Organisational Research
• Development of knowledge regarding range of organisational issues
• Carried our by academics to solve problems, describe environment, make predictions and
explain events
• Done through a range of quantitative methods
Data Collection Methods
• Informal method (Google, Wikipedia, journals…)
• Formal methods (Interviews, focus groups, observation, surveys, experiments or other
secondary data analysis)
Quantitative vs Qualitative Research
• Quantitative emphasises quantification in the collection of data and seeks to directly measure
the social world, assuming social reality is an objective reality. It is a deductive approach
• Qualitative emphasises words rather than quantification in the collection of data and it is
interested in the different interpretation that individual have of their social world. This approach
assumes social reality is a constantly changing environment and it is the individuals creation. It
is an inductive approach.
Induction and Deduction
• A theory attempts to formulate an explanation about some facet of reality
• Induction: the theory is formulated by first collecting data and then using data - qualitative
approach
• Deduction: the theory/idea comes first and then after you go collect data to compare it with the
theory previously formulated
Main assumptions of Quantitative Research Methods
• Objective and measurable reality
• Patterns, differences and relationships can be represented by numbers
• Causative statements are possible and they develop predictions
• Findings of good research can be generalised
‘Good’ qualitative research
• Based upon a theory
• Hypotheses developed which is testable and replicable
• Appropriate research design adopted
• Suitable data collected
• Sufficient and relevant samples of observation
• Data collection has validity and reliability
• Appropriate analysis of the data
• Statistical tests selected and conducted properly
• Implications drawn from findings
Different quantitative research questions
• Quantification/Description: Univariate statistics (eg What will UK’s GDP be in 2019?
• Difference: Bivariate analysis (eg Do men earn more than women?)
• Association: Bivariate or multi-variate statistics (eg Is a company’s marketing budget related to
product sales?)
,Week 2: Significance and Chi-square
Hypotheses
• Casual statements about the objective world around us
- e.g. Men earn more than women
• Null hypothesis (H0) v Alternative hypothesis (H1)
• Alternative hypothesis is the initial theory
• Null hypothesis is the opposite of the alternative hypothesis
- (H0): Men do NOT earn more than women
- (H1) Men earn more than women
• Is there sufficient evidence to reject the null hypothesis?
Probability
• Probability = Chance = Likelihood
• p= expression of probability
- p=1 : same as 100% probability
- p=0 : same as 0% probability
- A probability of 1 in 2 (eg coin toss) is 50% chance hence expressed as a p-value of 0.5
where p= 0.5
- A probability of 1 in 20 (very unlikely) os 5% chance hence expressed as a p-value of 0.05
where p= 0.05
Statistical Significance
• Statistical significance is the probability of finding a given deviation from the null hypothesis - or
a more extreme one - in a sample.
• It is often referred to as the p-value or simply p
• A small p-value means that your data are unlikely under some null hypothesis. A somewhat
arbitrary convention is to reject the null hypothesis if p<0.05
• Statistical significance is built on a few simple ideas: hypothesis testing, the normal distribution
and p values.
,Statistical Significance - key stuff
1. Significance tells us whether
- A particular finding is likely to be due to random chance alone
- A sample statistic is likely to reflect a genuine difference or association (eg greater than 0) in
the wider population
- We can reject or retain our null hypothesis
2. A p value of 0.05 or below indicates that a finding is significant (eg it is the threshold)
- p<0.05 means less than 5% likelihood that the finding is due to random chance
- Lower p-values = greater significance
3. To calculate p-values, we need inferential tests
Inferential tests
• Investigation of more than one variable (bivariate or multivariate)
• They examine the significance of differences and associations i.e. they produce p-values
• Many inferential tests exist and the choice defends on whether you are examining associations
or differences and on the type of data you have (i.e. categorical or continuous)
Types of inferential tests
1. Tests of association/dependency
- Chi-square (two categorical variables)
- Correlation (two continuous variables)
- Regression (multiple independent variables and single dependent variables
A. Multiple regression (continuous dependent variable)
B. Logistic regression (dichotomous dependent variable, which means that contains
precisely two distinct values)
2. Tests of difference
- T-test (dichotomous IV, continuous DV)
- ANOVA (polychotomous IV, continuous DV)
Chi-square test
• Test which examines the dependency or association between categorical data (i.e. nominal,
ordinal)
- SO, a Chi-square test indicates whether an observed association between two categorical
variables within a sample is likely to be due to random chance or reflective of a real
association within the population
- Examples of hypothesis involving two lots of categorical data (“tea tastes different if milk
goes in first” where DV is milk first and IV is the taste)
, Gender and promotion example (1)
• Sample of 200 employees (100 men and 100 women)
• After 2 years,100 had been promoted and 100 hadn’t
• Interest in whether gender influences likelihood of promotion
- Univariate frequencies:
- Bivariate contingency table
- Distribution within this sample data where everything is equal
- In this Chi-square there is no apparent relationship and this results are what we would expect
to get if the null hypothesis were true
Contingency tables
• Also called cross tabulation or cross tabs
• Compares distributions of frequencies (counts) within particular categorical conditions
• Cross tabs can display a relationship between two or more categorical variables
- If two dichotomous variable = 2x2 table
- A dichotomous variables and a trichotomous variable = 2x3 table
• Development of knowledge regarding range of organisational issues
• Carried our by academics to solve problems, describe environment, make predictions and
explain events
• Done through a range of quantitative methods
Data Collection Methods
• Informal method (Google, Wikipedia, journals…)
• Formal methods (Interviews, focus groups, observation, surveys, experiments or other
secondary data analysis)
Quantitative vs Qualitative Research
• Quantitative emphasises quantification in the collection of data and seeks to directly measure
the social world, assuming social reality is an objective reality. It is a deductive approach
• Qualitative emphasises words rather than quantification in the collection of data and it is
interested in the different interpretation that individual have of their social world. This approach
assumes social reality is a constantly changing environment and it is the individuals creation. It
is an inductive approach.
Induction and Deduction
• A theory attempts to formulate an explanation about some facet of reality
• Induction: the theory is formulated by first collecting data and then using data - qualitative
approach
• Deduction: the theory/idea comes first and then after you go collect data to compare it with the
theory previously formulated
Main assumptions of Quantitative Research Methods
• Objective and measurable reality
• Patterns, differences and relationships can be represented by numbers
• Causative statements are possible and they develop predictions
• Findings of good research can be generalised
‘Good’ qualitative research
• Based upon a theory
• Hypotheses developed which is testable and replicable
• Appropriate research design adopted
• Suitable data collected
• Sufficient and relevant samples of observation
• Data collection has validity and reliability
• Appropriate analysis of the data
• Statistical tests selected and conducted properly
• Implications drawn from findings
Different quantitative research questions
• Quantification/Description: Univariate statistics (eg What will UK’s GDP be in 2019?
• Difference: Bivariate analysis (eg Do men earn more than women?)
• Association: Bivariate or multi-variate statistics (eg Is a company’s marketing budget related to
product sales?)
,Week 2: Significance and Chi-square
Hypotheses
• Casual statements about the objective world around us
- e.g. Men earn more than women
• Null hypothesis (H0) v Alternative hypothesis (H1)
• Alternative hypothesis is the initial theory
• Null hypothesis is the opposite of the alternative hypothesis
- (H0): Men do NOT earn more than women
- (H1) Men earn more than women
• Is there sufficient evidence to reject the null hypothesis?
Probability
• Probability = Chance = Likelihood
• p= expression of probability
- p=1 : same as 100% probability
- p=0 : same as 0% probability
- A probability of 1 in 2 (eg coin toss) is 50% chance hence expressed as a p-value of 0.5
where p= 0.5
- A probability of 1 in 20 (very unlikely) os 5% chance hence expressed as a p-value of 0.05
where p= 0.05
Statistical Significance
• Statistical significance is the probability of finding a given deviation from the null hypothesis - or
a more extreme one - in a sample.
• It is often referred to as the p-value or simply p
• A small p-value means that your data are unlikely under some null hypothesis. A somewhat
arbitrary convention is to reject the null hypothesis if p<0.05
• Statistical significance is built on a few simple ideas: hypothesis testing, the normal distribution
and p values.
,Statistical Significance - key stuff
1. Significance tells us whether
- A particular finding is likely to be due to random chance alone
- A sample statistic is likely to reflect a genuine difference or association (eg greater than 0) in
the wider population
- We can reject or retain our null hypothesis
2. A p value of 0.05 or below indicates that a finding is significant (eg it is the threshold)
- p<0.05 means less than 5% likelihood that the finding is due to random chance
- Lower p-values = greater significance
3. To calculate p-values, we need inferential tests
Inferential tests
• Investigation of more than one variable (bivariate or multivariate)
• They examine the significance of differences and associations i.e. they produce p-values
• Many inferential tests exist and the choice defends on whether you are examining associations
or differences and on the type of data you have (i.e. categorical or continuous)
Types of inferential tests
1. Tests of association/dependency
- Chi-square (two categorical variables)
- Correlation (two continuous variables)
- Regression (multiple independent variables and single dependent variables
A. Multiple regression (continuous dependent variable)
B. Logistic regression (dichotomous dependent variable, which means that contains
precisely two distinct values)
2. Tests of difference
- T-test (dichotomous IV, continuous DV)
- ANOVA (polychotomous IV, continuous DV)
Chi-square test
• Test which examines the dependency or association between categorical data (i.e. nominal,
ordinal)
- SO, a Chi-square test indicates whether an observed association between two categorical
variables within a sample is likely to be due to random chance or reflective of a real
association within the population
- Examples of hypothesis involving two lots of categorical data (“tea tastes different if milk
goes in first” where DV is milk first and IV is the taste)
, Gender and promotion example (1)
• Sample of 200 employees (100 men and 100 women)
• After 2 years,100 had been promoted and 100 hadn’t
• Interest in whether gender influences likelihood of promotion
- Univariate frequencies:
- Bivariate contingency table
- Distribution within this sample data where everything is equal
- In this Chi-square there is no apparent relationship and this results are what we would expect
to get if the null hypothesis were true
Contingency tables
• Also called cross tabulation or cross tabs
• Compares distributions of frequencies (counts) within particular categorical conditions
• Cross tabs can display a relationship between two or more categorical variables
- If two dichotomous variable = 2x2 table
- A dichotomous variables and a trichotomous variable = 2x3 table
