Example (for Task 6) Variable Measurements and Sources… 151 Note: Make sure you do not lose your measurement list because you will need to report them in your thesis or journal article.
Another Example (for Task 6) Note: These measurements have been adjusted to suit the study’s context... “adapted”. 152
Task 7 After Presenting Task 6… (30 minutes) 1. Improve & finalize your questionnaire based on the comments received 2. If time permits, pre-test your questionnaire with 2 colleagues from other groups. Edit your questionnaire based on their feedback ------------------------------------------------------------------------------- 3. Print out your Model, Hypos, Definitions, Questionnaire: - make sure they align nicely with each other. - if they do; 50% of your study is complete! 4. Verify no. 3 and let’s develop your Mock Table (to confirm your Model is fit for Regression) To thoroughly understand this Chapter and get an idea of how to proficiently complete these tasks, please refer to GradEx Academy’s website and search for workshop Video Recordings: Number 4. 153
Example (for Task 7) The first few pages of a participant’s questionnaire … completed in 2 days only! 154 Note: The questionnaire was converted into an online survey in the NRA workshop and data collection resumed immediately.
To thoroughly understand this Chapter and A brief tutorial on how to get an idea of how to proficiently complete develop your Online Survey these tasks, please refer to GradEx is also provided in Video Academy’s website and search for workshop Recordings Number 4. Video Recordings: Number 4. Task 8 1. Start data collection. You have 1 Month to collect (1 hour) your data (optional: for my Publication Workshop) 2. Target around 200 responses (convenient sampling) via online survey (Google Forms) & person-administrated 3. Key-in or transfer your responses to SPSS 4. Bring your data to our Publication Workshop and we will run Regression Analysis together… to prove your hypos! ----------------------------------------------------------------------- Note: Find at least 5 journal articles related to your Focal Construct for your next task. 155
Chapter 5 The Role of Literature Review Stage 2: Applied Novel Research Section 156
“Write like it Matters, and it Will!” The role of the researcher is to justify the need for his or her study… it’s not what you know that is important, but what you can PROVE. 157
Highlights of Is Literature Review Enough? (The 2 missing ingredients) this Chapter Common Mistakes in LR (Pitfalls to avoid) General Guidelines for LR (Some important advice) How to write a LR (Linking past studies) Main, General, Overall LR (Strategize your write-up) Synthesizing the Literature (The past, present, future) Finding your Research GAPS (In just 1 simple Table) 158
Literature Search There will usually be earlier academic work related to your topic A secondary method for collecting information The first thing that you Facts gathered from previous publications should do in your literature review (LR) is to find and • mostly from journals learn about previous research • also from books, reports, articles published on your topic. Provides evidence of results of past empirical Note: The LR should observations (up to present stage) always begin with an exhaustive literature search… use keywords to find related articles. 159
Is Literature Review Enough? Searching out published research relating to the topic of interest (Literature Review) is an important early step of research. … but simply reviewing literature is not research. A good literature review does not just summarize the existing The critical review compares past research—it critically assesses findings, emphasizes where the gaps are, the literature related to a specific discusses their strengths, and highlights research topic. the shortcomings of the literature that The main goal of LR is to your study aims to fill. provide a clear picture of the state of knowledge on the subject of interest. The research process should always include a synthesis (analysis of the existing literature) & points out … How your research will move the existing literature forward (your contribution to knowledge) 160
Common Mistakes in LR • Review is not logically organized Reviewing the literature • Review did not focus on the most important aspect is a skill that is required in research, but PhD students of the study (the Focal Construct) are assumed to already hold • Reviews not related to the study those skills when they enroll • Few or outdated references in their studies. • Review reads like a series of split summaries • Did not cite all sources in reference section In fact, it’s not surprising • Recent references are omitted (last 5 years) to hear research students who do not know “how to do it right”, hence fall into the LR trap. The LR involves a sophisticated set of thought processes that often takes many months, if not years to learn. This chapter tries to clear this up so that you can manage your LR more efficiently. 161
General Guidelines for LR When you go through your key You need to articles, look at their reference list. demonstrate good knowledge of the Identify those that are relevant and search for them… this is known as field… the “snowballing” technique. • Need to cover research relevant to all the variables in your study (especially the focal construct) • Research that helps to explain the relationship between your variables is TOP priority (arguments for your hypos) • You need to plan how you will structure your literature review (strategize) • Be comprehensive (article: 50 references; thesis: >200 references) • Refer to quality journals (Scimago & JCR Listings) Leading Journal Publishers 162 Springer, SAGE, Emerald, Elsevier, Routledge, Taylor & Francis… indexed by Clarivate Analytics and Scopus. Note: Impact Factor scores are displayed on the journal’s website.
How to Write a LR It is common practice to identify areas of controversy or contesting views… this signifies the need for further research. • Identify critical issues in the literature A well-written LR - gaps, questions not yet resolved, different views, dilemma denotes familiarity with and understanding of • Draw links between previous works - defining “simultaneous launch”‒ different views, how the research in a are they different/similar, reflect on why this exist, and particular area. justify which definition is most appropriate - landmark studies that led to subsequent research • Discuss key findings from previous studies - concerning your focal construct and other variables • Report suggestions or calls for the study - the literature might suggest some aspects that has not been examined (e.g., contextual factors) “Use the literature to justify your stance… show that what you are looking at make sense.” 163
It’s equally important to Write your first draft early and know when to stop quickly. The earlier you start writing your LR the better. You reviewing your literature must accept that your first draft is (i.e., perfectionism)… exactly just that… a DRAFT. leads to delaying submission for viva 164 voce.
Planning your LR: The Process According to Different Sections of the research/thesis… MAIN Literature Review - write-up for your Focal Construct (Chap 2) GENERAL Literature Review - write-up for Theory & Hypo Arguments (Chap 2) OVERALL Literature Review - write-up for Intro, Method, Findings (Chap 1, 3, 4) You need to be able to showcase a structured and comprehensive LR. This can be achieved by determining the different sections and instructional steps to guide the write-up… in other words, “strategizing” your review. 165
Main LR: Focal Construct Note: The focal construct is the most important variable in your Model… either your IV(s) or DV. You need to be able to pinpoint your focal construct; it represents the primary attention of your study (the topic). This is the main purpose of the LR. 166
General LR: Theory & Hypothesis Note: Always place your model & hypos beside you (2-Page PhD in Chapter 1)… this will help you identify & include ONLY relevant info for your write-up (be selective). The LR section contains the easiest & hardest part of research… reporting/synthesizing your literature & arguing your hypotheses, respectively. 167
Overall LR: Throughout Paper Blueprints are key journal articles that you can use to get IDEAS on how to write your thesis or paper (story telling, flow, reasons why, how to argue, etc.) Note: The blueprint for your hypos is the most important one, have a look at: how they introduced their IV(s), MV(s), & DV(s) how they wrote arguments for their moderating hypos how they presented their regression table how they reported their results (statistics) how they discussed their findings (sub topics, flow) Keep your blueprints with you at all times! 168
Main LR: Synthesizing the Literature The synthesis Table Use a Table to review the literature allows you to organize your literature effectively Decide on your parameters (for Main LR only). • not practical to include all studies in your Table The Table helps guide • so, select only key empirical studies (Focal Construct) readers to make distinctions on the state of knowledge The Table summarizes… the past, present, future (i.e., past, present, future); it puts your literature into • What has been done (how the topic evolves) perspective and create meaning for your study. • What has not been done (gaps & shortcomings) • What needs to be done (your study!) “You need to show how your research will fit in the body of knowledge and why your study is important.” 169
Example of Main LR Table Note: Arrange your articles in descending order… to show how the literature has evolved or progressed over time. Can you find the focal construct in this Table? This table is used to both summarize & synthesize the literature (empirical papers). 170
Let’s get to Work! 171
To thoroughly understand this 1. From your Model, identify your Focal Construct Chapter and get an idea of how to proficiently complete these 2. Find at least 5 empirical journal articles related to tasks, please refer to GradEx your Focal Construct Academy’s website and search 3. Develop a Table with 5 columns - Authors & Year, for workshop Video Data Collection & Sample, Variables (IV, DV, MV, Recordings: Number 5. MedV), Moderator Effects (yes/no), Major Findings Task 9 4. Fill in the Table: extract the info from each article (1 hour) 5. Analyze the Table: find the GAPS in each COLUMN (any shortcomings? … hint: refer to Next slide) 6. How will your study contribute to the literature? 7. Show your Table, Gaps, and Your contribution ----------------------------------------------------------------------- Note: Did any of these articles mention about THEORY? If yes, you may have found the theory for your model! Same goes with ISSUE. Note: Although journal articles can help pinpoint the Theory and Issue for your study, doctoral students should refer to PhD dissertations for their write-ups… empirical papers typically will not explicate their theories and issue due to word count limits imposed for publication. 172
Synthesis: Finding the Gaps Assess Each Column (Table) Assessment of the progression rate, methodological limitations, contextual inadequacies, and controversial findings will assist in identifying the GAPS that need to be filled. Main LR Table (Column): find limitations, patterns, & draw conclusions • Author & Year (e.g., new area of interest, recently studied, pauses between years) • Data & Sample (e.g., most did qualitative approach, small sample size, data collected in developed countries, non in transition economies, focused on a similar industry or consumer group) • Variables (e.g., basic variables, most examined direct effects, not comprehensive, DV too specific/single construct, have not considered multiple dimensions for the IV/M/DV) • Moderator Effects (e.g., lack of studies, only 1 or 2 studies have considered contextual moderating effects) • Major Findings (e.g., mixed results, tried to examine moderators but were not significant, may not be applicable nowadays) Note: You should be able to “address” the limitations you highlighted… so, can you show how your study will fill the gaps? (in point form). 173
List: 322 Marketing Theories Theory is an integral part of research, especially at doctoral level. Scholars in the social sciences tend to rely on theory because it offers a broad or logical reasoning that explains the phenomena. 174
Research is all about searching for the truth – the theory(s) we choose should offer an explanation that enables us to understand, debate, and seek new knowledge about the area we are researching. Here are some examples of theories in marketing… Similarly, there are also a vast number of theories in management, education, health, and psychology. Note: You may have up to 3 theories in a study. Try not to go beyond that… 175
Remarks
Chapter 6 Data Analysis & Interpretation Stage 3: Publication Section 177
Statistical analysis provides a researcher with a wealth of information, but you need to be able to interpret the numbers first for them to be useful. “Looks Sophisticated but actually, Easy to Understand” 178
Highlights of Regression vs. PLS-SEM (When to use them?) this Chapter Statistical Procedures (The necessary ones) SPSS Data Preparation (Coding issues) Reliability Analysis (Initial detection of problem items) Confirmatory Factor Analysis (Establishing model fit) How to Use Syntax? (The hidden application in SPSS) Correlation Matrix (Avoid overlap between variables) Regression Analysis & Interpretation (Hypo testing) 179
Regression vs PLS-SEM? SPSS remains popular and used widely within the academic circle, mainly due to its stability & versatility. Most of the articles I publish uses SPSS (i.e., MRA, Hierarchical Regressions, Slope Analysis). In the last decade or so, there has been a strong movement to “discontinue” the use of • Is SPSS out of date? PLS; and many scholars have responded to • When should we use SPSS or Smart-PLS? clarify the allegations. My stand… “It’s not • What is Syntax (SPSS)? the issue of which is better, but knowing when to use them”. • Refer to your Mock Table… developed Note: Choosing the most appropriate earlier in Task 7 analytical tool for your study depends on your research objectives, type of Syntax is a specialized application in SPSS; not menu- model, & hypotheses. Rule of thumb: • Direct effect models = SPSS driven like normal but requires coding. It allows users to • Moderating models = SPSS • Mediating models = PLS perform tasks that are intricate such as testing multiple • Mod + Med models = PLS interaction effects, and documents how you analyzed your data. The Mock Table is used to structure your Syntax… “Many scholars recommend using regression analysis (SPSS) for models with moderators (e.g., Aiken & West, 1991; Barrick et.al, 2007; Sharma et al., 2010; Sheng et al., 2011) as it is the CLEAREST and BEST way to test moderating effects (Wu & Cheng, 2019).” 180
181
Statistical Necessary Procedures Analysis • Validity Analysis (CFA via Amos) The first two analytical • Common Method Bias (Herman's 1 Factor) procedures deal with achieving • Reliability Analysis (Cronbach’s Alpha) Model Fits and determining • Correlation Matrix (Multicollinearity) potential respondent bias. Main • Moderated Regression Analysis (via SPSS) objective is to detect “problem ------------------------------------------------------------------ items” in your data. Advanced Methods • Simple Slope Analysis (Statistics/Diagrams) Since there are many step-by- • Hierarchical Regression Analysis (Model Sequence) step tutorials on how to run CFA using Amos on YouTube, I will Note: Simple slopes (statistics) are rare and not direct my attention to SPSS required unless your moderators are hypothesized Syntax; to statistically test your at high and low levels. Hierarchical Regression is moderating hypotheses. The last optional, particularly for studies with a single DV three procedures can be done or models that have >6 interactions effects. using SPSS. 182
SPSS Data Preparation & Entering • Exclude questionnaires with missing values & same scores throughout • Assign a code (a Number) to each response for a question (item) in your questionnaire : Gender (male = 1, female = 2) : Loyalty (strongly disagree = 1 --- strongly agree = 5) Numerical values are • In SPSS, use simple names in Variable View needed for each : Loy1; Loy2; Loy3; Loy4; Loy5 questionnaire item; this will enable the software to statistically Variable View analyze your data. - Important: Name, Label (ok), Values, Measures Data View - to see no of respondents & responses for each question - to key in data (numbers) - responses for each quest item 183
Coding your Questionnaire: Example Coding should be manually done, usually on a “printed” questionnaire. The codes are written on the questionnaire to avoid confusion when entering your data into SPSS. Use a simple code for each item (e.g., BV1, VA1, or Vis1) 184
This is how your “variable view” should look like. Use a similar indicator for each column– short Name, Numeric, 8, 0, short Label, numeric Values, None, 8, Right, Scale/Nominal, Input. 185
This is how your “data view” should look like. From BV1 onwards, the numbers represent actual responses on a 7-point Likert scale by 357 survey participants. There are 357 samples in this dataset. 186
Reliability Analysis (SPSS) Note: A scale is said to have high reliability (internal consistency) if the items “hang together” and measure the same construct. The Cronbach’s Alpha score for a construct is viewed as the most appropriate measure of reliability for interval scales. Did you realize, so far, your actual variable/construct (in your model) is not present. Like in this case… “Visual Attribute”. This will be “created” in your syntax. 187
Reliability Scholars agree on a Analysis minimum Cronbach's (Output) Alpha score of .70. Note: If your total Alpha score is lower than .70, have a look at this Column: a high score for a particular item denotes a “problem item”. Run reliability test again without this item & DROP it from your regression analysis (syntax)… your CFA should also point to the same item as well. 188
Measures for Each Variable (CFA) An example of a CFA table with Factor Loading scores for each item (should be ≥ .60). Note: You can also show the items that you “dropped” in this table, but make sure you provide a note at the bottom… e.g., “Item was dropped during the scale purification process”. This is optional for a journal paper, but compulsory for a PhD thesis. 189
Sample Model for MRA Now, for the main event: TESTING your research hypos… the first thing to do is to “simplify” your IVs & MVs (a, b, c, d, e …). 190
MRA using Syntax (SPSS) This is an example of a SPSS Syntax that has been The Run completed and can be used Button. for regression analysis. Unlike the normal SPSS which is menu-driven, you need to manually type in the instructional codes. Do not include the “problem items” in your syntax Create a name for all variables (e.g., visual, info, function) The coding in syntax (e.g., BV1, BV2) has to be consistent with the Names you used in SPSS… directly linked Divide by the number of items (if you have 6 items: / 6 .) Insert the mean Run the data to get the Mean Scores for your variables… scores here & add Highlight the Instructions (1-2) “m” at the end of & Click the Run Button. your variable 191 name.
Simplifying the codes Coding instructions for for your IVs & MVs your Correlation Matrix… (use a, b, c, d, e …) Highlight the Instructions This is when you develop (1-4) & Click the Run the instructions for your Button. Interaction Effects. The syntax may seem to be a bit confusing in the beginning, but it’s actually very easy to do… it will take you 30-45 minutes to complete. Note: Because the codes are directly linked to your SPSS data and earlier commands you made, any “typo error” in the coding will not allow it to produce the required statistical output… so, you need to Find & Correct the typo error(s). Note: You are advised to attend my “hands-on” Publication Workshop to receive direct guidance on how to effectively write the commands for your syntax and interpret the results for your own study. 192
Instructions for your Regression Analysis… according to specific DVs. This model has 3 DVs (interest, purchase, repurchase). This will be the Most Exciting moment in your research journey… Highlight all the Instructions (1-7) & Click the Run Button. Hierarchical regressions will not be necessary for your analysis, unless advised otherwise by an experienced researcher. 193
Raw Output (Correlations) Examine the Pearson Correlation scores (should remain < .90). Are there any >.90, particularly for your IVs? If yes, this indicates that your data suffers from Multicollinearity. If this is present, you will have to resort to either “abandoning” one of the effected variables or combining them together as one construct in your model and analysis. Note: Look at the correlation values below the diagonal (1). No need to be concerned about the Significant levels and Directions. 194
Raw Output (Regression) R Squared?… the variance in the DV can be explained by the independent variables (including MVs). Studies that predict human behaviors tend to have R Squared values < .50, so don’t worry if you have a lower score here. Significant effects in a regression model are not effected regardless of whether your R Squared is 20% or 90%. 195
Raw Output (Regression) The raw output appears on a 2-tailed test. Hypothesis testing uses 1-tailed test… so to convert it to 1-tailed, divide the Sig (p- values) by half… (e.g., for “ad”, Sig p-value is .127 ÷ 2 = .064) Note: You have to calculate this manually for all the p-values (Sig. column). Note down the new scores beside the table. Once you have the scores recorded, we can now interpret your results. 196
This is the formula used to examine Note: Remember to use the p-values that whether your hypotheses are you calculated for the “supported” or “not supported”. A 1-tailed test. more stringent test is to use the threshold of p < .05. Significance †significant at p < .10 (t > 1.282) Levels *significant at p < .05 (t > 1.645) **significant at p < .01 (t > 2.326) ***significant at p < .001 (t > 3.090) Note: The t-values can be used to “double check” or confirm whether you assessed the p-values correctly or not (i.e., number of stars* & sword †). 197
Consistency: p-values & t-values Did you realize… the actual p-value stats are not reported: they are replaced with the Stars* & Sword †. †significant at p < .10 (t > 1.282) *significant at p < .05 (t > 1.645) **significant at p < .01 (t > 2.326) ***significant at p < .001 (t > 3.090) Note: Make sure the p-values (number of starts & sword) are consistent with the cutoffs for their respective t-values. 198
Interpretation: Star, Sig Level, & Direction Note: If you can rationalize your Interpreting your Regression Results “Refuted” hypo… this may turn 1. Look for Hypo variables with Stars* or †… out to be your most important this shows that the coefficient is significant, but may not necessarily be supported. Hypothesized finding (your hidden Cherry!) variables without any starts* or † are indeed not significant (= Not Supported) 2. Look at Sig Levels for Hypo variables… three stars*** indicate the strongest significant relationship. But bare in mind, regardless of the number of stars you received (one* two** or †), the result is still significant. 3. Look at the Direction of Sig Coefficients… this will determine whether your hypos are Supported or Not (direct & moderating hypos): If you hypothesized a +ve relationship & the “sig” result is +ve (= Supported) If you hypothesized a –ve relationship & the “sig” result is –ve (= Supported) If you hypothesized a +ve (or –ve) relationship but the “sig” result is –ve (or +ve) (= Not Supported)… also termed as “Refuted”. 199
Table: Correlation Matrix (SPSS) This is how you should report your Correlation Matrix Table. Also include your Mean scores and Standard Deviation scores in this Table (done manually)… refer to your syntax output. 200
Search
Read the Text Version
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
- 100
- 101
- 102
- 103
- 104
- 105
- 106
- 107
- 108
- 109
- 110
- 111
- 112
- 113
- 114
- 115
- 116
- 117
- 118
- 119
- 120
- 121
- 122
- 123
- 124
- 125
- 126
- 127
- 128
- 129
- 130
- 131
- 132
- 133
- 134
- 135
- 136
- 137
- 138
- 139
- 140
- 141
- 142
- 143
- 144
- 145
- 146
- 147
- 148
- 149
- 150
- 151
- 152
- 153
- 154
- 155
- 156
- 157
- 158
- 159
- 160
- 161
- 162
- 163
- 164
- 165
- 166
- 167
- 168
- 169
- 170
- 171
- 172
- 173
- 174
- 175
- 176
- 177
- 178
- 179
- 180
- 181
- 182
- 183
- 184
- 185
- 186
- 187
- 188
- 189
- 190
- 191
- 192
- 193
- 194
- 195
- 196
- 197
- 198
- 199
- 200
- 201
- 202
- 203
- 204
- 205
- 206
- 207
- 208
- 209
- 210
- 211
- 212
- 213
- 214
- 215
- 216
- 217
- 218
- 219
- 220
- 221
- 222
- 223
- 224
- 225
- 226
- 227
- 228
- 229
- 230
- 231
- 232
- 233
- 234