RMAS Lecture 2 - Three Level Experiments Identify the Variable . An IV is a variable that's value is not affected by something else . An IV can have any number of levels . A DV is that entity that is affected by the IV Quasi Experiment Compare the groups Self-Selection Participant variables - · Natural categories (age, gender, handedness) · Scores (personality types, clinical symptoms) Questionnaire-Based Selection Psychometrics · The measurement of unobservable psychological constructs typically using questionnaires . Key principles 1) validity of the measures constructed 2) reliability of the measures 3) standardised measures . It is a common skills and used extensively in the real world and research Reliability . Extent to which findings of studies can be replicated - not just by chance · If your study reproduces findings of previous research, findings are reliable · Types 1) inter-rater reliability (different people produce the same results) 2) test- retest reliability (the same result occurs over time) 3) inter-method reliability (the same result occurs if different methods are used) 4) internal reliability (all questions correlate with each other) Improving Reliability · Item analysis - examining reliability coefficients (Cronbach's alpha statistic - good reliability means Cronbach's alpha of at least .70) · Examining item-total correlations - correlations between participant's responses on individual items and participant's overall score on the scale Validity · Do findings show what they claim to show? · Testing validity involves assessing whether study findings have been correctly interpreted
· Types 1) face validity (does a test appear to measure what it should? 2) construct validity (does a test measure what it should?) 3) content validity (does a test measure all of what it should?) 4) predictive validity (does the test predict DVs later on in time?) 5) convergent validity (does a test correlate with similar tests?) 6) discriminant validity (does a test not correlate with different tests?) 7) criterion validity (does a test correlate with the DV as it should) 8) concurrent validity (does a test correlate with other tests measured at the same time?) Appropriate Questions · Methods for eliciting responses - rating scales, tick boxes, binary choices · Frustration at restricted response · Misinterpretation of questions meaning (ambiguous or unclear wording) . Lack of realism (little reason for agreement) · Likert scales (Likert, 1932) a set of statements with fixed points (anchors) . Problems with Likert scales (ambiguity of mid-point of scale and its anchors and relativity of responses and range-restricted data - how much difference is there between 1 and 2) · Semantic differential scale (Osgood et al., 1957) assesses meanings, bipolar adjectives for anchors containing fixed number of response points · Problems with semantic differential scale (ambiguity of mid-point of scale and unlabelled points mean you can't quantify response variability) · Response bias - individuals agree more than disagree with questionnaire items - deal with this problem (reverse-wording and scoring, having a mixture of positively and negatively worded items) · Social desirability bias (not giving true responses to appear good