Basics of statistical analysis using spss-

programme

Docent PhD Eija Metsälä and PhD Kaarina Pirilä ©.

1
SPSS for Windows –program (Statistical Package for
Social Sciences).
• Purpose of this course: To introduce the student to spss-program
and to help to get her/him started to use statistical methods and
spss-program.
• Contents of this educational package:
• Saving data
• Data transformations
• Descriptive statistics
• Performing and interpreting some basic statistical tests.

2
About SPSS-program:
• Easy to use.
• Operates in Windows environment.
• Saving and transforming empirical data.
• Statistical analysis with graphics.
• Wide variety of statistical analysis methods from
basic descriptive ones to time series analysis.

3
There are lots of spss-guides freely available in the
internet
Learn spss in 15 minutes 1: Tutorial video
https://www.youtube.com/watch?v=TZPyOJ8tFcI

SPSS Tutorials
https://www.spss-tutorials.com/basics/

A guide to spss the Basics

http://www.sussex.ac.uk/its/pdfs/SPSS_Brief_Guide_21.pdf

4
When you open the program you get first:
Data Editor –window:
-> Data view: data input and saving to the matrix.
-> Variabe view you get by clicking the respective sign in
the lower part of the matrix: defining the variable
properties like type, decimals, name, values, missing
values, scale…
• Own window for outputs.
• Filename endings: data.sav; output.spo

5
6
 DATA VIEW –window:
• File: making datamatrix, opening, saving, printing, exit.
• Edit: cut, copy, paste, clear –functions.
• View: extra options for viewing data.
• Data: operating with data matrix e.g. select cases, split data,
insert variable or case.
• Transform: making data transformtions like constructing sum
variables and making combined variable values.
• Analyze: all the stat analyses.
• Graphs: making graphics (not good looking in text, just for
analysis purposes)
• Utilities: looking at variable definitions (rarely needed).

7
8
VARIABLE VIEW –window
Variable name (e.g k1, k2, k3). Start with letter.
• Own row for each variable and if there are variables there the
respondent may select several options, also own row for each
variable values that are handled like separate varaibles e.g What
journals does the student read? Student reads Donald Duck 1=yes
0=no; Student reads Daily Mail 1=yes 0=no etc…
• Variable type: numeric, string (text), date…
• Width: number of characters in the column
• Decimals: number of decimals in the column
• Label: Proper variable name.
• Values: defining descriptive values for the numbers.

9
• Missing: defining numbers that you want to treat like missing
values. If you leave this empty the program deals with empty cells
like missing values.
• Scale: Must define variable scales, otherwise proper analysis not
possible.
• Columns: width of column.

Remember to save your data definitions!

10
Typing in data
• Hard copy questionnaire: one questionnaire is respective to one
row. Mark questionnaires by pen with respective row numbers->can
check up if type in errors suspected.
• Own column for each variable and if there are variables there the
respondent may select several options, also own row for each
variable values that are handled like separate varaibles e.g What
journals does the student read? Student reads Donald Duck 1=yes
0=no; Student reads Daily Mail 1=yes 0=no etc…
• If you collect the data by webquestionnaire you may convert the
data often directly to spss or Exel data file

11
Stages of data analysis
1. Checking out data
2. Making data transformations
3. Examining variable distributions: according to normal
curve or not?
4. Descriptive statistics
5. Choosing analysis methods
6. Making statistical analysis

12
Exploring the distribution of the variable
1. Analyze/Descriptive statistics/Frequencies.
2. Click Statistics. Mark v sign to: Std.Deviation , Mean,
Skewness and Kurtosis, then: Continue.
3. Press Charts: Put v to Histograms and v to with
normal curve - > makes hypothetic normal curve.
Then: Continue. OK.

13
Defining min and max-values: analyze->descriptive
statistics->descriptives

14
15
16
Normal curve

sd = sum (x-mean)2
N

Properties: sd less than half of the mean, kurtosis near 0 and true value of skewness<0,2

17
Data transformations: making classified variable values:

• There may be need to combine variable values to classified

variables in continuous variables like age (k1), years of work
experience (k11_v).
• Do not destroy the original one but just make a new one -> use
option transform into new variable.
• Stages: make transformation, define variable values in the variable
view, save.

18
Original distribution of the students’
age in years (k1)

19
Practice data transformation
• Define age ranges: <20y, 20-24y, 25-29y, 30-34y, 35 and over
• Make classification: transform/ recode/into different variables. Shift
variable k01 to operating cell, name new variable ”age ranges”, and
label ”Age in 5 ranges’’ or ’’classified age ”. Press change.
• Define age ranges pressing Old and new values.
• First old value = range lowest through 19 (< 20) and respective
”New value” 1 accept by add-button
• Next old value: range 20 through 24 and respective new value
2…When all the classifications have been made press: continue
and ok. The new variable will be formed last to data view –window.
• Remember to define values in Variable View –window. After that
take f:s from the new classified variable

20
21
Naming age ranges

22
Frequences of classified variables

IKÄLUOKA ikä luokiteltuna

Cumulative
Frequency Percent Valid Percent Percent
Valid 1,00 alle 20v 6 3,4 3,4 3,4
2,00 20-24 97 55,4 55,4 58,9
3,00 25-29v 43 24,6 24,6 83,4
4,00 30-34v 15 8,6 8,6 92,0
5,00 35 ja yli 14 8,0 8,0 100,0
Total 175 100,0 100,0

23
Data transformations: Forming Sum variables.
• Via Compute variable-option.
• E.g. when there are various items in some scale,
purpose is to make out of these scales a summated
variable. 1.k14_1 (measures how students may see the
entity of studies out of the curriculum), 2. k14_2 (how
students are able to see the progress of the studies
from the curricum) ja 3. k14_3 (how well the curriculum
guides planning of the studies)

24
Transform-option: compute variable with mean-method.

• Target variable: new name for the summated variable: PLANNING

• Numeric Expression choose from Function group boxista All.
• And from Functions and Special Variables –option – mean.
• To the function box add items=selected variables and separate
them with commas (k14_1,k14_2,k14_3)

Press ok.

25
Doing statistical analyses
• Scale and distribution of the dependent variable must
be taken into account when choosing methods of
statistical descriptives and tests.
• To study central tendencies: mode (for nominal and
ordinal variables), median and mean (interval and
scale), sd, skewness and kurtosis.
• Exploring associations, group differences…

26
Age by degree program means

Analyze/compare means/means, dependent k1 (age)

and independent k5 (degree program).

27
Exploring associations
• Correlation coefficient Pearson or Spearman
analyze/correlate/bivariate and then choose the variables those
association you want to explore.

E.g.
K14_1 The curriculum gives you an easy overview of your studies.
and
K14_3 Curriculum supports well planning my studies.
Look at the scale of the variable, which correlation coefficient to use?

28
Any association? Statistical significance?

29
Cross tabulation
• For exproring associations of nominal or ordinal scale
variables.
• Analyze/descriptive statistics/crosstabs, then select
column and row variables.
• Explore gender=sp (k2) by variable (k14_1).
• Rows var. k14_1, Columns var sp k2.

30
• For statistical tests press statistics→ chi-square, for
calculating percentages press cells, choose column
percentages and then you get 100% calculated by
columns.

31
32
Chi square
• Tests if there is any association between row and
column variables.
• Analyze/Descriptive statistics/Crosstabs/statistics/chi-
square
• H0 is that there is no
association.

Is there in this case?

33
T-test
• Testing group means of the dependent, two groups
• Analyze/compare means/independent samples t-test,
dependent (test variable) independent (grouping
variable)
• Esim. dependent k14_1 by independent k2 (gender).

34
Levene’s test for equality of variances >0.05 means that variances are equal
(H0 about the normality of variances apply). How about the test score stat.
significance?

35
ANOVA
• Testing group means and more than two groups
• Analyze/compare means/one-way anova then select
dependent and indipendent (Factor) variables
• E.g. dependent k1 and indipendent k5 (Factor)
variables

36
Interpret

37
ANCOVA
Analyze->General Linear Model->Univariate
Put the dependent (14_1) in the Dependent box and the Independent
(Degree programme) in the Fixed factor box. Covariate (Gender) into fixed
factor box

38
Levene´s test non significant-> equal variances in groups

39
P value 0.040 tells you that there is significant difference in opinions
of students from different DPs (k5) regarding how they see curriculum giving
Easy overview to their studies, whilst adjusting for gender (k2)

40
If ANCOVA is significant Post hoc test with Bonferroni should
be performed

How to do ANCOVA and post hoc tests https://www.youtube.com/watch?v=kY1XXORXl3s

41
42
Calculating Cronbach’s alphas
• Analyze/Scale/Reliability analysis
• Put to the Items box ordinal variables 1 (The
curriculum…) to 5 (Planning my studies)
• Give scale lable: Planning studies
• Open Statistics, select: Item, Scale, Scale if item
deleted
Press Continue, then Ok
THANK YOU!
www.metropolia.fi

www.metropolia.fi/en
www.facebook.com/MetropoliaAMK

www.facebook.com/MetropoliaAMK
[email protected]

[email protected]