Sei sulla pagina 1di 8

Methodology

How to justify the methods


Methods
Process which used to write a paper

1-Research design
Outlay of our research.
What are variables?
Data collecting method like questioners or interview
Four main part of RD
I. Research approach
II. Time horizon
III. Study setting
IV. Unit of analysis

1-Research approach
When you are working on such a variable which is not completely defined and no questionnaire
available then we to conduct Qualitative research by interviews
When variables are well defined and questionnaire are already developed then Quantitative approach is
used and we prefer it
2- Time horizon are of two types
Cross sectional
Only one time data is collected from one person.
Longitudinal
Data is collected from same person two or more times after some interval of time.
It is benefit able when cause & effect relation is present between variables like Training enhance the
performance
3- Studysetting are of two types
Contrived
Environment is artificially controlled by researcher.
Non-contrived
Data is collected in natural environment. Researcher interfere should be minimum.
We are student of social environment so we prefer non-contrived.
4-Unit of Analysis
UOA can be individual, dyads and groups .Here I am using dyads.
Individual
When data is collected from individuals
Dyads
When data is collected in the form of pairs like employee performance evaluation from supervisor and
supervisor support from sub-ordinate employee.
Group
When data is collected from groups.

2- Population
Target marketing
Not organization but their employees are population
Same like user of a brand
There are two types
Known
When list of population is available like students of a class
Employees of org when list is available
You can’t say known when only figure is available like 200 million people of Pakistan
2000 employees of an organization
Unknown
When only figure is available or may be figure is also not available

3-Sample and sampling technique


Sample is a subset of population
Sample is selected based on population type
krejcie and morgan 1970 formula for sampling
Download it and check page 2
As the population is known so sample is selected on basis of above formula and selected sample size
As population is not known so study use item response theory
If total questions (items) are 21
21 X 10 = 210 sample size
21 X 20 = 410
It depends on software requirement which we are using for hypothesis like SPSS which required mini
250 so here we used 21 X 20 = 410.
Sampling technique is also based on type of population.
If sample is known then we will use one of probability sampling technique.
i- Simple random sampling technique (ST) like balloting
ii- Systematic random sampling technique
Any nth number but should be extracted at-least sample size
iii- Stratified sampling technique
Population is divided in strata like employees of govt. and private org
Then simple random sampling is applied to each strata(group)
iv- Cluster sampling
Least acceptable technique
Mini number of persons in one cluster should be equal SS
May be one type of people selected in one cluster which is the failure reason for cluster sampling

If sample is unknown then we will use one of non-probability sampling


Convenience sampling technique
It depends on personal judgment of researcher which will select it common sense and collect data from
relevant peoples

In our research we collect data from individuals and Unknown sampling is used with convenience
technique

10-4-2019 lecture
SPSS software
Collect the respondent data and arrange with number for later on tracking
Variables are defined in Variable View in ist column
Arrange variables as per your questionnaire form
Abbreviations of var are used here as data is only for researcher .
Missing data Analysis
Analyze ----------Descriptive statistics -------------- Frequency
Demographics missing values not need to treated
Select all qs of one variable
Press arrow
Press Ok
If any missing value is found then treat them as following process
Missing values treatment
Transform ---------------- Replace Missing values
Select those variables which have missing value
Select “series mean” method
Press OK
New column was created for same variable with avg value at missing place. Now use this for further
analysis.
Respondent if answers less than 75% qs than it should not be consider for missing analysis just throw it.
All missing values columns may be selected for same time.
Outliers Analysis
Outliers only find in dependent variables including mediating var.
Outliers shows extreme data so it must be removed.
For outlier we have to take the avg of all qs of same variables
Transform------------ compute variable
Define Target Vale may as “Avg_POS”
Write formula
Mean (POS1,POS2,POS_3,POS4,POS5)
NOTe: We have used POS_3 instead of POS as POS_3 is formed after missing treatment.
Press OK
Analyze ----------Descriptive statistics -------------- Explore
Select “dependent variables avg“ as find above
Press statistics
Select “Outlier”
And accuracy will be 95%
Press “Plots”
Use “Stem and leaf”
Outliers shown around the points
Note all of them
List them in ascending order
Delete them from data by RMB on line and using command CUT (13 and 16 are liers)
After removing outliers counting of respondents should be greater than 250.

Reliability
Cronback alpha
Internal consistency
If our answer is greater than 0.7 then our questions are reliable
If less than 0.6 then data is not acceptable
So ist test a pilot study for it if results are reliable then continue otherwise form a new questionnaire.

Table 1 : Mean, standard deviation, Reliability and correlation


Variables means s.d α POS AC PCB
POS 3.35 0.64 0.82 1

AC 3.39 0.63 0.79 0.74** 1

PCB 3.62 0.53 0.63 0.55** 0.62** 1


Note: POS = perceived organization support , AC= affective commitment, PCB = psychological
contract
α = reliability, **P<0.01

Different arrangement can be used to avoid pledgerization. Only variables pattern should be
same in column and row so diagonal pattern formed for value 1
Perfect correlation is +_ 1
In SpSS
Analyze ------descriptive statistics ------descriptive
Select avg of all variables
Go to options
And select
Mean and S.d
Run command

Pick values after Round off up to 2 decimal


If mean is greater than 3.5 than consider it as agree
We don’t interpret s.d in research.
Analyze -----scale------- reliability
Reliability is related to questions scale
So select all questions of a variable
And press Ok
Note reliability of that variable
Cronbach’s alpha
Repeat same process for all variables
As per more statisticians, reliability above 0.6 is acceptable. Our value reliability is
greater than 0.6 so there are reliable
Correlation
Analyze -----correlate------ Bivariate correlation
If sig vale is less than .01 than its confidence is 99% and show as **
If sig value is less than .05 than it confidence is 95% and show as *

Interpretation of table
The values of table 1 show mean, standard deviation, reliability and correlation
among observed variables. Regarding mean values most of the respondents were
neutral about their perception of organizational support (M=3.35) ,affective
commitment (M=3.39) but agreed about their level of psychological contract
(M=3.62).This study followed the standard criteria of hair et al.(2017) and noted
all the variables are reliable as the values of reliability was noted well above the
value of 0.60.
Further, the values of table show that perceived org support has a positive
correlation with affective commitment (r=0.74, P<0.01), Psy contact and ….
Note : r is used for correlation and should be italic and p also in italic
To avoid pledgrization try to use abbreviations in above paragraph
Above mentioned style of interpretation is academic and as per apa-6 so don’t try
in formal way.
Lecture 24-04-2019
Regression analysis
Regression is used if we only testing impact of one variable on other variable.
Analyse____ regression --------linear
Only one dependent variable can be selected
More than one independent variable can be selected
But try to check one by one
Model summary
Linear regression then R square
Co-efficient
Value of un standard co-efficient beta… in front of independent variable
Sig means significant
Sig co related with P
if
P t
Less than < 0.01 Greater than >2.5
< 0.05 >1.98
Insignificant if > 0.05 <1.98
Table 2
Relationshi
R square β P Result
p
POS ----> AC 0.55 0.73 <0.01 Accepted
Accepted as result is significant.
Values in table 2 shows that when employees get POS it positively impacts their affective
commitment.
R-sq shows the percentage change in dep. variable due to change in indep. variable.
Beta show how much unit change occur in dep. variable. when one unit change occur in
independent variable.
Hierarchal regression
Form 3 models with dependent variable.
Model 1 select demographic variable all demographic variables
Model 2 select independent variable--- avg. values
Model 3 select mediating variable --- avg. value
According to barren and kanne 1986, we used hierarchical regression to check the impact of
mediating effect of AC
0.11 shows insignificant result which it is fully mediating if value is significant than it is partially
mediating.
Table 3 : Mediating role of AC between POS and WE
Variables M2 (β) M2 (β) M3 (β)
Control variables
Age -0.18* -0.11** -0.09**
Gender 0.07 0.1 0.09
Qualification 0.03 0.01
Independent Variable.
POS 0.62** 0.11
Mediating Variable
AC 0.69**
R square 0.02 0.32 0.48
Delta R square 0 0.3 0.16
Control variable means demographic variable
if sig value is less than 0.05 then add one star and if less than 0.01 than two star and if greater
than 0.05 than no star
Moderating analysis
Form 2 models with dependent variable
Model 1 select independent variable and moderating variable-- avg value
Model 2 select interactional term --- avg value
Interactional term will find by multiplying ind. variable and moderating variable.
For interactional term in spss
Transform --- compute variable
Target variable = POS X PCB
Formula = (POS)*(PCB)
Then go to
Analysis --- Regression ---- Linear
And select as per above mentioned model
Consider following model

POS --------------------------🡪 AC

PCB

Interactional term = POS X PCB

Table 4 : Moderating role of AC between POS and WE


Variables M1 (β) M2 (β)
Step 1
POS 0.57**
PCB 0.36**
R square 0.62
Step 2
POS X PCB -0.19**
R square 0.64
Delta R square 0.02

This study is carried out for moderating variable .Table 4 values are showing that when ind.
variable and moderating variable added then results are significant .and POS X PCB results are
significant which means that PCB is moderating POS and AC.

Potrebbero piacerti anche