PSY 450W Dr. Schuetze No one is ready 4 Levels Nominal Ordinal Ratio Interval Properties: Identity Classification data No ordering (makes no sense to state that M>F) Number assigned to each category is arbitrary (m/f = 0/1) 1
Properties: Identity and Magnitude Ordered but differences between values are not important. e.g., restaurant ratings Properties: Identity, magnitude, equal distance Ordered, constant scale No natural zero Difference makes sense but ratios do not (e.g., 30-20 =20-10, but 20 /10 is not twice as hot! 2
Properties: Properties: Identity, magnitude, equal distance, absolute/true zero E.g., height, weight, age, length Salary earned last year Quality of food: Good, Average, Poor Number of children Political Affiliation: Democrat, Republican, Independent Temperature in degrees Fahrenheit Marital status: Married, Single Reaction time Order people finished race Number correct on exam Score on intelligence test 3
Most statistical analyses have scale requirements Can no do means on ordinal or nominal data. Most analyses require at least interval scale. Will need to tell SPSS what scale of measurement each variable has Some statistical packages call both ratio and interval scales continuous Only certain operations can be performed on certain scales of measurement Can only examine if data are equal to some particular value or count the number of occurrences of each value E.g., gender can examine if gender of a person is m or f; can count the number of males in a sample. Can do everything we discussed with nominal data, plus Can exam if data point is less than or greater than another value Can rank ordinal data but cannot quantify differences between 2 ordinal values E.g., ratings of restaurants where 10=good, 1=poor. The difference between a 10 ranking and an 8 ranking can t be quantified. 4
Can quantify difference between 2 interval scales. E.g., temperature. 75 degrees versus 70 degrees. 5 degree difference has some meaning Does not make sense to say that 80 degrees is twice as hot as 40 degrees. Can take a ratio between 2 values. It is now meaningful to say that 24 pounds is twice as heavy as 12 pounds. 5
Degree to which a measurement is consistent and reproducible. Test-retest Reliability: compare scores of people who have been measured twice with same instrument. Reliability established when the two scores are very similar Reliability coefficient a correlation coefficient that ranges from 0.00 to 1.00 Highly similar scores are close to 1.00 6
Inter-item Reliability: extent to which different parts of a questionnaire or test assess the same variable. Sometimes you do have multiple measures, as in a 20-item personality measure Do the items correlate highly with one another? Interrater Reliability: level of agreement between measurements of different raters. A test of truth or accuracy Extent to which a procedure measures what it s intended to measure. 7
Agreement between a theoretical concept and a specific measuring device or procedure. Face validity Convergent validity Discriminant validity Criterion validity The degree to which a measurement device appears to accurately measure a variable 8
Do scores on the the measure relate to other measures in expected ways? Convergent validity: actual general agreement among ratings, gathered independently of one another, where measures should be theoretically related. Example: do people with high selfefficacy predict that they will perform better on a task? If so, this would be evidence for the construct validity of the measure. The measure of the variable is NOT related to other variables that it theoretically should not be related to. E.g., scores on the self-efficacy measure are not related to reaction time The degree to which a measurement device accurately predicts behavior on a criterion measure A paper-and-pencil measure of leadership ability predicts actual leadership behavior in a group 9
Content Validity: degree to which our measurements reflect the variable of interest. Face Validity: degree to which a manipulation or measurement technique is self-evident. Predictive Validity: degree to which a measuring instrument yields information allowing us to predict later behavior or performance. Concurrent Validity: degree to which scores on a measurement instrument correlate with another known standard for measuring the variable being studied. Reliability Versus Validity 10