Introduction to Research Methods (Bertelsmann Data Science challenge)

Statistics is necessary for decision making and essential to every one. In other taking a research, the subject matter, idea or theme measured is known as a construct. Operational definitions for a construct describes or defines its measurement. Constructs are not easy to define and measure. Examples of constructs include : i. boldness ii. daily happiness, etc.
Operational definitions for the above constructs include:

i.boldness:The ratio of the number of times an individual takes risk to the number of times the individual avoids risks.
iii.Daily Happiness: Percentage of hours a person is happy to hours an individual is not happy in a day.
Data is the most essential part of statistics. Decisions are mainly data driven.

In conducting research, one major hurdle is to ensure that difference between the population parameters and the sample statistics is minimal.
Definitions
Population parameters : Values that describe the entire population.
Sample statistics : Values that describe our sample; we use statistics to estimate the population parameters.

To achieve this, ensure that the sample size is big and the method of choosing sample could be random. Random sampling basically means that each sample has an equal chance of being selected.
Visualizing data in graph helps understand relationship between variables. There is the independent(predictor or x-axis) variable and the dependent(outcome or y-axis).

An important concept worth appreciating is that correlation does not mean causation. Example sleeping well positively correlates with good memory. The fact that there is a relationship between the two does not mean the cause of bad memory is lack of sleep or vice versa.

Types of Researches

The type of research conducted depends on the objective of the experiment. 1. To show relationship between variables, the type of research conducted is :
i. Observational Studies ii. Surveys
2. To show causation, the ideal research is the : i. Controlled experiment.

Using surveys are i. easy, ii.relatively inexpensive, iii.can be undertaken remotely iv. Can easily be analyzed by anyone.
Disadvantages are: i.wrong responses, ii.Biases in responses, iii. Respondents not understanding responses, iv. Respondents refusing to answer.

Controlled Experiments - Definitions
Placebo: Inactive pill to have a comparison group to those who took a pill(control)
Blind: Participants are not told if they received active or inactive pill to make them all believe that they are receiving treatment to reduce biases.
Double Blind: Participants are randomly assigned to the control group or the experiment group. The researchers equally do not know which participants to the active or inactive pill to limit bias.

Reference Material: Bertesmann Data Science Scholaship challenge(Udacity Data Science Nanodegree)