Module 5: t-Test and SEM Intro Rosseni Din Muhammad Faisal Kamarul Zaman Nurainshah Abdul Mutalib Types Independent-samples • Compare mean scores of 2 different groups Paired-samples • Compare mean of the same group on 2 different occasions Only comparing 2 groups or 2 conditions More than that use variance Independant It needs • One categorical variable / independent variable • One continuous variable / dependant variable What the test will do • It will tell you whether there is a statistically significant difference in the mean scores for the 2 groups. Assumptions needed Paired One group but 2 different occasion / conditions • E.g. pre/post test Requirements: the same as independent • One categorical independent • One continuous, dependent variable It will tell you whether there is a statistically significant in the mean scores Data Analysis Using SPSS t-test t-test Used to test whether there is significant difference between the means of two groups, e.g.: • Male v female • Full-time v part-time t-test Typical hypotheses for t-test: a) There is no difference in affective commitment (affcomm) between male and female employees b) There is no difference in continuance commitment (concomm) between male and female employees c) There is no difference in normative commitment (norcomm) between male and female employees Performing T-test Analyze → Compare Means → Independent-Samples T-test Analyze Compare Means Independent-Samples T Test Performing T-test Select the variables to test (Test Variables), in this case: • affcomm • concomm • norcomm And bring the variables to the “Test Variables” box Test variables are selected and carried to the box on the right by pressing the arrow The test variables: affcomm, concomm, and norcomm Performing T-test Select the grouping variable, i.e. gender; bring it to the “grouping variable” box Click “Define Groups” Gender is the grouping variable Performing T-test Choose “Use specified values” Key in the codes for the variable “gender” as used in the “Value Labels”. In this case: 1 - Male 2 - Female Click “Continue”, then “OK” Specified values for gender are: 1 (Male) and 2 (Female) T-Test: SPSS Output Group Statistics affcomm concomm norcomm GENDER OF RESPONDENT MALE N Mean Std. Deviation Std. Error Mean 357 3.49720 .731988 .038741 FEMALE 315 3.38016 .696273 .039231 MALE 357 3.18838 .756794 .040054 FEMALE 315 3.15159 .666338 .037544 MALE 357 3.24090 .665938 .035245 FEMALE 315 3.27540 .647409 .036477 Mean scores for “Male” on the three test variables The mean scores for “Female” T-test: SPSS Output Independent Samples Test Levene's Test for Equality of Variances F affcomm Equal variances as sumed Sig. 1.048 .306 Equal variances not assumed concomm Equal variances as sumed 5.353 .021 Equal variances not assumed norcomm Equal variances as sumed .656 .418 Equal variances not assumed 2 t-tes t for Equality of Means t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% C onfidence Interval of the Difference Lower Upper 2.116 670 .035 .117040 .055308 .008442 .225638 2.123 666.213 .034 .117040 .055135 .008780 .225300 .665 670 .506 .036788 .055335 -.071863 .145440 .670 669.997 .503 .036788 .054899 -.071006 .144582 -.679 670 .497 -.034500 .050813 -.134272 .065271 -.680 663.726 .497 -.034500 .050723 -.134097 .065096 1 3 (1) Sig. is 0.306 (> 0.05) so there is no significant difference in the variances of the two groups (2) so the row “Equal variances assumed” will be used to read the sig. of t-test (3) Sig. level for t-test is 0.035 (<0.05) Therefore there is a significant difference in the levels of affective commitment (affcomm) between male and female employees. From the SPSS output, we are able to see that the means of the respective variables for the two groups are: • Affective commitment (affcomm) Male 3.49720 Female 3.38016 • Continuance commitment (concomm) Male 3.18838 Female 3.15159 • Normative commitment (norcomm) Male 3.24090 Female 3.27540 T-test: Interpretation For the variable “affcomm” • Levene’s Test for Equality of Variances shows that F (1.048) is not significant (0.306)* therefore the “Equal variances assumed” row will be used for the ttest. * This score (sig.) has to be 0.05 or less to be considered significant. T-test: Interpretation Under the “t-test for Equality of Means” look at “Sig. (2-tailed)” for “Equal variances assumed”. The score is 0.035 (which is less than 0.05), therefore there is a significant difference between the means of the two groups. T-test: Interpretation Independent Samples Test Levene's Test for Equality of Variances F affcomm Equal variances as sumed Sig. 1.048 .306 Equal variances not assumed concomm Equal variances as sumed 5.353 .021 Equal variances not assumed norcomm Equal variances as sumed .656 .418 Equal variances not assumed 2 1 t-tes t for Equality of Means t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% C onfidence Interval of the Difference Lower Upper 2.116 670 .035 .117040 .055308 .008442 .225638 2.123 666.213 .034 .117040 .055135 .008780 .225300 .665 670 .506 .036788 .055335 -.071863 .145440 .670 669.997 .503 .036788 .054899 -.071006 .144582 -.679 670 .497 -.034500 .050813 -.134272 .065271 -.680 663.726 .497 -.034500 .050723 -.134097 .065096 3 1. Sig. is 0.021 (<0.05), there is significant difference between the variances 2. The row “Equal variances not assumed” is used for interpreting the t-test 3. The relevant significant level for t-test is 0.503 (>0.05) Therefore, there is no significant difference between the two groups T-test: Interpretation For the variable “concomm” • Levene’s Test for Equality of Variances shows that F (5.353) is significant (0.021)* therefore the “Equal variances not assumed” row will be used for the ttest. * This score (sig.) is less than 0.05, so there is significant different in the variances of the two groups. T-test: Interpretation Under the “t-test for Equality of Means” look at “Sig. (2-tailed)” for “Equal variances not assumed”. The score is 0.503 (which is more than 0.05), therefore there is no significant difference between the means of the two groups. T-test: Interpretation Independent Samples Test Levene's Test for Equality of Variances F affcomm Equal variances as sumed Sig. 1.048 .306 Equal variances not assumed concomm Equal variances as sumed 5.353 .021 Equal variances not assumed norcomm Equal variances as sumed .656 Equal variances not assumed 2 .418 t-tes t for Equality of Means t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% C onfidence Interval of the Difference Lower Upper 2.116 670 .035 .117040 .055308 .008442 .225638 2.123 666.213 .034 .117040 .055135 .008780 .225300 .665 670 .506 .036788 .055335 -.071863 .145440 .670 669.997 .503 .036788 .054899 -.071006 .144582 -.679 670 .497 -.034500 .050813 -.134272 .065271 -.680 663.726 .497 -.034500 .050723 -.134097 .065096 1 1. The sig. is 0.418 (>0.05) so there is no significant difference between the variances 2. “Equal variances assumed” will be used to determine t-test 3. The Sig. of t-test is 0.497 (>0.05) Therefore there is no significant difference between the means of the two groups T-test: Interpretation For the variable “norcomm” • Levene’s Test for Equality of Variances shows that F (0.656) is not significant (0.418)* therefore the “Equal variances are assumed” row will be used for the ttest. * This score (sig.) is more than 0.05, so there is no significant different in the variances of the two groups. T-test: Interpretation Under the “t-test for Equality of Means” look at “Sig. (2-tailed)” for “Equal variances assumed”. The score is 0.497 (which is more than 0.05), therefore there is no significant difference between the means of the two groups. Hands-on exercise Use survey3ED.sav from www.allenandunwin.com/spss OR http://rosseni.wordpress.com/2011/ 07/15/spss-for-beginners/ Procedure for independent-sample t-test 1. Analyze > Compare means > independent samples t-test 2. Move the dependent (continuos) variable (e.g. total self-esteem) > Test Variable Box 3. Move the independent (categorical) variable (e.g. sex) > Grouping Variable Procedure for independent-sample t-test 4. Click define groups > type in the numbers used in the data set to code each group. In the curent data file, 1=males, 2=females; therefore, in the Group 1 box type 1; Group 2 box type 2; * if you cannot remember the codes used, right click on the variable name and then choose Variable Information from the pop-up box that appears. This will list the codes and labels 5. Click continue > ok Intro to SEM Structural Equation Modeling Purpose of the Study The study development of a model for meaningful e-Training by blending conventional and computer mediated communication to cater to learners with differentiated LS preferences. In this study we call it the Hybrid eTraining method. The Extension: Conceptual Framework of a Hybrid E-Training System (HiTs) Overall Research Framework Develop, implement and evaluate a hybrid system implementation that caters learners with differentiated learning style preferences, achieve meaningful learning Overall Research Framework Content Cooperativity Delivery Intentionality Service Hybrid e-Training Meaningful e-Training Construction Structure (HiTs) (MeT) Activity Outcome Authenticity Individual Learning Style Preference Kinesthetic (LSP) Tactual Auditory Visual Group n = 213 ICT trainers/trainees studying as postgraduate students/graduating fourth year students participated in the Technology for Thinking/Computer Education course in the year 2008 Overview of the Analytical Approach Prelim Analysis 1a 1b Formulate hypotheses; operationaliz e variables; examine distributiona l assumption Item analysis; reliability analysis principal Componen t analysis; Modeling Procedures 2a FF analysis 2b Validate measurement models: HiT model specification; estimation; fit assessment; path adequacy; SMC Validate other CFA models : MeT and LSP model specification; estimation; fit assessment; path adequacy; SMC 2c Test structural model: (1) HiT MeT and (2) LSP HiT model specifications; estimation; fit assessment; path adequacy; SMC Confirmatory Modeling Strategy 3 Test the fullfledge d model Hypothesized HiT in relation to MeT and LSP . . . e1 e2 e3 e4 e5 1 1 1 1 1 content e19 e17 1 1 coop 1 inten deliver struc serv MeT HiTs const 1 activ authen outcm 1 e18 LSP 1 visual audio kines tactil group indiv 1 1 1 1 1 1 e11 e12 e13 e14 e15 e16 1 1 1 1 1 e6 e7 e8 e9 e10 Reliability of the Instruments Meaningful e-Training (MeT) Measure α =.89 Hybrid E-Training (HiT) Measure α =.93 Learning Style Preference (LSP) α =.88 -.52 HiT Measurement Model Normed Chi-Square RMSEA .101 CFI .993 TLI .975 p .024 e5 content .39 e4 3.155 deliver e3 struc e2 serv e1 outcm .77 .89 .82 .95 .79 HiTs MeT Measurement Model LSP Measurement Model Normed Chi-Square 1.249 RMSEA .034 CFI .998 TLI .994 p .288 Normed Chi-Square 1.095 RMSEA .021 CFI .999 TLI .998 p .357 LSP e16 .52 .74 .85 MeT .83 coop e4 inten e5 const e6 activ e7 .95 authen .74 .66 .85 .62 audio visual kines group e12 e13 e14 e15 .52 tactil -1.01 e16 e8 .57 Structural Relationship of HiTMeT Normed Chi-Square 2.509 RMSEA .084 CFI .972 TLI .956 p .000 e1 e16 content .35 e2 deliver -.08 -.24 .41 e3 struc e4 serv e5 outcm .85 .86 .89 .80 .52 .81 .74 HiTs .45 MeT .84 .83 .96 coop e6 inten e7 const e8 activ e9 authen e10 -1.06 Structural Relationship of LSPHiTs Normed Chi-Square 2.603 RMSEA .087 CFI .964 TLI .946 p .000 e1 content .72 .46 e2 deliver .25 -.34 e3 struc e4 serv e5 outcm .63 .80 .91 .82 group HiTs .57 .52 .18 LSP .66 e16 tactil e14 kines e13 audio e12 visual e11 .84 .74 .93 e15 Coverage I. Statement of problem Objectives of the study Extension of the current hybrid model II. Method Setting; sample; Modeling procedure III. Results Measurement model Structural model Full-fledged model IV. Conclusion Overview of the Analytical Approach Prelim Analysis 1a 1b Formulate hypotheses; operationaliz e variables; examine distributiona l assumption Item analysis; reliability analysis principal Componen t analysis; Modeling Procedures 2a FF analysis 2b Validate measurement models: HiT model specification; estimation; fit assessment; path adequacy; SMC Validate other CFA models : MeT and LSP model specification; estimation; fit assessment; path adequacy; SMC 2c Test structural model: (1) HiT MeT and (2) LSP HiT model specifications; estimation; fit assessment; path adequacy; SMC Confirmatory Modeling Strategy 3 Test the fullfledge d model Adequacy of the full fledge Integrated Meaningful Hybrid E-Training (I-MeT) Model Normed Chi-Square 2.394 RMSEA .081 CFI .945 TLI .929 p .000 e15 e17 content .37 -.25 .43 e14 deliver e13 struc e12 serv e11 outcm e16 .79 .52 .80 .75 .84 .49 HiTs .87 .89 MeT .84 .84 .95 -.25 .15 LSP .75 .67 .84 .51 .62 audio visual kines tactil Group e10 e9 e8 e7 e6 .57 coop e1 inten e2 const e3 activ e4 authen e5 -.95