Psych 225 1st Edition Lecture 12 Outline of Last Lecture 1) VP article review2) Chapter 14: Questionnaires Outline of Current Lecture 1) Paper Questions2) Reliability and Validity 3) Pinkerton Article Current LecturePaper Clarification Questions*include mean, standard deviation, f, and p results -example: The negative words (M=, SD=), neutral (M=, SD=), and positive (M=, SD=) were all close to equal in length F (,) p=Should be no difference in length, concreteness, etc.Should be difference in pleasantness and arousal (arousal should be higher in positive and negative than in neutral, and positive and negative should be equal in arousal)-in parentheses mention scale they are rated on, ex. (MRC database) -used MRC first, then cross-referenced with valence databases MRC Psycholinguistic Database-control for potential confounding variables -check dimensions you want to consider Slide: Is RWA related to sided jurors tended to support in well publicized cases?-used 2 versions of RWA scaleThese notes represent a detailed interpretation of the professor’s lecture. GradeBuddy is best used as a supplement to your own notes, not as a substitute.-P’s that sided with prosecution tended to have higher authoritarianism scores-then did test that was less reliable: average of response to 3 item instead of 15 item; not significant effect-did not necessarily have higher authoritarianism -reliability and power are so intertwined that we can fail to see an effect when we have a less reliable measure Reliability (consistency) does not equal validity (accurate)-already have looked at internal validity (which looks at possible cofounding variables) (really was IV that caused differences in the DV)-construct validity: how accurately have we manipulated a variable/or measured a variable; captured the construct as it really exists Person who weighs 200 lb. steps on bathroom scale 10 times (twice/day for 5 days) 1) 75, 250, 175, 150, 140, 225, 50, 100, 240, 125: not reliable or valid2) 175, 175, 174, 176, 174, 175, 174, 176 175, 174: reliable but not valid3) 200, 199, 200, 200, 199, 201, 200, 200, 200, 200: reliable and validConstruct validity:-accuracy, extent to which measure of manipulation capture the intended construct -in our study: did negative words and neutral words differ on pleasantness but were the same on arousal Strategies for assessing construct validity of a measure-did theoretically predicted results occur?-are results consistent with past research?-attempt to establishconvergent validitydivergent/discriminant validity -these two above terms are pathway to construct validity -pattern we are hoping to demonstrateConvergent validity-look for r if scores on constructs should be related/correlated Divergent/Discriminant validity-looking for an absence of r (correlation) if scores on construct should not be relatedExample~new measure of anxiety: constructed and administered(experimentally these self-report measures take a long time to administer and we want short form) -also administered established measures of depression, optimism, intelligence, and anxiety If we truly captured anxiety with new measure, what is Y? Depression, optimism, intelligence and anxiety ~should see no correlation with anxiety and intelligence~as depression increases, anxiety often also increases~as optimism increase, anxiety decreases~anxiety and anxiety are very strongly correlated Cecil and Pinkerton (1998)-reliability and validity of a self-efficacy instrument for protective sexual behaviors-what is self-efficacy?Beliefs about how effectively you can affect your own life; opposite of learned helplessness; clear connection between what I do and the outcomes; have control in environment -what did C & P view as weaknesses of past research on the relation between self-efficacy and condom use?Reliability issues: many used a single-item measure Only measured it in one domain; thought issues if more complex; look at range of self-protective behaviors-domains of protective sexual behavior (self-efficacy subscales)?ability to refuse sexual intercourseability to question potential partners ability to use condoms Reliability of published scales (in study)-self-esteem, political conservatism, attitude toward computers, mastery, anxiety, social desirability some measures (attitude toward computers) were included to show divergent/discriminant validity -Cronbach’s alpha values on this scale two that are the lowest values, look pretty lowAverage responses to subscales and their reliability Ability to refuse sexual intercourseAbility to question potential partnersAbility to use condoms Self-efficacy subscale M (1-5) alpha # itemsRefuse Intercourse 3.48 .85 9Question Partner 4.20 .80 5Use Condom 4.44 .81 8If scores are sufficiently variable, it is hard to show effect; for this example means are very close to each other; expand the scale (make bigger than 1-5); you need the scores to vary to potentially covary with something else -often when you have more items, you have more reliability; but can achieve decent reliability with decent items-for Cronbach’s typically want .70 or higherC& P did not remove any items from subscales, why not?-you can test if you can delete items and have significant changes; they didn’t find they wouldWhat do r’s between subscale scores suggest?Self efficacy subscale Ability to refuse sexual intercourseAbility to q potential partnerRefuse ~ ~Question .36** ~Condom .39** .44*-makes some sense because they do have some differences Convergent validity? Evidence that self-efficacy subscale scores were related to behavior?-found negative correlation: more strongly they endorsed they felt they could refuse intercoursethe less partners they had in the last 3 months-no correlation with ability to q and use of condoms with number of partners in last 3 months-condom use in last episode: more able they felt they could do all three when they wore condoms the last time they had sex (positive correlation for all three) -condom use in last 3 months only correlated with ability to use condom Partners last 3 monthsCondom use last episodeCondom use past3 monthsCondom use in future refuse -.22* .18* .02 .06question -.03 .22* .15* .03Condom -.11 .26* .17* .13*Convergent and divergent validity?Convergent:-should expect correlations with: self-esteem, mastery –should be positiveConvergent:-should expect correlation with: social anxiety –should be negativeDivergent:-shouldn’t expect correlation with: