? What time periods? What years? What is the dividing line between the two? With different measurements of "Then" and "Now, you would undoubtedly get different data for each time period.
Guideline #7
Definition: Guideline 7 is Check that results are fairly represented in graphics and concluding statements.
Example: Newsweek article about Who Owns the US?
Application: The last page of each Newsweek article (called "Backstory" is a 1-page graphical display of information/data. In link posted above, you see a graphical display of various countries who own US national debt, and the level of debt owned by each country. To represent the "pie", the article show a Buffalo superimposed with dotted lines "carving up" the Buffalo into pieces - the US debt. The problem is that the data is not accurately depicted proportional to the size of each piece. In other words, there are minor perceptual distortions that render the graphical information inaccurate, as compared to the corresponding numbers (levels of debt).
Guideline #7
Definition: Guideline 7 is Check that results are fairly represented in graphics and concluding statements.
Example: Jimmy Fallon, watch from minute 11:40, to minute 12:20
Application: What is the problem with the graph? There is no title, caption, vertical scale tick marks, or vertical scale labeling to indicate the quantity or frequency of each category.
Guideline #2
Definition: Guidelines 2 is Consider the source, particularly with regard to whether the researcher may be biased.
Example: Wall Street Journal Article, Washington Post Article
Application: The two news articles report how hackers posted on the internet thousands of emails and documents from prominent Global Warming scientists that raise ethical questions about the science and scientists. For example, the emails contain statements about trying to exclude contrary views from being published, to prevent data being available to dissenting scientists, and to punish dissenters. In regard to Guideline #2, there is clear bias in favor of Global Warming and against anyone who disagrees with that point of view. To be fair, however, the emails are private emails illegal hacked from their computers, and we may not fully understand the context of the emails.
#3: 'Lie' by misrepresentation
Definition: Damaging misrepresentation occurs by selective reporting: ♦ Presenting only part of the data ♦ Discarding unfavorable data ♦ Conducting multiple studies but only reporting the one(s) that confirm the hypothesis
Example: Wall Street Journal Article, Washington Post Article
Application: The two news articles report how hackers posted on the internet thousands of emails and documents from prominent Global Warming scientists that raise ethical questions about the science and scientists. For example, the emails contain statements about trying to exclude contrary views from being published, to prevent data being available to dissenting scientists, and to punish dissenters. In regard to "lying by misrepresentation", there is clear evidence in the emails that the scientists are trying to purposefully discard unfavorable data by preventing any data from being published that is contrary to their point of view. To be fair, however, the emails are private emails illegal hacked from their computers, and we may not fully understand the context of the emails.
4/5 Dentists recommend...
Problems:
- sampling
- sample size
- how worded
- bias
◄ Back to Descriptive Statistics PSY 202 Fall 2009 page
