WIXLIB

Data Analysis Using SPSSBy: Akmal Aini Othman

The key to GOOD descriptive research isknowing exactly what you want to measure andselecting a survey method in which everyrespondent is willing to cooperate and capable ofgiving you complete and accurate informationefficiently–Joe Ottaviani-

Uncertainty InfluencesThe Type Of ResearchCAUSAL LORATORYSource: Zikmund, 2009

ProblemdiscoveryProblem Discoveryand DefinitionSelection ofresearchtechniqueSamplingSelection ofexploratory gDataProcessingandAnalysisProblem definition(statement ofresearch objectives)ExperimentLaboratoryConclusionsand ReportSurveyFieldInterviewCollection ofdata(fieldwork)Editing andcodingdataDataprocessingSelection ofbasic researchmethodResearch aryData StudyInterpretationoffindingsReportSource: Zikmund, 2009

Thesis Contents Chap 1 - IntroductionChap 2 - Literature ReviewChap 3 – MethodologyChap 4 – Findings & DiscussionChap 5 – Conclusion andRecommendation

Thesis Contents Introduction – why & what this research Literature Review – who have done this research &how, what results, what shortcomings Research Framework & Data Collection – why thisframework, hypotheses; measurements, sample, howdata can be collected Data Collection & Analysis – what methods mostappropriate, findings Conclusion – have u achieved what you set out to do?

Thesis Contents Chap 1 – Introduction Background of the studyProblem StatementResearch QuestionResearch ObjectiveHypothesisSignificance of the studyLimitation

Thesis Contents Chap 4 – Findings and Discussion Descriptive Analysis Test of Goodness of Data – e.g Normality &Multicollinearity Factor Analysis Reliability and Validity Test Inferential Analysis / Hypothesis Testing

Data Preparation ProcessPrepare Preliminary Plan of Data AnalysisCheck QuestionnaireEditCodeTranscribeClean DataStatistically Adjust the DataSource: Malhotra, 2012Select Data Analysis Strategy

Questionnaire CheckingA questionnaire returned from the field may beunacceptable for several reasons. Parts of the questionnaire may beincomplete. The pattern of responses may indicate thatthe respondent did not understand or followthe instructions. One or more pages are missing. The questionnaire is received after thepreestablished cutoff date. The questionnaire is answered by someonewho does not qualify for participation.

EditingTreatment of Unsatisfactory Results Returning to the Field – The questionnaires withunsatisfactory responses may be returned to the field,where the interviewers recontact the respondents.Assigning Missing Values – If returning thequestionnaires to the field is not feasible, the editor mayassign missing values to unsatisfactory responses.Discarding Unsatisfactory Respondents – In thisapproach, the respondents with unsatisfactory responsesare simply discarded.

CodingData coding Coding the variables Coding the response/items for each variable Eg. Variable for gender sex Coding item 1 male, 2 female The numerical scale can be coded by using theactual number circled by the respondents(question 6 to 21) Random checks should be conducted to ensuredata are coded correctly

Table 12.1Coding of Serakan Co. Questionnaires1.Age (years)2. Education3. Job level4. Sex[1] Under 25[1] High school[1] Manager[1] M[2] 25-35[2] Some college[2] Supervisor[2] F[3] 36-45[3] Bachelor’s degree[3] Clerk5. Work shift[4] 46-55[4] Master’s degree[4] Secretary[1] First[5] Over 55[5] Doctoral degree[5] Technician[2] Second[6] Other (specify)[6] Other (specify)[3] Third5a.Employment Status[1] Part time[2] Full timeHere are some questions that ask you to tell us how you experience your work life in general.Please circle the appropriate number on the scales below.To what extent would you agree with the following statements, on a scale of 1 to 7, 1 denoting very low agreementdenoting very high agreement?6. The major happiness of my life comes from my job.1234567. Time at work flies by quickly.12345678. I live, eat and breathe my job.1234569. My work is fascinating.12345610. My work gives me a sense of accomplishment.12345611. My supervisor praises good work.12345612. The opportunities for advancement are very good here.12345613. My coworkers are very stimulating.12345614. People can live comfortably with their pay in this organization.12345615. I get a lot of cooperation at the workplace.12345616. My supervisor is not very capable.12345617. Most things in life are more important than work.12345618. Working here is a drag.12345619. The promotion policies here are very unfair.12345620. My pay is barely adequate to take care of my expenses.12345621. My work is not the most important part of my life.123456and 7777777777777777

Data TranscriptionFig. 14.4CATI/CAPIKeypunching viaCRT TerminalRaw DataOpticalRecognitionDigitalTech.Bar Code &OtherTechnologiesVerification: CorrectKeypunching ErrorsComputerMemoryDisksTranscribed DataOtherStorage

Data Cleaning Consistency ChecksConsistency checks identify data that are outof range, logically inconsistent, or have extremevalues. Computer packages like SPSS, SAS, EXCELand MINITAB can be programmed to identifyout-of-range values for each variable andprint out the respondent code, variable code,variable name, record number, columnnumber, and out-of-range value.Extreme values should be closely examined.

Data Cleaning Treatment of MissingResponses Substitute a Neutral Value – A neutral value, typically themean response to the variable, is substituted for the missingresponses. Substitute an Imputed Response – The respondents'pattern of responses to other questions are used to imputeor calculate a suitable response to the missing questions. In casewise deletion, cases, or respondents, with anymissing responses are discarded from the analysis. In pairwise deletion, instead of discarding all cases withany missing values, the researcher uses only the cases orrespondents with complete responses for each calculation.

Basic TermsLevels ofMeasurement NominalOrdinalIntervalRatio Key Terms VariableDimensionItemDefinition DictionaryOperationalVariables IndependentDependentModeratingMediating

Research Framework5 itemsManagement(Independent)5 items3 itemsAdvancement(Independent)4 itemsJob ry(Independent)Gender(Moderating)4 itemsWorkload(Independent)

ScaleNominalOrdinalIntervalRatioNumbersAssignedto RunnersFinish783Rank Orderof WinnersPerformanceRating on a0 to 10 ScaleTime toFinish, .19.615.214.113.4Source: Malhotra, 2007

What is Statistics – process of making senseof data Descriptive Stat – describe the basic featuresof data using tables, graphs, summary stats Inferential Stat – generalising from samples topopulations performing estimations, hypothesistests, determining relationships and makingpredictions

Descriptive Statistics Norminal data – frequencies, %, crosstabulation, mode, pie chart, bar chart Ordinal data - frequencies, %, cross tabulation,mode, median, pie chart, bar chart Interval & Ratio data – mean, variance, stddeviation, skewness, kurtosis, index number,histogram, box plot, stem and leaf plot

Inferential Statistics Statistical Techniques: Exploring differences between groups Exploring relationshipParametric – Data must be interval and thedistribution must be normalNonparametric – Data is categorical(norminal/ordinal) or interval butdistribution is not normal

Data analysis Basic objectives: Getting a feel for the data Testing the goodness of data Testing the hypotheses Feel for the data Checking for the central tendency and the dispersion If there is less variability, the questions could be notproperly worded Check for similar response for every questions Remember, if there is no variability in the data, then novariance can be explained

Data analysis It is always prudent to obtain: Frequency distributions for the demographic variables The mean, standard deviation, range and variance onthe other dependent and independent variables An inter-correlation matrix of the variables, regardlesswhether hypotheses are related to the these analyses. Ifthe correlation between two variables is high, say over.75, we should wonder whether they are really twodifferent concepts or we are measuring the sameconcepts.

Data analysis Testing goodness of data Reliability Cronbach’s alpha. The closer Cronbach’s alpha is to 1,the higher the internal consistency reliability Split-half reliability coefficient Stability measures include: Parallel from reliability Test-retest reliability Validity Criterion-related validity Convergent validity Discriminant validity

Choosing the TestDepends on: Data – Norminal or Interval/Ratio Data Samples – one/two/k-samples Purpose – Describing, Comparing twostatistics or Looking at relationship

A Classification of Univariate TechniquesUnivariate TechniquesNon-numeric DataMetric DataOne Sample* t test* Z testTwo or MoreSamplesIndependent* Two- Group test* Z test* One-WayANOVARelated* Pairedt testOne Sample*****Two or endent*****Chi-SquareMann-WhitneyMedianK-SK-W e: Malhotra, 2012

Univariate Analysis Univariate analysis is the simplest form of analyzingdata. “Uni” means “one”, so in other words your datahas only one variable. It doesn't deal with causes orrelationships (unlike regression) and it's major purposeis to describe; it takes data, summarizes that data andfinds patterns in the data. It explores each variable in a data set, separately. Itlooks at the range of values, as well as the centraltendency of the values. It describes the pattern ofresponse to the variable. It describes each variable on itsown.www.csulb.edu/./696uni.htm

A Classification of Multivariate TechniquesMultivariate TechniquesDependenceTechniqueOne DependentVariable* Cross-Tabulation* Analysis of Varianceand Covariance* Multiple Regression* 2-GroupDiscriminant/Logit* Conjoint AnalysisMore Than leInterdependence* Multivariate Analysis * Factor Analysisof Variance* Confirmatory* Canonical CorrelationFactor Analysis* Multiple DiscriminantAnalysis* Structural EquationModelingand Path AnalysisInterobjectSimilarity* Cluster Analysis* MultidimensionalScalingSource: Malhotra, 2012

Multivariate Analysis Multivariate Data Analysis refers to any statisticaltechnique used to analyze data that arises from morethan one variable. This essentially models reality whereeach situation, product, or decision involves more than asingle variable.

Steps Involved in Hypothesis TestingFormulate H0 and H1Select Appropriate TestChoose Level of SignificanceCollect Data and Calculate Test StatisticDetermine ProbabilityAssociated with TestStatistic (p value)Compare with Level ofSignificance, Determine Critical Value ofTest Statistic TSCRDetermine if TSCAL fallsinto (Non) RejectionRegionReject or Do not Reject H0Draw Research Conclusion

Hypothesis Testing – Hnull & Halternative A null hypothesis is a statement of the statusquo, one of no difference or no effect. If the nullhypothesis is not rejected, no changes will bemade. An alternative hypothesis is one in which somedifference or effect is expected. Accepting thealternative hypothesis will lead to changes inopinions or actions. The null hypothesis refers to a specified value ofthe population parameter (e.g., m , s , p ), not asample statistic (e.g., X ).

Hypothesis Testing – Hnull & Halternative A null hypothesis may be rejected, but itcan never be accepted based on a singletest. In classical hypothesis testing, there isno way to determine whether the nullhypothesis is true. The null hypothesis is formulated in such away that its rejection leads to the acceptanceof the desired conclusion. The alternativehypothesis represents the conclusion forwhich evidence is sought.H0: p 0. 40H1: p 0. 40

Hypothesis Testing – Hnull &Halternative The test of the null hypothesis is a one-tailedtest, because the alternative hypothesis isexpressed directionally. If that is not the case,then a two-tailed test would be required, andthe hypotheses would be expressed as:H 0: p 0 .4 0H1: p 0.40

One-Tailed & Two-Tailed Test

One-Tailed & Two-Tailed Test