Lecture Note
University
Middle Tennessee State UniversityCourse
Research MethodsPages
1
Academic year
2023
Talaija Hill
Views
0
p {margin: 0; padding: 0;} .ft00{font-size:21px;font-family:LatoLatin;color:#202122;} .ft01{font-size:60px;font-family:LatoLatin;color:#202122;} .ft02{font-size:28px;font-family:LatoLatin;color:#202122;} .ft03{font-size:24px;font-family:Arial;color:#000000;} .ft04{font-size:24px;font-family:ArialMT;color:#000000;} .ft05{font-size:24px;font-family:Arial;color:#660000;} .ft06{font-size:24px;font-family:Arial;color:#000000;} .ft07{font-size:24px;font-family:ArialMT;color:#0000ff;} .ft08{font-size:24px;font-family:ArialMT;color:#551a8b;} .ft09{font-size:37px;font-family:LatoLatin;color:#202122;} .ft010{font-size:24px;font-family:LatoLatin;color:#ffffff;} .ft011{font-size:21px;font-family:LatoLatin;color:#202122;} .ft012{font-size:28px;font-family:LatoLatin;color:#202122;} .ft013{font-size:21px;font-family:LatoLatin;color:#006fbf;} .ft014{font-size:28px;font-family:LatoLatin;color:#202122;} .ft015{font-size:10px;font-family:.SFUI;color:#808080;} .ft016{font-size:10px;font-family:.SFUI;color:#808080;} .ft017{font-size:60px;line-height:73px;font-family:LatoLatin;color:#202122;} .ft018{font-size:24px;line-height:30px;font-family:ArialMT;color:#000000;} .ft019{font-size:24px;line-height:30px;font-family:Arial;color:#000000;} .ft020{font-size:24px;line-height:29px;font-family:ArialMT;color:#000000;} Reflect in ePortfolio Download Print Langston, Research Methods, Notes 5 --Measurement (Ch. 5; Ch. 6) You have viewed this topic Last Visited Sep 22, 2023 8:54 AM I. Goals A. Introduction B. Numbers C. Psychometric Scaling D. Reliability and Validity II. Introduction The problem is this: Your experience is private. I can't look at you and know what is inside your mind. But, I want to measure that stuff. The more accuratelyI can measure it the better. Your textbook says that “Behavior is the road to subjective experience.” In other words, I can infer from your behavior what mustbe in your head. This unit is going to present you with just a taste of how this works in psychology measurement. III. Numbers First, we have to talk about what numbers are and what they can stand for. Depending on where the numbers come from (what sorts of measurementsproduce them), they can be more or less informative. There are four types of measurement scales: A. Nominal/Categorical The data are names or labels for categories. If they're presented as numbers, the numbers have no meaning outside of their function as labels. Forexample, if I collect people's soft-drink preference and assign Pepsi = 1, Coke = 2, Mountain Dew = 3, etc., I couldn't say that Pepsi is better than Cokebecause it has a lower number. All I can do is report the frequencies of people's preferences. Analysis: We can compare frequency distributions using a statistic called chi-square. For example, is the distribution of soda preference the same in thepsychology building as it is in the art building? We will see how to do this analysis in Notes 7. B. Ordinal These numbers add the order property. So, 1 is now ahead of 2. But, I don't know anything about the distance between the numbers. The space between 1and 2 could be huge, but the space between 2 and 3 could be small. If you think about finish positions in a race, you have this kind of data. The one whofinishes first is first, but we don't know by how much they beat the one who was second. Analysis: We have non-parametric versions for all of the common statistics that we use in psychology. These can be used with ordinal data. The problem isthat they are relatively weak. If possible, we would rather collect interval or ratio data. C. Interval These data add the equal intervals property (the gaps between all of the numbers on the scale are equal). So, 2 - 1 = 4 - 3. Temperature in degrees Celsiusis an example of this. The difference between 15 and 16 degrees is the same as the difference between 20 and 21 degrees. If you have this kind of data,you can meaningfully do math (like take an average). Analysis: Most of the statistics with which you are familiar require data that are at least interval (r , t , F ). D. Ratio These data add a true zero (the score of zero corresponds to none of the quantity being measured). So, if I said something had zero length, it has no length.Note that on a Celsius scale, you don't get this property. Zero is the freezing point of water, not “no heat.” Analysis: With ratio data you can make ratio statements. So, a line of two inches is twice as long as a line of one inch. You can check your understanding of the types of numbers by completing comprehension check 5.1. Comprehension check 5.1 E. Terms 1. Nominal/categorical2. Ordinal3. Interval4. Ratio Comprehension check 5.1 evaluation : My answers are below. a. I ask people how long they’ve been at MTSU. This would be ratio (it has order, equal intervals, and a true zero).b. I rank buildings based on how much people like them. This would be ordinal (ranking implies an order, but we don't know anything about the gaps).c. I score people’s major by 1 = Psychology, 2 = Biology, etc. This is nominal (there is no order).d. I have people report their class year. This is interval (order and equal intervals).e. I ask people for their age. This is ratio (it has order, equal intervals, and a true zero).f. I ask people to rank their gen ed classes from best to worst. This would be ordinal (ranking implies an order, but we don't know anything about the gaps). IV. Psychometric Scaling A. Introduction Psychometric scaling is a way to attach numbers to psychological constructs. The procedures here will help us solve two problems: 1. We can attach numbers to constructs that cannot directly be observed. 2. The data we get can be interval or ratio. B. Likert scales This procedure gives us a way to measure stuff that's hard to scale physically. For example, I have a theory that people can be extroverts or introverts. Iwant a measure that's at least interval (to allow statistics) that assesses whether you're an extrovert or an introvert. The steps are: 1. Generate a pool of items that (ideally) code behaviors related to the construct (introversion-extroversion). 2. Produce positive and negative versions of those items. a. At a party, I frequently talk to as many strangers as possible (positive) b. At a party, I usually only talk to the people I already know (negative) 3. Attach the scale in the top row of Table 5.1 to each item and have people rate the items using that scale. StronglyAgree Agree Neither Agree nor Disagree Disagree Strongly Disagree NegativeItems 1 2 3 4 5 PositiveItems 5 4 3 2 1 Table 5.1. Rating anchors and scoring for a Likert scale. 4. Score the positive items using the numbers in row 3 of Table 5.1 and score the negative items using the numbers in row 2. This will result in higher scoresgoing with behaviors more associated with extroversion. 5. Add up the scores to get the total for the scale. A Likert scale should produce interval numbers. There is some disagreement on that, but in psychology you will usually find people treating Likert data asinterval, so for the purposes of this class we will declare that to be the case. C. Terms 1. Psychometric scaling2. Likert scales V. Reliability and Validity A. Reliability Reliability assesses whether or not a measure is consistent. We're worried about consistency over time and whether or not the items in a scale all measurethe same underlying construct. If a measure is not reliable then there will be a lot of variability in the data and it will be hard to interpret what the numbersare telling us. For example (for reliability over time), I want to know if therapy is effective. I measure a person before and after therapy to see if they change.If the measure is unreliable (meaning I get a different score every time I measure it), how do I know if changes were due to therapy or random changes inthe measure? Assessing reliability: 1. Test-retest reliability : I correlate people's score on a test the first time they take it and their score the second time they take it. If the correlation is high, the test is reliable. If it's low, the test is not reliable. We usually look for this correlation to be .8 or above. 2. Internal reliability: One measure of this is coefficient alpha . This score essentially correlates all of the items with one another simultaneously. If it is above .8, then the items are all measuring the same thing and you have a good scale. When you look at a research report, you want to verify that their measures had high reliability. This should be in the materials section of the method sectionof the paper. If they don't report it, that can be troubling. If you are looking for a measure, you want one that has a demonstrated reliability. The only way toknow is to test it, so you have to find measures for which these data have already been collected and the scores showed high reliability. B. Validity Validity means the test measures what I want it to measure. For example, my extroversion item has face validity. It looks like a measure of extroversion.However, I usually want more than that. If my measure is not tapping into the right underlying construct, then the data are worthless. Assessing validity: Validity is much harder to assess. If you knew that your measure was getting at the right construct, then you wouldn't need to assess validity. If you don'tknow if you're measuring the right thing, then how will you be able to tell when you get it right? 1. Face validity : On the surface, it looks like a good way to measure what I'm interested in. A lot of the time I don't want high face validity because I don't want participants to know what response is expected of them. 2. Criterion validity : If we have a measure that we know measures the construct (a criterion, e.g., physiological arousal accompanies anxiety) we can correlate the score on the criterion with our new measure. If the correlation is high, then we can argue that the two are measuring the same thing. 3. Predictive validity : If your measure is valid, then it should reliably predict certain behaviors. For example, people who are anxious should get lower scores on measures of working memory. If your test predicts who gets lower scores on measures of working memory, then you are probably tapping into theright construct. 4. Convergent validity : If two tests are measuring the same thing, then the scores on those tests should correlate with one another. Note that there are many tests of validity, and a properly evaluated measure will look at more than one. C. Terms 1. Reliability2. Test-retest reliability3. Coefficient alpha4. Validity5. Face validity6. Criterion validity7. Predictive validity8. Convergent validity Try It for Yourself When you're selecting a measure for a research project, one of the most important steps is to evaluate its reliability and validity. You want measures thatmaximize both. Luckily, all you have to do to find out is look up articles that have done the work. For example, Stephenson et al.(2003; Stephenson,_Hoyle,_Palmgreen,_&_Slater_2003.pdf ) developed a measure of sensation seeking that was shorter than the original and could be used in large-scale research projects. They reported reliability for the new measures (e.g., coefficient alpha is reported in Table 2 on p. 283). For the 4-itemmeasure, the value was .66 (which they described as "very good" even though you would like at least a .80). As one way of measuring validity, they lookedat the relationships between sensation seeking and substance use (Table 3 on p. 283). Because higher substance use should go with higher sensationseeking, these correlations suggested that sensation seeking has predictive validity. They also discussed convergent validity because the new sensationseeking measures correlated well with the existing measures (they converged). Take a moment to review their article. For this assignment, find an article describing the reliability and validity of a personality measure of your choosing. Go to PsycInfo (on the front page of thelibrary site they have "Databases A-Z" in the "Quick Links" section; click that and then "P" and scroll down to PsycInfo; visit the library page here: https://library.mtsu.edu/home ). Type in a personality trait that you want to measure (e.g., "introversion") and look through the articles until you find one reporting a scale to measure your personality trait. (You can add search terms to narrow it down; if you put your personality trait in the first box and "validity"or "reliability" in the second box, that might get you there faster). When I put in "introversion" and "validity," I hit the jackpot on the third article. Zebb and Meyers (1993; Zebb_&_Meyers_1993.pdf ) reported the reliability and validity for an introversion-extroversion scale. The coefficient alpha (.82), test-retest reliability (.88) and description of validity measures were right in theabstract. It would be a bonus if the actual items were included (with the instructions for how to administer the test), but that is not part of this assignment. Once you find your article, look for at least one measure of reliability and one measure of validity. Report your personality trait, the reference for your article,and your reliability and validity data in the discussion forum Notes 5 Try It for Yourself comprehension check . Include enough detail so that we can understand what data you are reporting without having to look up the original article. Langston Research Methods Notes 5 Research Methods - PSY-3070… TH TaLaija Hill Course Home Content Examity Zoom Videos My Evaluations ePortfolio Assessments Communica ! on Help Table of Contents Langston, Research Methods, Notes 5 -- Measurement Langston, Research Methods, Notes 5 -- Measurement (Ch. 5; Ch. 6) Activity Details 10/3/23, 2 : 59 PM Page 1 of 1
Measurement
Please or to post comments