About Validity and Reliability: Examples TABLE OF CONTENTS Types of Validity ................................................................................................................ 1 Face Validity................................................................................................................... 1 Content Validity.............................................................................................................. 1 Criterion Validity ............................................................................................................ 1 Concurrent Validity ........................................................................................................ 1 Predictive Validity .......................................................................................................... 1 Construct Validity........................................................................................................... 2 Types of Reliability............................................................................................................. 2 Test-Retest Reliability .................................................................................................... 2 Spearman-Brown and Cronbach’s Alpha ........................................................................... 2 Internal and External Validity............................................................................................. 3 History............................................................................................................................. 3 Maturation....................................................................................................................... 3 Testing............................................................................................................................. 3 Instrumentation ............................................................................................................... 3 Selection.......................................................................................................................... 3 Experimental Attrition .................................................................................................... 3 What is External Validity?.............................................................................................. 4 Reactive Effects of Testing......................................................................................... 4 Unrepresentative samples ........................................................................................... 4 Reactive Settings......................................................................................................... 4 Multiple-Treatment Interference................................................................................. 4 About Validity and Reliability: Examples Types of Validity Face Validity ► You create a test to measure whether people with the name Brandon are generally more intelligent than people with the name Dakota. If everyone shown the proposed test agrees that it seems valid, the face validity of the test has been established. Content Validity ► You wish to measure an individual's sleepiness. Noting that sleepiness is affected by a number of external and internal influences (e.g. amount of sleep the previous day, amount of activity during the day, etc.), a valid measure of sleepiness would account for these various factors. Criterion Validity ► You wish to measure risk of failure amongst first-year chemistry students, with a 25-item questionnaire. Criterion validity is established if your proposed measure correlates with a previously validated measure on risk of failure amongst first-year chemistry students. Concurrent Validity ► Your test which measures the average number of bananas consumed by an infant monkey correlates strongly with a validated test measuring the same information, when applied to a sample in the San Francisco Zoo. Predictive Validity ► The predictive validity of the MCAT for students applying to medical school is measured through the correlation of student MCAT scores to their undergraduate marks. If high-scorers on the MCAT perform better in their undergraduate courses than low-scorers, then the MCAT is presumed as a predictively valid measure of academic ability. 1 Construct Validity ► You wish to develop a new measure to assess euphoria. Construct validity is established by demonstrating that the new measure accurately predicts the myriad hypotheses that can be derived from a theory of euphoria. Types of Reliability Test-Retest Reliability ► The chart given below represents the data associated with two applications of the same test. r = 0.98 The r, Pearson's correlation coefficient, indicates a strong positive correlation. Spearman-Brown and Cronbach’s Alpha ► A researcher has developed a new 50-item scale to measure an individual's attentiveness. In evaluating the reliability of the measure, she decides to use the split-half reliability approach. After she collects data from an administration of the measure, she divides the scale equally, treating each half as its own scale in the application of correlational analysis. It is found that r = 0.73, which is too low to show adequate reliability. What should she do? She can apply the Spearman-Brown formula! 2 ► When the researcher plugs her values into the formula: r = 2 (0.73) = 1.46 = 0.84 1 + 0.73 1.73 Therefore, we see that after correcting with the Spearman-Brown formula, the researcher can now consider her measure of attentiveness to be reliable. Internal and External Validity History ► A 20-year study of Internet crime may be affected by a change in related laws, during the study period. Maturation ► A child may have a longer stride due to physical growth, as opposed to continual correction from an adult (independent variable). Testing ► A good example of this is multiple testing effects. Instrumentation ► One might measure a subject's propensity for laughter at the start of a study on happiness, while measuring a number of physical characteristics observed at the end of the same study. Selection ► In a study on innate athletic ability, two groups are selected – one composed entirely of NBA starters, the other made up of couch potatoes. Experimental Attrition ► In a test comprising two groups of 100 subjects each, the first group has 11 drop-outs, while the other has 50 drop-outs. 3 What is External Validity? Reactive Effects of Testing ► Measurement of a paper and pencil test will likely yield different results for those who have written them previously, as opposed to those who have not. Unrepresentative samples ► A group of German midgets is selected to measure average height tendencies in Europe. Reactive Settings ► A subject may feel depressed based on the peeling wallpaper and black ceiling of the room, while the experimenter may provide clues to their expectations of how the experiment will go based on their demeanour. Multiple-Treatment Interference ► Cue cards are presented to subjects in a particular order in a loud room, while these same cue cards are presented in the same order to another group of subjects in a quiet room. Unnoticed by the experimenter was that the first room had three mirrors on the walls within the subjects’ immediate view and peripheral vision, in addition to a few stains on the carpet, while the second room was clean, with one mirror out of view of the subjects. 4