Management Research Methods 2
(MRM 2)
Pre-master Business Administration
UvA
Block 2
Grade 8,7
Lecturer: Abhishek Nayak
,Lecture 1: Conceptual models & Analysis of Variance (ANOVA) 3
Lecture 2: Moderation in ANOVA 13
Lecture 3: Regression (basics) 21
Lecture 4 Regression (advanced) & mediation 33
Lecture 5: Logistic regression 50
Lecture 6: Factor Analysis 57
2
,Lecture 1: Conceptual models & Analysis of Variance (ANOVA)
Some Terminology
- Outcome Variable (OV)
What you are testing e.g. what affects the sales. Sales is dependent variable or outcome variable.
E.g. Nr. Of sales> positive effect/ nr of products> Positive effect/ price of the unit> negative effect
price has negative effect on quanitty/ advertising $ spent> positive/ nr. Of competitors> negative
These examples are outcomes of sales and are the independent variables (IV) or predictors (PV)
o Or DV = Dependent Variable
§ Test variable, variable to be explained
- Predictor Variable (PV)
o Or IV = Independent Variable
§ Variable that explains
ANOVA only has 1 IV and 1 DV
Regression has multiple.
- The p-value
o Stands for the Probability of obtaining a result (or test-statistic value) equal to (or ‘more extreme’ than) what was
actually observed (the result you actually got), assuming that the null hypothesis is true
o P< 0,005 à Null Hypothesis rejected à Alternative Hypothesis is accepted
o Why 0,05 because we deal with 0,95
o 99% means there is more confidence
- A low p value indicates that the null hypothesis (H0) is unlikely
Conceptual models
- Visual representations of relations between theoretical constructs (and variables) of interest
- In research: by “model” we mean a simplified description of reality
- OV = outcome variable = dependent variable
- PV = predictive variable = independent variable
- Variables can have different measurement scales:
o Categorical (nominal, ordinal) – subgroups are indicated by numbers;
o Quantitative (discrete, interval, ratio) – we use numerical scales, with equal distances between values.
§ In social sciences we sometimes treat ordinal scales as (pseudo) interval scales, e.g. Likert scales.
Quantitative:
Price, length, age, numbers very unlikely – very likely.
Important to know whether your dependent variable is quantitative or something else!
You can do a count
Categorical:
There is no number in it
Male/Female 0 1 or 1 0 – price 1 to ..(categories 0, 1..) Because there are only two levels
Income (Low, medium, high 0 1 2 or 1 2 3..)
You can not do a count here
Depending on what your variable is, the technique is completely different.
For most techniques DV or OV needs to be Quantitative!
The PV or IV for ANOVA needs to be categorical because you compare groups. (you compare 2 or 3 groups on a certain
quantitative variable: price, emotional intelligence) therefore PV or IV needs to be categorical.
For Regression, the PV could be continuous as well, how two variables affect each other.
- Consider the following situation
- Research question: What factors influence student satisfaction?
3
, o Commitment of teacher
o Course content
o …
RQ: What factors determine students satisfaction?
H0: Teachers that are more committed will not affect/ there is no effect between .. and .. (no effect).
H1: Teachers that are more committed will increase the satisfaction level of students (in comparison to uncommitted teachers).
Conceptual models – Moderation
- What if our proposed effect is stronger in certain settings?
H2: Teachers that are more committed will increase the satisfaction level of students (in
comparison to uncommitted teachers), when they have good communication skills.
- “Communication skills” is a moderating variable à one variable moderates the
relationship between two other variables.
- More in week on interaction in Factorial ANOVA (Moderation = Interaction)
Conceptual models – Moderation
- What if the proposed relationship “goes via” another variable?
H3 =The positive effect of teacher’s commitment on student satisfaction
is mediated by quality of the course material
- “Lecture slides quality” is a mediating variable à one variable mediates the
relationship between two other variables.
Moderation and mediation
A affects B and is moderated by C
A leads to B and B leads to C is mediation
4
