Sharon Seidman Ph D CAS 301 Group Comparisons Sharon Seidman Ph D CAS 301 Week 7 Today s Agenda Review exam 1 Collect article 1 worksheets Learn more about group comparisons But First Thinking about Exam 1 Week 7 1 Sharon Seidman Ph D CAS 301 Research Process Develop hypothesis or question State specific testable prediction Collect data Variables Basis for research Individual elements of hypotheses predictions Must vary A variable Has levels Is a category Is a member of category Can change Cannot change Different Types of Variables ec bj u S t Re s pons e n atio Si t u Week 7 2 Sharon Seidman Ph D CAS 301 Variable Definitions Concept Valid Face Predictive Concurrent Convergent Discriminant Operation Reliable Internal Test retest Inter rater Scale Ratio Interval Ordinal Nominal Likert Scales Special type of ordinal scale Ordered rating categories Example 1 2 3 4 Not at all 5 A Lot Describing a Group Norms Central Tendencies Mean Median Mode Individual Differences Variability Range Standard Deviation Week 7 3 Sharon Seidman Ph D CAS 301 Scenario 1 Questions Variable identification X Axis Y Axis Type of graph Causality Variables Causality Types of variables Subject Situation Response Created vs pre existing variables Subject always pre existing Situation response may be Pre existing Created Scenario 2 The Administration for Children and Families decided to use the Peabody Picture Vocabulary Test 3rd Edition PPVT III to assess children s receptive English vocabulary They listed the following data about the test as the reason for their choice Cronbach s for the PPVT III 95 When children are tested twice the correlation between their scores is 94 The correlation between Head Start children s PPVT III score and their kindergarten knowledge test is 79 Week 7 4 Sharon Seidman Ph D CAS 301 Construct Validity Face validity subjective Criterion validity objective Predictive in future Concurrent between groups Convergent between measures Discriminant divergent between measures Construct Reliability Inter item internal consistency Split half Cronbach s alpha Test retest Inter rater Alternate forms Scenario 3 The US Department of Health and Human Services conducted a study of Head Start s impact on children s development in 2004 2005 They compared the cognitive and academic skills of children who did and did not enroll in Head Start programs and used multiple regression to determine how much Head Start participation was associated with improved child outcomes Outcome Measure Week 7 3 Year Olds 4 Year Olds Peabody Picture Vocabulary Test III 12 03 Woodcock Johnson III Letter Word Ident 24 22 Letter Naming Task 19 24 Color Naming Task 10 11 5 Sharon Seidman Ph D CAS 301 Change in Desired Result 1 Fall to Spring Scenario 4 W W WW WW W W W W W W W W WW W WW W W W W WW WW W W WW W W W WW W W WWW WW W W W WW WWW W W W WW W W W W W 2 00 1 50 1 00 WW W W W W W 0 50 W 0 00 0 00 1 00 W W W W W W W 2 00 3 00 Desired Result 1 in Fall Correlation Strength Direction 1 0 to 70 Strong Negative 69 to 30 Moderate Negative 29 to 00 Weak Negative 00 to 29 Weak Positive 30 to 69 Moderate Positive 70 to 1 0 Strong Positive Not the same as percentage Scales of Measurement Ratio Interval Ordinal Nominal Mean Median Mode Median Mode Mode Apply to quantitative measures Week 7 6 Sharon Seidman Ph D CAS 301 Some More Review Goals of Research Description Prediction Causation Descriptive Statistics Inferential Statistics Normal Skewed Week 7 7 Sharon Seidman Ph D CAS 301 Differences in Range Differences in Standard Deviation Kinds of Predictions Week 7 Linear Relations Correlations Group Comparisons Experiments 8 Sharon Seidman Ph D CAS 301 Linear Relation Class Size Fights 10 9 8 7 6 5 35 30 25 20 15 10 Correlations Linear Relation Fights Per Day 11 10 9 8 7 6 5 4 0 10 20 30 40 Class Size Week 7 9 Sharon Seidman Ph D CAS 301 Something New Group Comparisons Class Size 35 30 25 20 15 10 Fights 10 9 8 7 6 5 Between Group Comparison Group 1 Classes with 20 Kids Group 2 Classes with 10 Kids Question Is the amount of fighting significantly different Week 7 10 Sharon Seidman Ph D CAS 301 Describing Groups Review Norms mean median mode Variability range standard deviation New question how similar are two groups How many group members are similar How much do two graphs overlap Is difference between groups bigger than chance Number of Classes Frequency of Fighting in 20 Student Classes 4 5 6 7 8 9 10 Number of Fights Number of Classes Frequency of Fighting in 10 Student Classes 2 3 4 5 6 7 8 Number of Fights Week 7 11 Sharon Seidman Ph D CAS 301 Properties of a Normal Distribution 2 5 Area 2 2 2 5 13 6 34 1 34 1 13 6 2 2 Standard 3S 2S Deviation 1S Mean 1S 2S 3S Standard Assumptions in Group Comparisons Normative vs Atypical Group Members Middle 95 normal Extreme 5 atypical Very high scores 2 5 Very low scores 2 5 Group Comparison Process Compare normative group members Means middle 95 Frequency of Fighting in 20 Student Classes Number of Classes 95 4 5 6 7 8 9 10 Number of Fights Week 7 12 Sharon Seidman Ph D Comparing Groups 4 CAS 301 95 5 6 7 8 9 10 T statistic Numerical representation of difference between groups Difference between groups divided by variability within the groups Mean1 Mean2 SD Group Comparisons 2 Possibilities Groups are different Research Hypothesis Groups are the same Null Hypothesis Week 7 13 Sharon Seidman Ph D CAS 301 Sampling Frequency of Fighting in 20 Student Classes Number of Classes 95 4 5 6 7 8 9 10 Number of Fights Reality vs Conclusion Inferential statistics determination about probability of observed difference between groups Small probability accept research hypothesis Large probability accept null hypothesis Week 7 14 Sharon Seidman Ph D CAS 301 Reality Our Conclusion Accept Null Null is True Null is False Reject Null Reality Our Conclusion Accept Null Reject Null Null is Null is True False Type II Error Type I Error Probability Criterion 05 5 Probability of Type 1 Error Accepting research hypothesis when null is true Week 7 15 Sharon Seidman Ph D CAS 301 95 2S 99 3S Keys to comparison Norms M Variability Standard deviation SD Sample size Degrees of freedom Degrees of freedom Related to sample size Possible variability in sample How much freedom is there to get a different score Week 7 16 Sharon Seidman Ph D CAS 301 Writing Results Need all the information Degrees of Freedom t statistic Probability criterion alpha t df p 05 t 18 3 75 p 05 Effect Size Similar to variance in correlations Measures