You are on page 1of 23


Dr. Ferlis Bin Bullare @ Bahari
Head (Senior Lecturer)
Psychology and Social Health Research Unit
Faculty of Psychology and Education
: 016 520 2070
Facebook : Ferlis Bahari
Research Method in Education (TT60104)

"Hard" science
Literature review must be done early in study
Tests theory
One reality: focus is concise and narrow
Reduction, control, precision

"Soft" science
Literature review may be done as study
progresses or afterward
Develops theory
Multiple realities: focus is complex and broad

Mechanistic: parts equal the whole

Discovery, description, understanding, shared
Organismic: whole is greater than the parts

Report statistical analysis.
Basic element of analysis is numbers

Report rich narrative, individual interpretation.
Basic element of analysis is words/ideas.

Researcher is separate
Context free
Reasoning is logistic & deductive
Establishes relationships, causation
Uses instruments
Strives for generalization
Designs: descriptive, correlational, quasiexperimental, experimental

Researcher is part of process
Context dependent
Research questions
Reasoning is dialectic & inductive
Describes meaning, discovery
Uses communication and observation
Strives for uniqueness
Designs: phenomenological, grounded theory,
ethnographic, historical, philosophical, case
Sample size is not a concern; seeks "information
rich" sample
Provides information as to "which beans are
worth counting"

Sample size: 30 to 500
"Counts the beans"

2007. . or 4) to test a theory about the nature of underlying processes (Tabachnick & Fidell. 608). and to 3) provide an operational definition (a regression equation) for an underlying process by using observed variables.FACTOR ANALYSIS   The purpose of using factor analysis is to 1) summarize patterns of correlations among observed variables. 2005). p. prior to using them in other analyses such as multiple regression or multivariate analysis of variance (Pallant. 2) to reduce a large number of observed variables to a smaller numbers of factors. Factor analysis can also be used to reduce a large number of related variables to a more manageable number.

 Confirmatory factor analysis is a more complex and sophisticated set of techniques used later in the research process to test (confirm) specific hypotheses or theories concerning the structure underlying a set of variables.EXPLORATORY VS.  . CONFIRMATORY FACTOR ANALYSIS There are two main approaches to factor analysis that you will see described.  Exploratory factor analysis is often used in the early stages of research to gather information about (explore) the interrelationships among a set of variables.

50. 4) Communalities of the variables must be greater than . 5) The factor loadings of . 1999) have suggested that those items/variables that have communality values less than 0. Hair et al. (. & Hong. 7) Percentage of varianced explained usually 60% or higher. If the low communality item contributes to a well-defined factor.5 can be retained when the sample size is over 500. Tabachnick & Fidell. 6) Factors with eigenvalues greater than 1 are considered significant. (. six assumptions need to be met according to the guideline recommended by Hair et al. a researcher should consider retaining it. Field (2005) and others scholars (MacCallum. 2003). Note: In terms of communalities.APPROPRIATENESS OF FACTOR ANALYSIS   In order to ensure the appropriateness of factor analysis.50.60. Zhang. (2006. 2010). 2008) 2) The result of the Bartlett’s test of sphericity should be at least significant at .70 Neuman.05. Widaman.30 or above for each item are considered practical and statistically significant for sample sizes of 350 or greater. (2006) also noted that a researcher may take into account whether to retain or remove those items/variables which have a low communality. 3) Anti-image correlation matrix of items should be at least above .50. 1) Kaiser-Meyer-Olkin measure of sampling adequacy (KMO) values must exceed . .


50. Neuman. 2008) KMO Bartlett’s test of sphericity . 2003). (. Tabachnick & Fidell.FACTOR ANALYSIS  Kaiser-Meyer-Olkin measure of sampling adequacy (KMO) values must exceed .70. (.60.

70.50 (. Jika keputusan menunjukkan nilai yang signifikan iiatu k < . 2008) Ujian Bartlett’s Test of Sphericity digunakan bagi mengenal pasti sama ada korelasi antara item memadai untukdilakukan analisis faktor. Jika nilai korelasi yang sama yang wujud antara dua atau lebih item. Ujian KMO juga membantu pengkaji mengenal pasti sama ada item-item tersebut sesuai atau tidak sesuai untuk dilaksnakan analisis faktor. Tabachnick & Fidell. .60. item-item itu mengukur aspek yang sama. maka korelasi antara item-item memadai untuk dilakukan analisis faktor. Neuman. (.KMO & BARTLETT’S TEST      Kaiser-Meyer-Olkin measure of sampling adequacy (KMO) mempamerkan multicollinearity. Analisis faktor boleh dijalankan jika nilai KMO lebih besar daripada 0. 2003).05.

808 -.023 -.154 .811a a.059 -.072 -.155 -.139 -.186 .075 .051 -.508 -.155 -.041 LOYf irst -.377 -.126 .198 -.140 -.373 -.237 -.919 -.069 -.816 -.091 a LOYf riends -.113 LOYrepeat -.026 LOYrecom -.50 .377 .490 -.077 .140 -.598 -.072 -.216 .077 -.542 -.042 -.041 -. Measures of Sampling Adequacy(MSA) Anti-image correlation must above .041 -.069 -.059 -.095 a LOYf irst -.876 -.095 -.026 -.059 .051 -.042 .023 -.139 .091 -.126 -.023 -.154 -.075 -.186 -.FACTOR ANALYSIS – ANTI IMAGE CORRELATION MATRIX Anti-image Matrices LOYpositiv LOYf riends LOYrecom LOYf irst LOYrepeat LOYcontinu Anti-image Covariance LOYpositiv .042 -.551 .401 -.551 LOYcontinu -.059 a LOYrecom -.383 a Anti-image Correlation LOYpositiv .876 -.237 a LOYrepeat -.042 LOYf riends -.113 -.041 -.075 -.373 .075 .216 LOYcontinu -.023 -.198 .

000 1.573 .649 .556 . Communalities LOYpositiv LOYf riends LOYrecom LOYf irst LOYrepeat LOYcontinu Initial 1.000 1.COMMUNALITIES OF THE VARIABLES MUST BE GREATER THAN .677 Extraction Method: Principal Component Analy sis.000 1.539 .5 .50.640 . Note: you need to take note to those variables below 0.000 Extraction .000 1.000 1.

Note: cumulative% should not below 50%.EIGENVALUES AND % TOTAL VARIANCE Factors with eigenvalues greater than 1 are considered significant. usually 60% or higher .

EIGENVALUES AND % TOTAL VARIANCE Factor 1 had an eigenvalue of 6. Factor 2 captured 16. .40.25 and explained 31.98% of the total variance with an eigenvalue of 3.25% of the total variance.

538 .810 .625 .727 .355 .558 .728 .831 .749 . after remove it you need to re-run the data reduction process again until you fulfill Hair et al.596 .838 . a.795 .340 Extraction Method: Principal Component Analy sis. Rotation Met hod: Varimax with Kaiser Normalization.654 . . (2010) guideline a Rotated Compo nent Matr ix Component SOBimpres SOBaccept SOBimprov SOBapprov EBf eelgood EBdelight EBpleasure EBconf iden EBf eelsexy FBsuit able FBreliable FBconv nien FBsolution FBremov e FBbeaut y SY Bprest ig SY Bt ast e SY Bdesire SY Bf itsocia 1 .611 .837 .772 .317 2 3 4 Crossloaded .647 .Need to remove item if it cross-loaded on other factor(s) : one by one.786 .774 .360 . Rotation conv erged in 5 iterations.810 .




.7 is considered acceptable and less than 0. 2003).8 is considered good.6 is considered to be poor (Sekaran.0 to 1. A Cronbach’s alpha close to 1. The most common measure of reliability is internal consistency of the scale (Hair et al.RELIABILITY TEST    Reliability analysis was performed in order to determine whether a group of items (i. .0. 0.e. 2005). items measuring a construct generated from factor analysis) consistently reflected the construct it is measuring (Field. Cronbach’s alpha was calculated in order to examine the internal consistency of the scales used in this study. Cronbach’s alpha coefficient can range from 0. 2006).0 indicates that the item is considered to have a high internal consistency reliability. above 0.



AFTER CHECKING RELIABILITY ANALYSIS when you are satisfied with reliability analysis of each of the dimensions and/or constructs that was generated from the factor analysis  You need to compute the mean scores for each of the dimensions and/or construct(s).  .

1. then click on Compute. In the Target variable box type in the new name you wish to give to the total scale scores Double-check that all items are correct and in the correct places. From the menu at the top of the screen click on: Transform. 2. then find Mean . Click OK. Click All.

DESCRIPTIVE ANALYSIS The mean and standard deviation values for all of the study variables/dimension/construct.  a mean score between 2 to 4 is rated as average.  Based upon the scale of 1 to 5. and  a mean score of greater 4 is rated as high.  . the mean scores can be explained as:  a mean score that is less than 2 is rated as low.