Week-10

COURSE ON Advanced

RESEARCH METHODOLOGY
(MSBA/MBA, FACULTY OF MANAGEMENT SCIENCES)
PROFESSOR

Dr.Alyas Qadeer Tahir E-Mail: [email protected]

Variable Types:

Qualitative (categorical/Attributes) 1* Data that refers only to

name classification (done using numbers) 2* Can be placed into distinct categories according to some characteristic or attribute.

Nominal Data (cant be rank) Gender, race, citizenship. etc

Use code numbers (1, 2,)

Ordinal Data (can be rank) Feeling (dislike like), color (dark bright) , etc.

Types of Data

Quantitative
(Ratio/Scale/Numerical) 1* Data that represent counts or measurements (can be count or measure) 2* Are numerical in nature and can be ordered or ranked.

Discrete Variables Assume values that can be counted and finite Ex : no of something

Continuous variables Can assume all values between any two specific values & it obtained by measuring Ex: weight, age, salary, height, temperature, etc.

 Exploring Relationship
(Finding out the strength of relationship between variables) Partial Correlation Multiple Regression Correlation (Pearson/Spearman) Logistic Regression Factor Analysis

Partial Correlation
Example of Research Question: Is there a relationship between the level of satisfaction of employees and the amount of production of a industry? OR Is a industry with higher level of satisfaction of its employees experience more production? OR Is a industry with lower level of satisfaction of its employees experience less production?

Conti. Partial Correlation

What you need: Two variables, both continuous, or one continuous and the other dichotomous (e.g, sex M/F). What it does: This clarifies the relationship between two continuous variables, in terms of both the strength of the relationship and the direction. Major Assumptions: 1. Normality 2. Linearity

Multiple Regression
Example of Research Question: How well do the two measures of variables (level of satisfaction and salary) predict the amount of production of a industry? OR How much variance in a factory production measure can be explained by the measures on the level of employees satisfaction and their salaries.? OR Which is the best predictor of a factory production: the employees level of satisfaction or their salaries?

Conti. Partial Correlation

What you need: One continuous dependent variable (factory production) Two or more continuous variables (satisfaction score and salaries). You can also use dichotomous independent variables (e.g, sex, M/F). What it does: This tells us how much of the variance in you dependent variable can be explained by our independent variables. Major Assumptions: 1. Normality 2. Linearity

Exploring Differences between Groups

(Whether statistically significant difference exists among a number of groups)

Non-Parametric Test Chi-square for independence Parametric Tests T-tests ANOVA MANOVA ANCOVA

Chi-square for independence (comparing groups)

Research Questions: Are males more likely to drop out of therapy than females? Is the proportion of permanent staff that recommends the organization as a good place to work significantly different from the proportion of casual staff that recommend? This test is used to explore relationship between two categorical variables. Each of these variable can have again two or more categories. This test compares the observed frequencies or population of cases that occur in each of the categories, with the values that would be expected if there was no association between the two variables being measured. It is based on cross-tabulation, with cases classified according to the categories in each of variable. (e.g. male/female; dropout/clear etc.)

For Chi-Square test for independence From our file: staffsurvey3Ed.sav We need: Two categorical variables; Employees status (permanent/casual) Recommendation (no/yes)
Assumptions: The lowest expected frequency in 80% cells should be more than 5.

T-Tests
Independent-sample t-test: used to compare the mean scores of two different groups of peoples or conditions. Paired-sample t-test: used to compare the means scores for the same group of people on two different occasions. Research Question: Is there a significant difference in the mean total scores for permanent and casual staff. We need: Two variables One categorical, independent variable (per./cas) One continuous, dependent variable (total satis. score)

ANOVA (One-way Analysis of Variance)

In T-Tests, we compare mean scores of two different groups. For comparing mean scores of more than two groups, we use ANOVA.
ANOVA involves one independent variable (called as a factor), which has a number of different levels. These levels correspond to the different groups or conditions. In ANOVA, we compare the variance (variability) in scores between the different groups with the variability within each of the groups. For ANOVA, we calculate F ratio (variance between the groups divided by variance within the groups). A large F ratio indicates that there is more variability between the groups than there is within groups. A significant F test indicates that we can reject the hypothesis, which states that the population means are equal. It does not, however tells us which of the groups differ. For this , we conduct Post-hoc analysis. There are two types of One-way of ANOVA. 1.Between groups ANOVA (independent group design). 2.Repeated-measures ANOVA (within-subject design).

ONE-WAY ANOVA
Research Question: Is there a significant difference in total scores of employees having service of one year or more, 3 to 5 years and 6 years or more? We need: Two variables One categorical (factor), independent variable with three or more distinct categories or a continuous variable with three groups One continuous, dependent variable (totalsatis. score), Assumption: Normality

TWO-WAY ANOVA
In 2-Way ANOVA between groups, we explore two-way between groups analysis of variance. There are two independent variables and there are different peoples in each of the groups This techniques allowed us to look at the individual and joint effect of two independent variables on one dependent variable. The advantage of using a two-way design is that we can test the main effect for each independent variable and also explore the possibility of an interaction effect. An interaction effect occurs when the effect of one independent variable on the dependent variable depends on the level of a second independent variable.

Two-WAY ANOVA
Research Question: What is the impact of service groups and service status on staff satisfaction? Does service status moderate the relationship between service group and staff satisfaction? OR Conduct a two-way ANOVA with post-hoc tests (if appropriate) to compare staff satisfaction scores (totsatis) across each of the length of service categories (servicegp3) for permanent versus casual staff (employeestatus). We need: Three variables Two categorical independent variable; servicegp3 and employeestatus One continuous, dependent variable (totalsatis score) Assumption: The variance should be homogeneous (Apply Levenes Test).

MULTIVARIATE ANALYSIS OF VARIANCE

(MANOVA)
MANVOA is an extension of ANOVA for use when we have more than one dependent variable. MANOVA compares the groups that tells whether the mean differences between the groups on the combination of dependent variables are likely to have occurred by chance. There is way to conduct ANOVAs separately for each dependent variable but in such case there is chance of occurring Type 1 error . Therefore, it is not recommended. MANOVA can be used in one-way, two-way and higher-order factorial designs (with multiple independent variables) and when using ANCOVA. What MANOVA does: Compares two or more groups in terms of their means on a group of dependent variables. Tests the null hypothesis that the population means on a set of dependent variables do not vary across different levels of a factor or grouping variable.

For MANOVA Example We use data File: survey3ED.sav In this example, the difference between males and females on a number of measures of wellbeing is explored. These include a measure of negative mood (-ive Affect scale), positive mood (+ive affect scale) and perceived stress (total Perceived Stress scale). Research Question: Are males better adjusted than females in terms of their positive and negative mood states and levels of perceived stress? Do males and females differ in terms of overall wellbeing? We need: One-way MANOVA One categorical independent variable (e.g, sex) and Two or more continuous, dependent variable (-ive affect, +ive affect, perceived stress).

Assumptions: MANOVA has following assumptions.

1.Sample size (Minimum dependent variables 3, more cells are preferred) 2.Normality (a sample size of 20 in each cell should ensure normality) 3.Outliers (data points or scores that are different from the reminder of the scores, we need to check this) 4.Linearity (presence of a straight-line relationship between each pair of dependent variables) 5.Homogenearity of regression 6.Multicollinearity and singularity 7.Homogeneity of variance-covariance matrices

ANALYSIS OF COVARIANCE
(ANCOVA)
ANCOVA is an extension of ANOVA. This allows to explore differences between groups while statistically controlling for an additional variable. This additional variable is called covariate. This variable we may suspect influencing scores on the dependent variable. ANCOVA can be used as part of one-way, two-way and multivariate ANOA techniques. USES of ANCOVA: ANCOVA can be used when we have a two different interventions pre-test/post or post-test/delayed post-design. That is for comparing the impact of two different interventions taking before and after measures for each group. The scores on the pre-test are treated as a covariate to control for pre-existing differences between the groups. This makes ANCOVA very useful in situations when we have very small sample size and only small or medium effect size. ANCOVA is also handy when we are unable to randomly assign our subjects to the different groups but instead have had to use existing groups (classes of students etc.)

Assumptions:
ANCOVA has following assumptions.
(In addition to the one used for one-way ANOVA)

1. Influence of treatment on covariate measurement 2. Reliability of covariates 3. Correlations among covariates 4. Linear relationship between dependent variable and covariate 5. Homogeneity of regression slopes

ONE-WAY ANCOVA
File Name: experim3ED In this example, we will explore the impact of the math.skills class (Group 1) and the confidence building class (Group 2) on students scores on the Fear of Statistic Test, while controlling for the scores on this test administered before the program (Pre-test, covariate). Research Question: Is there difference in the Fear of Statistics Test scores for the math.skills (group 1) and the confidence building group (group 2), while controlling for their pre-test scores on this test?

We Need:

Three Variables (Note: Here we are going for Time2)

One categorical independent variable with two or more levels (group 1/ group 2); One continuous dependent variable (score on the Statistics Test at Time 2 i.e post-test); One or more continuous covariates (scores on the Fear statistics Test at Time 1).

What it does: ANCOVA will tell us if the mean Fear statistics Test
scores at Time 2 (post-test) for the two groups are significantly different after initial pre-test scores are controlled for. Assumptions: All normal one-way ANOVA assumptions apply in ANCOVA. Additional ANCOVA assumptions are as under: 1. The covariate is measured prior to the interventions or experimental manipulation; 2. The covariate is measured without error (or as reliably as possible);

3. 4.

The covariates are not strongly correlated with one another; There is a linear relationship between the dependent variable and the covariate for all groups (linearity); 5. The relationship between the covariate and dependent variable is the same for each of the groups (homogeneity of regression slopes). Testing Assumptions: (Only additional one)
Assumption 1: measurement of the covariate Yes, covariate (pre-test) is measured before the experiment. Assumption 2: Reliability of the covariate Yes, the reliability Cronbach alpha in this case is .78. Assumption 3: Correlation among the covariates Here, we have only one covariate, therefore need not to do. Assumption 4: To check the assumption of a linear relationship between the dependent variable and the covariates for all our groups, we proceed as:

Two-WAY ANCOVA Introduction

In one-way ANCOVA, we were interesting to determine that which intervention (maths skills or confidence building) was more effective in reducing students fear of statistics. We found no significant difference between the groups. Now, suppose that in reading further in the literature on the topic, we found some research that suggested that there might be a difference in how males and females respond to different interventions.

Cont from two-way ANCOVA

Therefore, in literature you will see this additional variable (e.g, sex) described as moderator. That is, it moderates or influences the effect of the other independent variable. Often, these moderator variables are individual difference variables, characteristics of individuals that influence the way in which they respond to an experimental manipulation or treatment condition. Important: If you obtain a non-significant result for your one-way ANCOVA, then you need to consider for a moderator variable and go for a 2-way ANCOVA.

Two-WAY ANCOVA
File Name: experim3ED Research Question: Does gender influence the effectiveness of two programs designed to reduce participants fear of statistics? OR Is there a difference in post-intervention Fear of Statistics Test scores between males and females in their response to a math.skills program and a confidence building program.
We Need: Four variables: 1. Two categorical independent variables with two or more levels (sex: M/F, group: math.skills/confidence building). One continuous dependent variable (score on the Statistics Test at Time 2 i.e post-test); One or more continuous covariates (scores on the Fear statistics Test at Time 1).

2.
3.

What it does:
ANCOVA will control for scores on our covariate (s) and then perform a normal two-way ANOVA. This will tell us if there is : a significant main effect for our first independent variable (group). a main effect for our second independent variable (sex); a significant interaction between the two. Assumptions: All normal two-way ANOVA assumptions apply (normality, homogeneity of variance). These should be checked first. Additional ANCOVA assumptions: Same as discussed under one-way ANCOVA. These should be checked first.