Chapter 3Chapter 5 & 6Chapter 7Exam 2 Review ***Know the listed concepts. Be able to define, apply, and provide examples of them. ***Chapter 3 Converging operations Scales of measurement (nominal, ordinal, interval, ratio)o Nominal- does not meet any of the 3 properties Labels or groups Gender, ethnicity, zodiac sign, religion o Ordinal- rank or order (the different variables have some meaning to one another) Magnitude Military rank, educational level, race or contest resulto Interval- magnitude and equal intervals Magnitude of difference Temperature, sea level, year, IQo Ratio- magnitude, equal intervals and absolute zero Test zero, money, height, weight, distance, age Adds to absolute zeroo If you have a choice, you would want to go with Interval and Ratio Examples:- Year- Interval- has magnitude, equal intervals, NO absolute zero- Military Rank- Ordinal- give meaningful comparison- Ethnicity- Nominal- no magnitude, no interval, no absolute zero- Money- Ratio- there is an absolute zero Measurement properties (magnitude, equal intervals, absolute zero)o Scales differ on properties of magnitude, equal intervals and absolute zero Magnitude- does this scale imply a size or quantity difference?- Does one value on the scale represent more, less, or equal amounts value on the same scale?- Examples:o Weight- Ann is 140 lbs, Paul is 185lbs. Results: Paul is bigger than Anno Sex- 1=women, 2=men. Given arbitrary values- do not imply magnitude Equal Intervals- does the difference between 2 values on the scale have thesame meaning on any part of the scale. - Examples:o Weight- Ann lost 10 lbs (140-130), Paul lost 10 lbs (185-175). Results: difference means the same (both lost 10 lbs)o Depression level- 0 is more than 1and 2 is more than 3 Difference between 0 and 1 is not the same as 2 and3 Absolute zero- represents the absence of this behavior in the characteristic.Does the property being measured have a zero quantity - Zero equals absence of the property- Examples:o Weight- 0 lbs= no weighto Intelligence measures- 0 does not necessarily mean no intelligence Observed score = True score + Measurement erroro Want to measure to assess behavioral variability accurately True score falls under systematic variance Measurement error falls under error variance Sources of measurement error o Participant factors Transient (State) Factors: mood, health, anxiety, fatigue Stable Attributes: intelligence, ability, motivationo Situational factors: animal handling, room temperature, lightingo Measure characteristics: ambiguous questions, too long, induce fearo Mistakes: experimenter sneezes or distracted, computer glitcho Example: True intelligence score (hypothetical since there is no such thing as a true intelligence score)= 125 Observed intelligence score= 112 Error= 13- Could be internal or external erroro Test taker could be sick, depressed, stressed outo The room could be too coldo The test could be poorly designedo The experimenter could have made a scoring error Systematic vs. random error: types of erroro Systematic error Consistent error Less of a concerno Random error Unpredictable error Major concern ( want to eliminate this type of error even more than systematic error) Reliability= true score variance/ total variance in a set of scoreo How much of that variance are we capturing with that measureo Consistency or dependability of a measuring techniqueo Assessing reliability of a measure Correlation coefficient (r) - The bigger the number, the more correlated that number iso The closer it is to 1, the better!o The closer to 0, the worse..o .00-1.00o Sufficient reliability (>/) .70 Test-retest reliabilityo Measure participants on two occasionso Depending on the variable of interest, the time we test is more important Correlate time 1 and time 2 scores with anothero Only used on stable traits and characteristics Examples: IQ, attitudes, personality Interitem reliability (item-total, split-half, Cronbach’s alpha)o Subjected to questionnaireso Assess the degree of consistency among the items on a scale Want each item to measure the same general construct (correlation?) Internal consistencyo Techniques (assess Interitem reliability) Item-total correlation- Short questionnaire, ideal approach- Correlation between single item and sum of all other itemso Ex: #2 vs. all other numbers- For each question, we’re looking for a .30 correlation or better, if itdoesn’t qualify, we drop that item with a low correlationo r >.30 Split-half reliability- For a longer questionnaire- Divide scale items into two setso Top: 1-6 vs. bottom 7-12o Odd vs. Even- Correlate total scores- Problems- small number of items, variation due to split- r > .70 Cronbach’s alpha coefficient- Average of all split-half reliabilities- Most commonly used Interitem reliability technique- r >.70 Interrater reliabilityo Consistency among researchers who observe and record participants’ behavior Observe and record responses in the same way, with the same judgment- Percentage of time raters agreed- Correlate raters’ recordings Observation measures, content analysis Ways to increase reliabilityo Will need clear and precise operational definitionso Raters should practice together Validityo Face- a measure appears to measure what it’s supposed to Involves the judgment of the researcher or of research participants Never enough evidence, but it’s a start- Ex: a national store chain paid $1.3 million to job applicants who sued the company because they were required to take a test that contained bizarre personal items such as “I would like to be a florist” and “ Evil spirits possess me sometimes.” These items werefrom commonly used, well-validated psychological measures (MMPI & CPI) but they lacked face value Qualifications- Just because something has face validity, doesn’t necessarily mean it’s actually valid- Many measures that lack face validity are in fact, valid- May design things purposely to lack face validity and thereby conceal the purpose of the test although it is, in fact, valido Construct (convergent and discriminant) Assessment: look at the correlation between multiple measures AKA- converging operations Do we get the same results with these different measures?- Convergent validity- correlations