180 likes | 263 Views
Things I’ve Tried to Measure. Remo Ostini Healthy Communities Research Centre University of Queensland Australia. Overview. Community Based Health Organisation Activities Moral concepts Using Polytomous Item Response Theory models. MEASURED PROGRESS – Jan. 2009: 2 of 18.
E N D
Things I’ve Tried to Measure Remo Ostini Healthy Communities Research Centre University of Queensland Australia
Overview • Community Based Health Organisation Activities • Moral concepts • Using Polytomous Item Response Theory models MEASURED PROGRESS – Jan. 2009: 2 of 18
Community Based Health Organisations • “Self-help” organisations for people with chronic illnesses • Diabetes Australia • Arthritis Queensland • Prospective survey, two time periods, 4 months apart • PS1 N=323; PS2 N=306 respondents surveyed again MEASURED PROGRESS – Jan. 2009: 3 of 18
CBO Activities Asked nine questions at both surveys: • Member of CBO? • Estimate times phoned CBO • Estimate times read newsletter or printed info • Estimate times in seminar, workshop, info session • Estimate times talked with other CBO members • Estimate times attend CBO support group or social outing • Estimate times volunteer with CBO • Estimate times counselling, exercise, discount products • Estimate times used CBO info raise others' awareness • Dimensionality? MEASURED PROGRESS – Jan. 2009: 4 of 18
CBO Activities • Factor analysis suggested 1 or two factors • Theory supported 2-factor structure • Factor I: High intensity activities • Factor II: Low intensity activities • Qualitatively different, stage-like relationship MEASURED PROGRESS – Jan. 2009: 5 of 18
Rotated Factor Matrix Factor 1: High Intensity Activities; Factor 2: Low Intensity Activities MEASURED PROGRESS – Jan. 2009: 6 of 18
IRT Results – Scaled together • Model • Partial Credit model • Results • All items together (assume one dimension); both time periods (9 + 10 items) -- 5 items with poor fit (< 0.001) MEASURED PROGRESS – Jan. 2009: 7 of 18
IRT Results – Scaled separately • Results • Low Intensity Activity items both time periods - 2 items with poor fit • Low Intensity Activity items PS1 - 2 items with poor fit • No High Intensity Activity items with poor fit (at both times or PS1) • Really just want q for further analyses MEASURED PROGRESS – Jan. 2009: 8 of 18
Comparing Organisations • Mean q across activities and time periods • Diabetes Aus and Arthritis QLD significantly different on low intensity activities MEASURED PROGRESS – Jan. 2009: 9 of 18
Predicting PAM PAM = Patient Activation Measure; a self-report measure of patient confidence in managing their own health MEASURED PROGRESS – Jan. 2009: 10 of 18
Moral Conceptualization • Initial question: What is morality? • Lots of answers. Who knows? Everyone! • So ask them… MEASURED PROGRESS – Jan. 2009: 11 of 18
Moral Conceptualization - Background • 20 Interviews – talkative people • 1269 statements – some (~ half) redundant • 4 × 150 item questionnaires - MCS • 5-point Likert scale (1 – Nothing like what you think; 5 – Exactly what you think) MEASURED PROGRESS – Jan. 2009: 12 of 18
Moral Conceptualization - Background • Example statements: • Caring about others is the dominant feature of a good person • Everyone is born good • Evil does exist and people can do it • The boundaries between right and wrong are defined by social custom • What aspects of morality does the questionnaires cover? MEASURED PROGRESS – Jan. 2009: 13 of 18
Moral Conceptualization Structural Analysis • Data Screening – bound to be noise • Four linear analyses: • Principal Components Analysis • Factor Analysis • Multidimensional Scaling • Cluster Analysis • One non-linear analysis • Mokken scaling analysis – nonparametric IRT with h test for dimensionality assessment • 10-14 factors MEASURED PROGRESS – Jan. 2009: 14 of 18
Moral Conceptualization Structural Analysis • Types of Factors • Life definition • Relativism • Responsibility • Socially defined morality • Conscience • Tolerance MEASURED PROGRESS – Jan. 2009: 15 of 18
Moral Conceptualization Item Analysis • Item-total correlations < 0.20 • Serious skew (some items still useful?) IRT for Item analysis • Partial Credit Model • Reversed boundaries • Evidence of a problem? • Item model fit • p <0.001 • Too soon for information? MEASURED PROGRESS – Jan. 2009: 16 of 18
Moral Conceptualization Construct Validity • Concurrent validity • Personality (Agreeableness & Conscientiousness related to Relativism) • Internal State Awareness – multiple MCS • Liberalism & Conservatism – some MCS • Multiple Social Values – multiple MCS • Not Emotional Intelligence; Empathy; Guilt; Religion • Predictive validity • Predicting moral judgements? MEASURED PROGRESS – Jan. 2009: 17 of 18
Moral Conceptualization Construct Validity • Next? • Reduced number of questionnaires with best items from first four • More sophisticated test development, scoring and validation… MEASURED PROGRESS – Jan. 2009: 18 of 18