Final Exam MKTG2001
Week 7
Sampling
Role in research process
Def: Selecting a small number of participants from a larger group and generalising the result.
Used when it’s unreasonable to do a census.
Less time-consuming and cheap than census
Plays an indirect role in designing questionnaires
Distinguish between non-probability and probability sampling
Probability:
Each sampling unit has the same probability of being picked.
Unbiased sample.
Proper sample representation.
Generalisable results.
Simple Random sampling:
Every sampling unit has equal chance
Systematic Random
Organised into a list and skipping a certain number every time.
Stratified sampling:
Separates the target population into groups and uses random sampling to create a new single sample
Cluster sampling:
Divides groups into clusters, then uses random sampling.
Non-Probability
Probability of selecting sampling unit is unknown
Sampling error is not known
Selection based on researcher and may/may not be representative of target population
Convenience:
Draws sample based on convenience of researcher
Judgement/purposive sampling:
Selected because they fit a criteria set by researchers.
Quota:
Selection based on pre specified quotas for: eg demographics, behaviours, attitudes.
Snowball:
Participants are selected by researchers and then they refer people for the study.
Explain the factors determining sample size
Based on:
Research objective: eg qualitative/quantitative
Degree of accuracy: Insights or inferences
Resources: Budget?
Time frame: When does it need to be done by?
Knowledge of target population: Are there lists? How easy is it to develop a sampling frame?
Scope of research: local, international etc
Statistical Analysis needed: Hypothesis testing or projection
Probability:
Population standard deviation
Level of Confidence desired
Degree of precision (acceptable amount of error)
Non-probability:
Not be used for statistical inferences
Determined by budget and previous studies
Made on initiative decision of researchers
Sampling plan steps: ensure data represents population
Step 1: Define target population
Step 2: Select data collection method
Step 3: Identify sampling frames needed
Step 4: Select appropriate sampling methods
Step 5: Determine necessary sample sizes
Step 6: Creating a plan for selecting sampling units
Step 7: Execute this plan
Scales/Measurement
Role of measurement in marketing research
Applies to abstract things such as people’s preferences and accurate measurements are essential to effective decision making.
Measurement process consists of two steps
Construct selection/development
Determine what specific data solves research problem, pick out most relevant objects
Done by sorting out objective and subjective properties of each object
Scale measurement (how to measure each construct)
Assigning set of descriptors to represent possible responses
Combination of labels that determine degree of intensity (SCALE POINTS) eg strongly agree/disagree or 1-7
Explain the four basic levels of scales
Nominal
Uses labels to classify into groups
Frequency distributions and mode
Ordinal
Ranked via preference
Mean, mode, frequency distributions and scales can be applied.
Interval
Demonstrates absolute differences between each scale point
Means and standard deviation
Ratio
Allows researcher to identify absolute differences but also make comparisons
Enables true natural zero
Means and standard deviation
Describe scale development and its importance in gathering primary data
Designing a scale requires
Understanding research problem
Identifying / developing constructs
Establishing detailed data requirements
Understanding scale properties
Selecting appropriate measure scale
discriminatory power of scale descriptors is the scale’s ability to differentiate between the scale responses (more scale points = greater disciminiatory power)
Balanced scale = just as many positive and negative responses
Forced choice = no neutral descriptor
3 main scales to use
Likert Strongly Agree/Disagree
Semantic differential scale is a unique bipolar ordinal scale format that captures a person’s attitudes or feelings on a given object
Behavioural intention probability/predictability
Discuss comparative and noncomparative scales.
Comparative
Discusses feelings/attitudes on the basis of another object
Rank order scales
Constant sum scales
Non-comparative
Discusses feelings of a singular object
Week 8
Data prep/analysis/results
Describe process for data preparation and analysis
Accurate data that helps with decision-making
Step 1: Validation
Surveys were done correctly and without bias
Measurement bias
Respondent bias
Researcher bias
Step 2: Editing
Raw data is checked for mistakes
4 main concerns
Asking proper questions
Accurate recording of responses
Correct screening of respondents
Recording of open ended questions
Step 3: Coding
Involves grouping and assigning values to various responses
4 step method
Generate list of as many responses as possible
Consolidate responses
Assign a numerical value as a code
Assign a coded value to each response
Step 4: Data entry
Entering data into a file for analysis
Error detection identifies errors from entry
Errors can be identified through error edit routines
Missing Values
Data tabulation
One way
Shows responses for a single variable
Descriptive statistic (Summary)
Can spot missing values
Two way tab
Two or more variables
Most commonly used with ordinal and nominal data
Descriptive stat
Week 9
Basic Data Analysis for Quantitative data
Explain basic tendencies and dispersion
Central tendencies
Describes a set of data by identifying the central position
Mean (average) : Best for Interval/ratio variable (NOT SKEWED)
Median (Middle number) : Best for ordinal (preference) and Interval/ratio skewed
Mode (Most frequent) : Best for nominal (categorical)
Dispersion
Describes extent of variability
Range, Interquartile range, Standard Deviation, Variance
Standard deviation (average distance of distribution variables from the mean) for symmetrical numerical data (mean is middle)
Ordinal or Skewed data median and interquartile range
Describe how to test hypotheses using univariate and bivariate statistics
Form hypothesis regarding population characteristics. First you do frequency distributions and average etc. Then you test hypothesis
If you test one variable at the time it's a univariate test. You test the variable against for example a mean.
Null Hypothesis (H0 ): The null hypothesis is no difference in the group means
Or Alternate Hypothesis (H1 ): Opposite of null hypothesis
Significance level normally set at 0.05 (confidence level = 95%)
Bivariate tests: Testing two variables, how one variable influences another variable. (each sample is independent)
Cross Tabs
A frequency distribution of responses on two or more sets of variables
To conduct cross tabulation, the responses for each of the groups are tabulated and compared
Chi-square test
Analyses nominal or ordinal data
Enables analysis of statistical significance between frequency distributions between 2 or more variables
How actual frequencies fit expected frequencies
Null = no difference
Alternative difference
If level of confidence is lower than 0.05 we can reject null
Only larger sample sizes
Independent t-test
Statistical difference between 2 means/interventions/change scores
Dependent variable that is continuous eg interval/ratio
Independent variable that is categorical eg gender
Assumption: Normal distribution, no outliers (check with box plot), random sample, no relationship between samples.
Paired T-test
Determines if difference between means is 0
Null hypothesis true mean difference = zero
Assumptions
dependent variable is continuous (interval/ratio), normally distributed and no outliers
Variables have to be independent of one another.
ANOVA
Statistical difference between 3 or more means
One-way anova the comparison involves means but with only one independent variable
F-test tells us if tests are statistically significant
Total variance = set of responses is made up between group and within group variance
between-group variance measures how much the sample means of the groups differ from one another
within-group variance measures how much responses within each group differ from one another.
Does not identify which pairs of means are different
N-way Anova
Analyse several independent variables at the same time
Multiple independent variables act together to affect the dependent variable group means (interaction effect)
Used for experimental design
Perceptual mapping
Graphic representation from analysis
Perception based of other brands in comparison to key attributes
Parametric tests
Normal distribution
Interval/ratio data
Results affected outliers
More statistical power
T-Test, Anova, Regression,
Week 10
Evaluate the types of relationships between variables
Strength of association (weak, moderate, strong)
Direction (positive or negative)
Presence (systematic and consistency)
Type of relationship (Linear and nonlinear)
Linear: Changes in one variable changes the second variable
Curvilinear: Increases to a certain extent then starts to decrease (curve)
Explain the concepts of association and covariation
Association: Numerical measure of strength between relationships
Covariation: Amount of change in one variable that consistently changes in another variable.
Discuss the differences between Pearson correlation and Spearman correlation
Pearson correlation:
Statistical measure of strength between two metric variables.
Varies between -1 and 1 (0 is no correlation)
Null hypothesis : is no association
0.81-1 = strong
0.00-0.20 = weak
May be a non-consistent (nonlinear) relationship
Significance level must be calculated: might falsely reject null
Interval/ratio, normally distributed, linear relationship
Spearman rank order
Two variables used ordinal scales
Either one is ordinal
Explain the concept of statistical vs. practical significance
Statistical:
Statistical significance is used to examine the coefficient for multiple regression
Not all of them are statistically significant
Because some of the procedures involved in determining the statistical significance of a statistical test include consideration of the sample size, it is possible to have a very low degree of association between two variables show up as statistically significant
However, by considering the absolute strength of the relationship in addition to its statistical significance, the researcher is better able to draw the appropriate conclusion about the data and the population from which they were selected.
Explain when and how to use regression analysis
Estimates relationship between one dependent variable and one or more independent variables
Which of the variables have an impact?
Most important variables
How they interact with each other
How certain are we about the variables
Any point not on the line is unexplained variance
Bivariate analysis focuses on one dependent and one independent variable
Multiple regression analysis has multiple independent variables
Each variable has a separate regression coefficient is calculated which describes its relationship with the dependent variable
Linear relationship, Homoscedasticity and normal distribution
Week 11
List the objectives of a research report
Clear concise interpretation (logical explanation, make complex information easy to understand) of project
Accurate, easy to understand, credible (create believability through good quality and organisation)
Guide future research and serve as an information source
Logical recommendations
Describe the format of a marketing research report
Title page
table of contents
Exec summary, introduction
Research Method and procedures (research design, sampling, sampling process, primary data and secondary data)
Data analysis and findings (the body, tables/figures/graphs,
conclusions and recommendations
Limitations
appendix
Discuss the techniques of graphically displaying research results
Frequencies: bar chart, pie chart or table
Several thematic related variables can be presented in same table
Bar charts are used to compare groups
ANOVA/t-tests be presented in tables (include correlations)
Regression analysis displays predictor and outcome with arrows
Address the problems encountered in preparing reports
Lack of data interpretation
Unnecessary use of complex statistics
Lack of relevance
Placing too much emphasis on few statistics
Appreciate the significance of presentations in marketing research
Helps to make good decisions
Seen by only those commissioning the report
Seniors rely on short, concise presentations
Final Exam MKTG2001
Week 7
Sampling
Role in research process
Def: Selecting a small number of participants from a larger group and generalising the result.
Used when it’s unreasonable to do a census.
Less time-consuming and cheap than census
Plays an indirect role in designing questionnaires
Distinguish between non-probability and probability sampling
Probability:
Each sampling unit has the same probability of being picked.
Unbiased sample.
Proper sample representation.
Generalisable results.
Simple Random sampling:
Every sampling unit has equal chance
Systematic Random
Organised into a list and skipping a certain number every time.
Stratified sampling:
Separates the target population into groups and uses random sampling to create a new single sample
Cluster sampling:
Divides groups into clusters, then uses random sampling.
Non-Probability
Probability of selecting sampling unit is unknown
Sampling error is not known
Selection based on researcher and may/may not be representative of target population
Convenience:
Draws sample based on convenience of researcher
Judgement/purposive sampling:
Selected because they fit a criteria set by researchers.
Quota:
Selection based on pre specified quotas for: eg demographics, behaviours, attitudes.
Snowball:
Participants are selected by researchers and then they refer people for the study.
Explain the factors determining sample size
Based on:
Research objective: eg qualitative/quantitative
Degree of accuracy: Insights or inferences
Resources: Budget?
Time frame: When does it need to be done by?
Knowledge of target population: Are there lists? How easy is it to develop a sampling frame?
Scope of research: local, international etc
Statistical Analysis needed: Hypothesis testing or projection
Probability:
Population standard deviation
Level of Confidence desired
Degree of precision (acceptable amount of error)
Non-probability:
Not be used for statistical inferences
Determined by budget and previous studies
Made on initiative decision of researchers
Sampling plan steps: ensure data represents population
Step 1: Define target population
Step 2: Select data collection method
Step 3: Identify sampling frames needed
Step 4: Select appropriate sampling methods
Step 5: Determine necessary sample sizes
Step 6: Creating a plan for selecting sampling units
Step 7: Execute this plan
Scales/Measurement
Role of measurement in marketing research
Applies to abstract things such as people’s preferences and accurate measurements are essential to effective decision making.
Measurement process consists of two steps
Construct selection/development
Determine what specific data solves research problem, pick out most relevant objects
Done by sorting out objective and subjective properties of each object
Scale measurement (how to measure each construct)
Assigning set of descriptors to represent possible responses
Combination of labels that determine degree of intensity (SCALE POINTS) eg strongly agree/disagree or 1-7
Explain the four basic levels of scales
Nominal
Uses labels to classify into groups
Frequency distributions and mode
Ordinal
Ranked via preference
Mean, mode, frequency distributions and scales can be applied.
Interval
Demonstrates absolute differences between each scale point
Means and standard deviation
Ratio
Allows researcher to identify absolute differences but also make comparisons
Enables true natural zero
Means and standard deviation
Describe scale development and its importance in gathering primary data
Designing a scale requires
Understanding research problem
Identifying / developing constructs
Establishing detailed data requirements
Understanding scale properties
Selecting appropriate measure scale
discriminatory power of scale descriptors is the scale’s ability to differentiate between the scale responses (more scale points = greater disciminiatory power)
Balanced scale = just as many positive and negative responses
Forced choice = no neutral descriptor
3 main scales to use
Likert Strongly Agree/Disagree
Semantic differential scale is a unique bipolar ordinal scale format that captures a person’s attitudes or feelings on a given object
Behavioural intention probability/predictability
Discuss comparative and noncomparative scales.
Comparative
Discusses feelings/attitudes on the basis of another object
Rank order scales
Constant sum scales
Non-comparative
Discusses feelings of a singular object
Week 8
Data prep/analysis/results
Describe process for data preparation and analysis
Accurate data that helps with decision-making
Step 1: Validation
Surveys were done correctly and without bias
Measurement bias
Respondent bias
Researcher bias
Step 2: Editing
Raw data is checked for mistakes
4 main concerns
Asking proper questions
Accurate recording of responses
Correct screening of respondents
Recording of open ended questions
Step 3: Coding
Involves grouping and assigning values to various responses
4 step method
Generate list of as many responses as possible
Consolidate responses
Assign a numerical value as a code
Assign a coded value to each response
Step 4: Data entry
Entering data into a file for analysis
Error detection identifies errors from entry
Errors can be identified through error edit routines
Missing Values
Data tabulation
One way
Shows responses for a single variable
Descriptive statistic (Summary)
Can spot missing values
Two way tab
Two or more variables
Most commonly used with ordinal and nominal data
Descriptive stat
Week 9
Basic Data Analysis for Quantitative data
Explain basic tendencies and dispersion
Central tendencies
Describes a set of data by identifying the central position
Mean (average) : Best for Interval/ratio variable (NOT SKEWED)
Median (Middle number) : Best for ordinal (preference) and Interval/ratio skewed
Mode (Most frequent) : Best for nominal (categorical)
Dispersion
Describes extent of variability
Range, Interquartile range, Standard Deviation, Variance
Standard deviation (average distance of distribution variables from the mean) for symmetrical numerical data (mean is middle)
Ordinal or Skewed data median and interquartile range
Describe how to test hypotheses using univariate and bivariate statistics
Form hypothesis regarding population characteristics. First you do frequency distributions and average etc. Then you test hypothesis
If you test one variable at the time it's a univariate test. You test the variable against for example a mean.
Null Hypothesis (H0 ): The null hypothesis is no difference in the group means
Or Alternate Hypothesis (H1 ): Opposite of null hypothesis
Significance level normally set at 0.05 (confidence level = 95%)
Bivariate tests: Testing two variables, how one variable influences another variable. (each sample is independent)
Cross Tabs
A frequency distribution of responses on two or more sets of variables
To conduct cross tabulation, the responses for each of the groups are tabulated and compared
Chi-square test
Analyses nominal or ordinal data
Enables analysis of statistical significance between frequency distributions between 2 or more variables
How actual frequencies fit expected frequencies
Null = no difference
Alternative difference
If level of confidence is lower than 0.05 we can reject null
Only larger sample sizes
Independent t-test
Statistical difference between 2 means/interventions/change scores
Dependent variable that is continuous eg interval/ratio
Independent variable that is categorical eg gender
Assumption: Normal distribution, no outliers (check with box plot), random sample, no relationship between samples.
Paired T-test
Determines if difference between means is 0
Null hypothesis true mean difference = zero
Assumptions
dependent variable is continuous (interval/ratio), normally distributed and no outliers
Variables have to be independent of one another.
ANOVA
Statistical difference between 3 or more means
One-way anova the comparison involves means but with only one independent variable
F-test tells us if tests are statistically significant
Total variance = set of responses is made up between group and within group variance
between-group variance measures how much the sample means of the groups differ from one another
within-group variance measures how much responses within each group differ from one another.
Does not identify which pairs of means are different
N-way Anova
Analyse several independent variables at the same time
Multiple independent variables act together to affect the dependent variable group means (interaction effect)
Used for experimental design
Perceptual mapping
Graphic representation from analysis
Perception based of other brands in comparison to key attributes
Parametric tests
Normal distribution
Interval/ratio data
Results affected outliers
More statistical power
T-Test, Anova, Regression,
Week 10
Evaluate the types of relationships between variables
Strength of association (weak, moderate, strong)
Direction (positive or negative)
Presence (systematic and consistency)
Type of relationship (Linear and nonlinear)
Linear: Changes in one variable changes the second variable
Curvilinear: Increases to a certain extent then starts to decrease (curve)
Explain the concepts of association and covariation
Association: Numerical measure of strength between relationships
Covariation: Amount of change in one variable that consistently changes in another variable.
Discuss the differences between Pearson correlation and Spearman correlation
Pearson correlation:
Statistical measure of strength between two metric variables.
Varies between -1 and 1 (0 is no correlation)
Null hypothesis : is no association
0.81-1 = strong
0.00-0.20 = weak
May be a non-consistent (nonlinear) relationship
Significance level must be calculated: might falsely reject null
Interval/ratio, normally distributed, linear relationship
Spearman rank order
Two variables used ordinal scales
Either one is ordinal
Explain the concept of statistical vs. practical significance
Statistical:
Statistical significance is used to examine the coefficient for multiple regression
Not all of them are statistically significant
Because some of the procedures involved in determining the statistical significance of a statistical test include consideration of the sample size, it is possible to have a very low degree of association between two variables show up as statistically significant
However, by considering the absolute strength of the relationship in addition to its statistical significance, the researcher is better able to draw the appropriate conclusion about the data and the population from which they were selected.
Explain when and how to use regression analysis
Estimates relationship between one dependent variable and one or more independent variables
Which of the variables have an impact?
Most important variables
How they interact with each other
How certain are we about the variables
Any point not on the line is unexplained variance
Bivariate analysis focuses on one dependent and one independent variable
Multiple regression analysis has multiple independent variables
Each variable has a separate regression coefficient is calculated which describes its relationship with the dependent variable
Linear relationship, Homoscedasticity and normal distribution
Week 11
List the objectives of a research report
Clear concise interpretation (logical explanation, make complex information easy to understand) of project
Accurate, easy to understand, credible (create believability through good quality and organisation)
Guide future research and serve as an information source
Logical recommendations
Describe the format of a marketing research report
Title page
table of contents
Exec summary, introduction
Research Method and procedures (research design, sampling, sampling process, primary data and secondary data)
Data analysis and findings (the body, tables/figures/graphs,
conclusions and recommendations
Limitations
appendix
Discuss the techniques of graphically displaying research results
Frequencies: bar chart, pie chart or table
Several thematic related variables can be presented in same table
Bar charts are used to compare groups
ANOVA/t-tests be presented in tables (include correlations)
Regression analysis displays predictor and outcome with arrows
Address the problems encountered in preparing reports
Lack of data interpretation
Unnecessary use of complex statistics
Lack of relevance
Placing too much emphasis on few statistics
Appreciate the significance of presentations in marketing research
Helps to make good decisions
Seen by only those commissioning the report
Seniors rely on short, concise presentations