Individual Differences Week 2 notes
Individual Differences Week 2 notes PSY 345
Popular in Individual Differences
verified elite notetaker
Popular in Psychology (PSYC)
This 12 page Class Notes was uploaded by Cimmi Alvarez on Thursday September 1, 2016. The Class Notes belongs to PSY 345 at University of Texas at Austin taught by Elliot Tucker-Drob in Fall 2016. Since its upload, it has received 26 views. For similar materials see Individual Differences in Psychology (PSYC) at University of Texas at Austin.
Reviews for Individual Differences Week 2 notes
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 09/01/16
Measurement, Normal Distribution, Correlation, Regression, Reliability, Validity Measurement The quantification of the attributes of objects Quantifying- makes psychology scientific, Turing observations into quantitative date Qualitative- sitting in a room talking to people and making decisions. Applies to attributes of objects, rather than objects themselves Quantifying: extraversion, achievement motivation, intelligence ect. Not just a person but a characteristic Whereas physical attributes are directly observable, psychological attributes are Theoretical constructs Theoretical constructs Must be inferred from what are presumed to be their behavioral manifestations Trying to understand something about psychology. Can’t measure these psychological traits themselves. So observe to measure Examples Work ethic of an employee Sociability of a college students Measurement To measure psychological constructs, psychologists me asure observable behaviors that are presumed to reflect the consturcts Constructs- we construct them and are not able to be directly observed. Theoretical constructs-> Sample of observable behaviors People behave and they are behaving that way because of these constructs we cannot see. First step in taking the theoretical to the lab is by establishing an Operational definition. Operational Definition The process of creating rules that relate unobservable constructs to observable behaviors is called establishing an operational Definition EX: Sociability-> Number of Facebook Friends Measurement In order to compare individuals to one another in how much (variation) of a construct they possess, we need to be able to quantify the observable behaviors that we have measured Number of FB friends Number of questions answered true to a self-report test of sociability. Levels of Measurement (4) Nominal if numbers are used they are used as lables, and do not have any meaningful order Ordinal People can be ordered on the amount of the construct, but the difference between scores is not meaningful 2 The numbers still have an order but the difference between the orders are arbitrary. Ex: Swimming: rank 1, 2, 3 but the difference between 1 and 2 is 4 milliseconds ; difference between 2 and 3 is 12 milliseconds. Interval The difference between scores is meaningful Fahrenheit scale of measuring temperature Psychological test scores are often assumed to be measured on interval scales. But, Can’t compare the ratios because there is no meaningful 0 point. 100 Degrees F is not twice as hot as 50 degrees F Ratio Measurement Most meaningful. There is a meaningful zero point. This allows nonzero values to be expressed as ratios of one another. Weight, height, age Take the Differences between interval values and that difference is on a ratio scale Ecological Fallacy Compare groups of people what happens on a group scale doesn’t affect the individual individually. Trying to make a conclusion on individual people make sure that the country doesn’t mess it up. Standard Scores & the Normal Distribution When a number of different people take the same test, they will get different scores A distribution is characterized by what the scores look like when they are distributed on a graph 3 Want data to fall into this type of distribution Symmetrical graph Can take a non normal distribution and redo it to be normal. Normal distribution Most psychological tests are a normal distribution Most people score near the average Number of people with scores that differed from the average becomes few as scores get more extremes Holds for both high and lows low scores (symmetry) Normal Distribution can be characterized by two statistics Mean Average of the scores Add up all the scores and divide by the number of scores Variances Spread of the scores (how wide the distribution is) Average squared difference between each scores and the mean Divide by the number of scores Standard deviation Square root of the variance Standard scores and Normal Distribution Standard scores (z-scores) Important for putting scores on different tests on the same metric Makes scores directly comparable Much easier to interpret than raw scores 4 Tables with percentiles are usually in a z-score metric. How many standard deviations from the mean is the score? –z score tells you where you are relative to other people/scores. 1.96 standard deviations above- 97.5% higher than most people 1.96 Standard deviations below 2.5% higher than that percentage of people To convert raw scores into standard scores, subtract the mean and divide by the standard deviation Standard scores, or z-scores, have a mean of 0 and a standard deviation of 1 Correlation & Regression Indexing how two different variables are related to each other. Correlation and linear regression are ways we can quantify the relation between two variables Not about a definite but about an on average Indexing the tendency but not predicting it perfectly Correlation of 0 Flat slope Correlation of + Positive scores indicate positive scores Correlation of - Positive scores indicate negative scores Correlation of 1 or -1 are perfect and 0 is no correlation Closer to zero less correlation. Example Knowledge=482+4.5Xeducation + U 5 U is unknown. Factors that contribute that are unknown and are not controlled for R squared is what percentage is explained by the correlation What might a correlation imply? DOES IMPLY CAUSATION just doesn’t tell you what it is coming from but it does show how something is causing the two variables to be related to each other One variable causes another variable x->Y Y->X A third variable causes both variables Z->X AND Y Both variables represent the same underlying construct C->X AND Y The two variables represent two different constructs that are related Z-> C1 AND C2 AND C1-> X C2-> Y SOME CAUSATION IS HAPPENING SOMEWHERE No causation when the correlation happens completely by chance Correlation would not exist if you did the same experiment with different people CAUSED BY ERROR P values Confirm that a correlation is unlikely due to chance (low p value) then all of the causations hold. P=.05 is the cut off that is acceptable but you want to get it as low as possible Magnitude of a correlation is an index of how strongly two variables are related to each other 6 Correlation and Regression What is the difference between correlation and univariate linear regression? Correlation is simply a linear regression with standardized scores, rather than raw scores. Correlation is with z scores Regression is with raw scores Classical Test Theory Theoretical Construct -> Sample of Observable Behaviors <- Error Formula Y=T+E Y is score T is true score/ Theoretical Construct E is Error Error comes from Measurements are made with some level of imprecision Psychology magnifies the issue that attributes that we measure are only proxies for unobservable construct Only sample a finite set of behaviors during discrete period of time in hopes of generalizing to an infinite set of possible behaviors that potentially occur over a lifetime. Not observing all the time but only for a period of time Each item makes up a test may have idiosyncrasies that elicit responses based on participant characteristics unrelated to the construct of interferes. Contaminating sample with this. True scores + Error = Score Classical Test Theory Cant perfectly test anything 7 Errors- white noise in our measurement Reliability of test How well that test represents true scores Reliability Used to estimate proportion of variation in the test scores can be attributed to the construct that the test was designed to measure (as opposed to error) Classical Test Theory Internal consistency How much do the items within the test correlate wi th one another? How much does one question associate with another question in the same sect. What is the average correlation of all the different questions in the Openness category Called Cronbach’s Alpha Average correlation between each in the tests. Split Half Reliability Correlation between two halves (odd and even numbered items) of the test. Want it to be arbitrary and random Correlate them together to see how the two half’s then compare. IF High Correlation then good reliability Random stuff will not correlate. Take the correlation and subtract from one equals the percentage of error Test-Retest Reliability 8 Correlation between scores on two administrations of the test. The exact same test usually When might it be more appropriate to estimate reliability using internals consistency vs. Test retest? Could be practice effect by looking up answers you didn’t know. Trait vs. State distinction Traits Characteristics of an individual that are relatively stable ad enduring Intraindividual changes happen to traits More or less enduring changes that are construed as developmental by virtue of the nature of their antecedents, their consequences and their correlates Relatively slow, changes gradually Directional States momentary moods, feelings, or capabilities that are prone to fluctuation, and are often sensitive to physiological situational influences. Intraindividual variability happens to states Relatively short-term changes that are construed as more or less reversible and that occur more rapidly than Intraindividual changes. Test retest for traits Internal consistency for states Dilemma Theoretical constructs cannot be directly observed 9 We therefore need an operational definition delineating the correspondence between he theoretical construct Are constructs simply whatever we decide they are? Validity Extent to which a measurement assess the characteristic that it is supposed to assess Issues Does it measure what it is supposed to Four basic types 0. Face Does the test appear to measure what it is supposed to? It has meaning to the researcher but not to the participant Have a lot of experts vote in order to decide if it is an objective test (math) or objective (reading) Criterion-Oriented Validity 1 and 2 Whether or not the test scores predict some criterion of interest SAT scores – predicts how well you will do in college Companies making hiring decisions Insurance companies 1. Concurrent A form of criterion-oriented validity in which the test score and the criterion are determined at essentially the same time. 2. Predictive A form of criterion-oriented validity in which the test score issued to predict a later outcome Wait a period of time 10 Take the SAT wait for end freshmen year of college to see GPA 3. Content Validity Are the test items a representative sample of the universe of items that the investigator is interested? Test should have ALL features of intended characterizes, NONE of the irrelevant features Work sample tests: applicant for job p erforms tasks they would have to typically perform were they to be hired for the job. Should have all the tasks they would perform. 4. Construct Validity Process of better understanding exactly what the construct is that is being measured Occurs whenever no criterion or universe of content is accepted entirely adequate to define the characteristic of interest Undertaking process of trying to understand what you are Measuring-Theoretical Have theories about ow different constructs relate to one another, find our tests of those constructs do not relate as expected. Conclusion Tests do not represent the construct Theories about the constructs are incorrect Sometimes tests constructed basis of specific criterion, but later judged to have greater construct validity than the original criterion. 11 12
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'