How to analyze survey research data?

SPSS Interface Data Editor

• Variable view – Name – eg id, gen, schcat, a1, … (space bar, ?, ! *, !, …

considered as illegal characters) – Type – Width – Decimals – Label – ID, Gender, School Category, Item 1a, etc – Values – eg 1 – Strongly Disagree, 2 – Disagree, 3 – Neither

agree nor disagree, 4 – Agree, 5 – Strongly Agree – Missing – Columns – Align – Measure – Role

SPSS Interface Data Editor

• Data view

– Entering data

– Save file

Reliability Analysis

• Analyze > Scale > Reliability Analysis

• Overall

–Corrected Item –Total Correlation. Omit item with r < .3

• Construct a

• Construct b

• Construct c

Descriptive Persepsi Pelajar Terhadap Sekolah

• Jantina & Tingkatan

– Analyze > Descriptive Statistics > Frequencies > Jantina & Tingkatan

• Jantina Mengikut Tingkatan

– Analyze > Descriptive Statistics > Crosstabs > Rows (Jantina) & Column (Tingkatan)

Descriptive Persepsi Pelajar Terhadap Sekolah

• Score & percentage for each item

– Analyze > Descriptive Statistics > Frequencies > (All items)

Calculate Means (Persepsi Pelajar Terhadap Sekolah)

• Recode negative items

– Transform > Recode

– 1 5

– 2 4

– 3 3

– 4 2

– 5 1

Calculate Means (Persepsi Pelajar Terhadap Sekolah)

• Means

– Overall

– Transform > Compute variable > Target variable (MEAN) > Numeric Expression

Calculate Means (Persepsi Pelajar Terhadap Sekolah)

• Means

– Prasarana Sekolah

– Tenaga Pengajar

– Kepimpinan Sekolah

Compare Means (Persepsi Pelajar Terhadap Sekolah)

• Gender (Male, Female)

– Overall

– Prasarana Sekolah

– Tenaga Pengajar

– Kepimpinan Sekolah

T Test

• One-Sample T Test

– To compare sample mean and population mean

• Independent-Samples T Test

– To compare means of 2 different groups

• Paired-Samples T Test

– To compare means of 2 sets of data of the same independent variable

Jika Ujian Levene signifikan (p>.05), gunakan baris pertama. Jika Ujian Levene tidak signifikan (p<.05, gunakan baris kedua.

Ujian menunjukkan t(34) = .537,

p=.595 adalah tidak signifikan.

Keputusan ujian menunjukkan tidak

terdapat perbezaan skor min

persepsi terhadap sekolah yang

signifikan antara pelajar lelaki dan

pelajar perempuan

Assumptions

• Normality – The assumption of normality is a prerequisite for

many inferential statistic.

– To explore normality graphically • Histogram

• Box-plot

• Stem-and-leaf plot

– To explore normality statistically • Shapiro-Wilk statistic

• Skewness

• Kurtosis

Assumptions

• Normality (graphically)

– Analyze > Descriptive Statistics > Frequencies

– Move variable (eg Min Keseluruhan) into the Variable box.

– Click Charts >`Histogram’, `Show normal curve on histogram’

Assumptions

• Normality (statistically) – Shapiro-Wilk

– Analyze > Descriptive Statistics > Explore

– Move variable (eg Min Keseluruhan) into the Dependent List box.

– Click Plots > Normality plots with test

– If the significance level is >.05, then normality is assumed.

Assumptions

• Homogeneity of variance

– The groups should come from populations with equal variances

– Levene test

• If the test is significant (p < .05), it shows that the variances are unequal.

• If the test is not significant (p > .05), it shows that there is no significant differences between the variances of the groups.

Compare Means (Persepsi Pelajar Terhadap Sekolah)

• Tingkatan (Ting 1, Ting 2, Ting 3)

– Overall

– Prasarana Sekolah

– Tenaga Pengajar

– Kepimpinan Sekolah

Analysis of Variance

• One Way ANOVA

– To compare means of more than 2 groups of an independent variable.

– Analyze > Compare Means > One-Way ANOVA

– Options > Descriptive > Homogeneity of variance test

Analysis of Variance

• One Way ANOVA

– To compare means of more than 2 groups of an independent variable.

– Analyze > Compare Means > One-Way ANOVA

– Options > Descriptive > Homogeneity of variance test

The analysis showed that there is no statistically significant difference at the level of p < 0.05, F(2, 35) = 1.461, p = .247.

Correlation

• Relationship between 2 variables.

– Value of correlation coefficient, r: -1 < r < 1.

– r < 0 – negative correlation

– r > 0 – positive correlation

– r = 0 – no correlation

– r + 1 – perfect correlation

– R > 0.8 – strong correlation

– R < 0.5 – weak correlation

Correlation

• Relationship between 2 variables. – Pearson product-moment coefficient

• The relationship between 2 continuous variables

– Phi coefficient • The relationship between 2 categorical variables

– Point-biserial correlation • The relationship between a continuous and a

categorical variable

– Spearman’s rank-order correlation • Assumption underlying correlation cannot be met

adequately

Correlation

• Assumptions: – Related pairs: data must be collected from related

pairs

– Scale of measurement: data should be interval or ratio in nature

– Normality

– Linearity: the relationship between the 2 variables must be linear

– Homoscedasticity: the variability in scores for one variable is roughly the same at all values of the other variable

Correlation

– Normality

• Normality – Shapiro-Wilk

–Analyze > Descriptive Statistics > Explore

–Move variable into the Dependent List box.

–Click Plots > Normality plots with test

– If the significance level is >.05, then normality is assumed.

Correlation

– Linearity: the relationship between the 2 variables must be linear

– Homoscedasticity: the variability in scores for one variable is roughly the same at all values of the other variable

– Scatterplot: Graphs Legacy Dialog Scatter/Dot Simple Scatter

There is a linear relationship between pre-test and post-test

The scores cluster uniformly around the regression line, the assumption of homoscedasticity has not been violated

Correlation

Analyze Correlate Bivariate Pearson

Non-Parametric techniques

• Assumptions:

–Random sampling

–Variability across distribution

– Independence i.e. subjects appear in anly one group and the groups are not related in anyway

Non-Parametric techniques

• Mann-Whitney test

• Kruskal-Wallis test

• Spearman’s rank order correlation

• Chi-square

Non-Parametric techniques

• Chi-square

– To discover if there is relationship between 2 categorical variables

• Mann-Whitney test

- To test that 2 independent samples come from populations having the same distribution.

• Kruskal-Wallis test

– To examine differences between two or more groups.

• Spearman’s rank order correlation (Spearman rho)

– A non-parametric alternative to the parametric bivariate correlation

Chi-square

• To discover if there is relationship between 2 categorical variables

–Variable 1: Gender (Male, Female)

–Variable 2: Hometown (Kuching, Kota Kinabalu, Others Urban, Rural)

Chi-square

• Hypotheses:

–Ho: Hometown and gender are independent

–Ha: Hometown and gender are related

• Analyze Descriptive Crosstabs Statistics Chi square

Conclusion: Hometown and gender are related.