SPSS LEVEL 2

SPSS Level 2

By Hafiza Abas

26 Mac 2015Advanced Informatics SchoolUniversiti Teknologi Malaysia

Descriptive and Inferential Statistics

• DS – used to summarize, organize and simplify data (using mean and SD)

• IS – make generalization of the population from the sample (using estimation and hypothesis testing)

• Why make generalization of the sample?

Because Our RQ is about the population, the issues lies in the population.

Variable

• A variable is any measured characteristic or attribute that differs for different subjects.

• Variables can be quantitative or qualitative.

• Quantitative variables are measured on an ordinal, interval, or ratio scale

• Qualitative variables are measured on a nominal scale

Measurement Scale

• Measurement is the assignment of numbers to objects or events in a systematic fashion. Four levels of measurement scales are commonly distinguished:

nominal, ordinal, interval, and ratio.

Hypothesis Testing

• IS requires hypothesis

• Hypothesis about the population not the sample

• State the hypothesis first

• Go out and collect dataThe statement of H very important to guide us to search the info/data/evidence to support or to refuse the H.

Statement of the Hypothesis

• H0 (Null H)State that such as nothing happen

That is no relationship between X and Y

No effective

No significant

No difference

No influence

But you can also state Mean IQ for registered voter in Malaysia is-> mean=110

Statement of the Hypothesis

• H1 (alternate H)

• Opposite of the H0

• That is a relationship between X and Y

• Relate to ALPHA- type one error (type 2 error is beta)

Alpha = 5% -----tolerance level

• Measure a construct – 5 or more items (don’t go beyond 10)

• Always prepare good items straight away

• Icon to switch code to label

SPSS

• This is wrong – WHY?

• No ( …. )

• Transform -> Compute

Variables

Output

Write Objective

• Try to avoid – to find out

Objective

• To investigate

• To examine

Objective To examine if there is a significant difference among teachers of different GENDER on the first source of stress (Relationships with colleagues/student s’ parents)

Research Question

Is there a significant difference among teachers of different GENDER on the first source of stress (Relationships with colleagues/student s’ parents)?

Hypothesis

Null There is NO significant difference between among teachers of different GENDER on the first source of stress (Relationships with colleagues/students’ parents)

Alternative There is A significant difference between among teachers of different GENDER on the first source of stress (Relationships with colleagues/students’ parents)

Statistical Tool

Independent sample t-test

Decision Fail to REJECT the null hypothesis (p=0.108, t=1.615)

Note: If p-value<0.05 REJECT NULLIf p-value>00=.05 FAIL TO REJECT NULL

Summary The difference among teachers of different GENDER is not statistically significant.

T-test

• 3 types of t-test

– One sample t-test

– Paired Sample t-test

– Independent t-test

Why use independent sample t-test? (in this case)

Because independent group

T test allow to compare two independent group/level within the variable at a time

ANOVA

• Allow you to compare more than two variables (eg. malay, chinese, indian and others)

Analyze

• Analyze-> Compare Means -> independent sample t-test

• How SPSS do the comparison between male and female – as the first source of stress?

SPSS will compare the male and female.

T-TEST is about comparing the MEAN.

• Why ?? Appear?

Inform the computer which two groups are comparing…

The result

• This is the descriptive table

• female more influence from the first source of stress compare to male.

• The dif between mean is about 0.2

• Is that the dif 0.2 large enough ?

• Refer to P Value

• SIG = significant level = P Value

• Whether to reject or not refer to Sig (2 tailed)

• Which to refer – top or bottom?

• Whether top or bottom refer the

• How to read P Value? Sig ->0.565 -- if it is larger than 0.5 read from the top.

• So read from the top - > 0.108

• P Value larger – the decision = fail to reject the NULL hypothesis. NULL is still true.

• Our NULL is still good

• So : There is NO significant difference between among teachers of different GENDER on the first source of stress (Relationships with colleagues/students’ parents)

Objective To examine if there is a significant difference among teachers of different GENDER on the SECONDsource of stress (WORKLOAD)

Research Question

Is there a significant difference among teachers of different GENDER on the first source of stress (WORKLOAD?

Hypothesis

Null There is NO significant difference between among teachers of different GENDER on the SECOND source of stress (WORKLOAD)

Alternative There is NO significant difference between among teachers of different GENDER on the SECOND source of stress (WORKLOAD)

Statistical Tool

Independent sample t-test

Decision Fail to REJECT the null hypothesis (p=0.175, t=-1.360)

Note: If p-value<0.05 REJECT NULLIf p-value>00=.05 FAIL TO REJECT NULL

Summary The difference among teachers of different GENDER is not statistically significant on the second SOURCE OF STRESS

RUN THE ANALYSIS

• Analyze-> Compare Means -> independent sample t-test

• Sig= 0.255 , larger than 0.5 .. Read from the top -> Sig (2 tailed)

So the decision

FAIL to REJECT

Conclusion:

The difference among teachers of different GENDER is not statistically significant on the second SOURCE OF STRESS

Run the Analysis

One Sample T-Test

• Mean = 20

• Test value= 25

• Sig 2 tailed=0.00

• Mean different -4.5

Conclusion= REJECT NULL

* To compare the mean to a certain value.

Paired Samples T-Test

• Paired – example: pre test n post test.

• Collect the data twice from the same source.

• Data – taken in pairs from the same post

• Repeated measures – compare more than 2.

• T Test – can compare two groups at a time

Objective To examine if there is a significant difference between the first and fourth source of stress.

Research Question

Is there a significant difference between the first and the fourth source of stress?

Hypothesis

Null There is NO significant different on how the first and the fourth source influence tress among teachers

Alternative There is a significant different on how the first and the fourth source influence tress among teachers

Statistical Tool

Paired Sample / dependent sample t-test

Decision REJECT the null hypothesis (p=0.000, t=-18.179)

Note: If p-value<0.05 REJECT NULLIf p-value>00=.05 FAIL TO REJECT NULL

Summary The fourth source (students’ issues) has greater influence compared to the first source (relationships) on stress among teachers.

Run the Analysis

This is for correlation

Mean difference If p-value<0.05REJECT NULL

Repeated Measures

• Measures are being repeated.. Taken in jan, feb, mac, april ..and want to compare all the data simultaneously ..then use repeated measures.

• Reject the NULL

• Conclusion:

Workload significantly higher in giving stress to the teacher

ANOVA

• To compare more than 2 groups simultaneously

• Also about comparing the means

Objective To investigate if there are significant differences among teachers of different ethnicities (MALAY, CHINESE, INDIAN, OTHERS)

Research Question

Are there any significant differences among teachers of different ethnicities (M,C, I, O) on the first source of stress?

Hypothesis

Null There are NO significant differences among teachers of different ethnicities M,C, I, O) on the first source of stress.

Alternative There is at least ONE significant differences among teachers of differentethnicities M,C, I, O) on the first source of stress.

Statistical Tool Analysis of Variance (One-way ANOVA)

Decision FAIL to reject the NULL hypothesis (p=0.960 >0.05, F=0.099)

Note: If p-value<0.05 REJECT NULLIf p-value>00=.05 FAIL TO REJECT NULL

Summary There are NO significant differences among teachers of different ethnicities M,C, I, O) on the first source of stress.

Run the analysis

• Why One-Way ANOVA?

Because we use only ONE independent variable

• Why Two-way ANOVA?

• No way ANOVA – close your computer

• Do not define the groups.

• Larger than 0.05 ..fail to reject.

• The significant differences are not available.

If p-value<0.05 REJECT NULLThe overall P. No need to proceed anymore

• (Biggest # 157 / smallest # 4)<1.5

Good generalization of the result… if the ratio is larger than 1.5 – so the generalization is very limited

There are NO significant differences among teachers of different ethnicities M,C, I, O) on the first source of stress.

Inflated Alpha

• The main problem that designers of post hoc tests try to deal with is alpha -inflation

• It refers to the increase in the nominal alpha level when the number of statistical tests conducted on a given data set is increased.

Run the Analysis

• Open working example 2.

• In this study, perceived support received from the principals was operationalized in terms of THREE dimensions, namely Support for Career Progression, Support for General Welfare and Support for Instructional Processes.

Objective To investigate if there are significant differences among teachers of different age groups on job satisfaction

Research Question

Are there any significant differences among teachers of different age groups on job satisfaction?

Hypothesis

Null There are NO any significant differences among teachers of different age groups on job satisfaction

Alternative There are at least one significant differences among teachers of different age groups on job satisfaction

Statistical Tool Analysis of Variance (One-way ANOVA)

Decision FAIL to reject the NULL hypothesis (p=0.018 , F=2.622)

Note: If p-value<0.05 REJECT NULLIf p-value>00=.05 FAIL TO REJECT NULL

Summary Of all the differences, significant difference on job satisfaction is only found between age group 20-24 (mean=4.25, SD=0.58) and 45-49 (mean=3.29, SD=0.81)

• The result is to reject the NULL.

• There is at least one significant –somewhere.

• Run again.

• Run Post Hoc is the overall P is found to be smaller than 0.05.

• SD – how the spread is.

• Is large -> indicate …

• Mean n SD.. Must be together.

• P smaller than 0.05

• P = 0.024

• Reject NULL

CORRELATION

• Relationship –positive – same direction

(berkadar terus)

• Seeking for the relationship among the variable

• Characterized by the strength

• Correlation coefficient

Objective To investigate if there are significant relationship between job satisfaction and perceived support received on instructional processes

Research Question

Is there any significant differences between job satisfaction and perceived support received on instructional processes

Hypothesis

Null There is NO significant relationship between job satisfaction and perceived support received on instructional processes

Alternative There is A significant relationship between job satisfaction and perceived support received on instructional processes

Statistical Tool Correlation analysis using Pearson correlation coefficient

Decision REJECT the NULL hypothesis (P=0.000, r=0.680)

Note: If p-value<0.05 REJECT NULLIf p-value>00=.05 FAIL TO REJECT NULL

Summary There is a statistically significant relationship between job satisfaction and perceived support received on instructional processes.

Why bivariate?

• Because two variable

• Must correlation between -> both continuous scale..ratio or interval

• A two-tailed test will test both if the mean is significantly greater than x and if the mean significantly less than x.

Pearson correlation

• Pearson correlation coefficient tell how strong the relationship is.

• A change of one variable may not results of other variable –not very strong relationship

• Will definitely change the result – very strong relationship

• The slope does not indicate the strength

Pearson correlation

• How much y changes due to one unit change of X = SLOPE

• Strength = max is 1

• r=0.68

• Slope important in

regression.

The closer to 1 the stronger the relationship

• Positive relationship- the closer to 1 the stronger the positive relationship is.

• Negative relationship- the closer to -1 the stronger the negative relationship is

• Look at P value. Smaller than 0.05.. Significant.

• The relationship between ….

• Relationship exist – then run the regression.

• The purpose of running the regression is to get the prediction equation

• Simple linear regression – One predictor (IV) and one outcome (DV)

• In reality – DV requires more than one IV -> Multiple regression analysis.

Objective To examine the predictive ability of PS (instructional Processes) as the predictor of JS

Research Question

What is the predictive ability of PS as the predictor of JS?OrHow well the variation in JS is explained by PS (instructional Processes)?

Hypothesis

Null PS (instructional Processes) is not a significant predictor of JS.

Alternative PS (instructional Processes) is a significant predictor of JS.

Statistical Tool REGRESSION ANALYSIS

Decision REJECT THE NULL HYPOTHESIS

Summary R2 value of 0.463 indicates that 46.3% of the variations is the JS is explained by the predictor, PS. The remaining 53.7% of JS is explain by other predictor which are not included in the model. (The adjusted R2 the R2 which would have been obtained if the data came from population from which the sample was chosen) was found 0.460, indicating and almost similar value to the R2 found in the sample.From the anova table (in regression) PS was observed to be a significant predictor of JS. (P=0.000; F=175.676)

Regression analysis

• Regression analysis – is to create regression equations. That we will use to predict the outcome.

• R2– tell how good the predictive

• 46.3% of the outcome is explain by the predictor.

• Adjusted R2 – that r2 that you would have obtain if the data came from the population which the sample was chosen.

• P value smaller – reject NULL

The regression equation

• JS=1.278+0.716*PS

• One unit of change is PS will result 1.278 in JS

slope

• The strength based on R

• And the slope

HOW to RUN Multiple Regression

• JS=0.918+0.441*PS(IP)+0.125*PS(CP)+0.278*PS(GW)

• To rank the strength of the predictor by rank the Beta (Standardize Coefficients) or t value but disregards the sign (- or +)

• IP->GW->CP

Special Thanks to:

Dr. Mohd Burhan Ibrahim

Assistant Professor,

Kulliyyah of Education,

International Islamic University Malaysia (UIAM)