1. Tshwane University of Technology
A GPS to navigate succesfull primary quantitative research
Dr Marthi Pohl

2. The journey from planning to execution
What is the best road to take in order to achieve the intended result
( arrive at destination) within the best time frame?

3. Context Where do I begin? - Statistical Requirement
The departure point = survey methodology
The pivotal role of your sample and questionnaire design
Questionnaire design
Revisit your statistical requirement
What methods can I use to achieve my objectives?
Basic inferential statistics
Multivariate analysis
Planet of other quantitative techniques/ models

4. Statistical Requirements
Differences
Relationships
Prediction
Classification
Data reduction
Profiling/ segmentation
Pattern Recognition
Time series analysis ( or time and e.g. company – panel data)

5. Objectives of research
The departure point = survey methodology
Sample method
Sample size
Data collection
Questionnaire design
Objectives of research

6. Sample and questionnaire: The corpus callosum of
Sample and questionnaire: The corpus callosum of quantitative primary research
Corpus Callosum: Links the two brain hemispheres to align a person’s actions with his/her visions and intentions
Objectives
and Literature Review
Results, conclusions

7. Sample methods Probability samples Random?
Key for generalization
Very difficult to achieve in a survey
Stratified random sampling vs cluster sampling
Systematic sample
Representative?
Critical to ensure unbiased samples
Know or acknowledge the existence of bias if not representative

8. Sample methods Non-probability
Purposive
Convenience
Snowball
Acknowledge shortcomings
Multi- stage (select eg departments and then individuals)

9. Sample size Determined by Key pitfall: The magic 30?
Margin of error and confidence level required
Statistical Techniques to be applied
Key pitfall:
Determining sample size without keeping non-response into account
The magic 30?
Realized sample vs initial planned sample
When it make sense to use a census – total population

10. Data scale type fundamentally impact analysis methods that can be applied

11. Questionnaire design – scales
The key to enable analysis required to answer research question
Need different map types for different destinations (roads, rivers, flight paths)
Types
Dichotomous
Rating (3 -7, 10) - absolute
Semantic differential scales (opposite words)
Side by side matrix (importance and satisfaction asked next to each others)
Ranking (relative)
Tick all that apply

12. An example of a semantic differential scale

13. Data collection Method of execution
Online
But beware = exit survey at any point in time
Structured interviews
But beware availability
Fieldworker knowledge and skills
But beware the dangerous delete button
Weigh the advantages and disadvantages of each method before making a final decision (might result in a cul de sac!)

14. Statistical Requirements (revisit)
Differences – between two or more groups w.r.t a variable of interest
Relationships – between two variables
Prediction – direction of relationship and potential multiple predictors
Classification – can group membership (eg voters/non-voters) be predicted by means of a list of variables
Data reduction – Can a large number of variables actually be represented by a much smaller number of underlying factors
Profiling/ segmentation
Pattern Recognition
Time series analysis ( or time and e.g. company – panel data)

15. A peek into Statistical Techniques
Parametric and non-parametric test
Parametric
Key to ensure assumptions are met
Assumption violation - determine to what extent test to apply is robust w.r.t violations of the assumptions
Non parametric techniques
Distribution free
Can be applied to small samples

16. Generally well -known statistical techniques
Means tests ( t test, ANOVA, MANOVA, Mann-Whitney, Kruskal- Wallis)
Relationships ( Correlation, Pearson Chi-square test for independence, several measures of associations)
Regressions ( a multitude of types)
Discriminant analysis
Exploratory and confirmatory analysis
Structural equation modelling
Cluster analysis

17. The lesser known ones Game Theory Neural Networks
Self Organising Maps (type of clustering)
Data mining techniques for fraud detection and control

18. A view of techniques applied in fraud detection and control

19. But beware, data manipulation…
Get the facts first, then you can distort them as much as you please ( Mark Twain)
To be continued….techniques and reporting