Unformatted text preview: Now that you have Now that you have data, what do you do with it?
Comm 300 On the On the menu today… How we how we interpret what we’ve found in research Why i
nterpretation matters Why interpretation matters Already discussed how numbers carry authority Also, instrumental and symbolic value How we interpret data can therefore have:
Policy; Ethical; and Researchrelated consequences What we look for in doing What we look for in doing analyses We look for patterns—do they reflect what our RQs and hypotheses posited we would, based on theory and on prior work?
What presentation of the results best reflects the reality of our findings? How can we compare our findings to others’ results? What do we do with data once we have it? Say you’re conducing a study of online behavior Say you’re conducing a study of online behavior and one of your respondents said that they spend 9 hours online yesterday—is that a lot, or a little?
To figure out what is “average” in a population, we need more information. Then we can compare individuals to the overall sample Measures of Central Tendency: Measures of Central Tendency
Measures that describe the center point of a distribution of quantitative data. Mean: a.k.a., “average” Median: a.k.a., “middle point” Mode: a.k.a., most frequently occurring Your participants in your study of online behaviors report the following # of hours spent online yesterday: 1, 3, 3, 5, 5, 5, 7, 7, 9 What is the mean, median, and mode of these numbers?
Mean: Add them up and divide by the total number of responses: 1+3+3+5+5+5+7+7+9 = 45/9 = 5 Median: Find number that is in the middle: 5 Note: Odd total number means there will be one number in the middle Mode: Which number appears most often = 5 When do you report them?
When do you report them?
You will report means most often
~ E.g., to report the average number of hours participants spent online yesterday
Medians (Md or Mdn) are reported when you have some extremely high or extremely low values that might affect your mean
~ ex: Income
Modes (or frequencies) are generally reported with nominal data Let’s go back to our earlier Let’s go back to our earlier question:
One of your respondents said that they spend 9 hours online yesterday—is that a lot, or a little?
How do figure out where an individual score falls, relative to other scores in the sample? Measures of Dispersion
Measures of Dispersion Measures that report how far a set of scores are spread around the center point of the data and across the distribution. #1: Range: distance between highest and lowest scores in a distribution Ex: 6, 2, 1, 3, 5, 9, 8, 9, 11 Highest number? Lowest number? Distance between them? Measures of Dispersion Measures of Dispersion #2: Standard Deviation (SD): It’s a complicated concept, but the easiest way to think of it is as the "mean of the mean.”
Standard deviation measures how much on average individual scores vary (or deviate) from the mean score for the group. In other words, the standard deviation is useful because it shows us how many subjects in the group scorewithin a certain range of variation from the average for the entire group. Knowing this can help us make decisions about how to interpret our results. Say these were your midterm Say these were your midterm scores:
# students What would each of these distributions tell me about the exam I gave you? # students 0
100% What could each of these standard What could each of these standard deviations tell us?
Low SD= Low dispersion High SD= High dispersion How do we understand the How do we understand the significance of what we’ve found? Statistical significance as a way to understand Statistical significance as a way to understand if our results are due to chance (null hypothesis); or actually due to our manipulation of the IV (research hypothesis) Statistical significance vs. social significance Correlation vs. causation Real world example: Do we have ESP? The situation:
Dr. Bem (Cornel Univ.) tested a classic memory study in reverse.
The original study: participants study 48 words and then divide half those words into categories. They are more likely to remember the subset than those they didn’t practice. Dr. Bem’s version:
Dr. Bem’s version:
Gave 100 college students a memory test before they did the categorizing — and found they were significantly more likely to remember words that they practiced later. Conclusion: “The results show that practicing a set of words after the recall test does, in fact, reach back in time to facilitate the recall of those words.” Proof #2?
Proof #2? Subjects chose one of two curtains on a computer screen—one hid a photograph, the other had just a blank screen.
After students picked a curtain, a software program randomly posted a picture behind of the curtains.
Finding: students beat chance, by 53 percent to 50 percent—when the photos being posted were erotic (did not do better than chance on other photo types). Bem’s conclusion: “Subjects could sense the Real world example: Do we have ESP? All four reviewers said the paper met the journal’s standards, even though “there was no mechanism by which we could understand the results.” Many experts say that is precisely the problem. Claims that defy almost every law of science are by definition extraordinary and thus require extraordinary evidence. Neglecting to account for this — as conventional social science analyses do — makes many findings look far more significant than they really What does this reveal about the review What does this reveal about the review process? Dr. Bem’s reputation may have played a role in the paper’s acceptance. Peer review is anonymous—but all four reviewers were social psychologists, and would have known whose work they were checking. None were topflight statisticians. One journal editor said, “The problem was that this paper was treated like any other, and it wasn’t.” The null hypothesis here is ESP does not exist. Refusing to give that hypothesis weight makes no sense, experts say; if ESP exists, why aren’t people getting rich off it? What efforts have been made at What efforts have been made at selfcorrection? Unusual move by the journal—the article was published along with other articles that claim to rebut the findings. So far, at least three efforts to replicate the experiments have failed. But more are inthe works, Dr. Bem said, adding, “I have received hundreds of requests for the materials” to conduct studies. Alternative explanation?
“He’s got a great sense of humor,” said Dr. Hyman, of Univ. Oregon. “I wouldn’t rule out that this is an elaborate joke.” Summary: Why interpretation Summary
matters Interpretations can vary, and are result from decisions made during the research procedure and analysis
Honest assessments of a study’s limitations Need for replication Findings have consequences Instrumental and symbolic value Vested interests in outcomes? Consequences for policy, ethics, public perception, and future research ...
View Full Document
- Fall '11
- Real world example, Dr. Bem