Fresh Data Analyst Interview Preparation Guide

Download PDF

Short and long-term outcomes can be measured by using different methodologies for collecting data.
★ Cross-sectional study
★ Longitudinal study

Long-term outcomes are persistence of behaviors and broader lifestyle changes.

Medium-term outcomes include changes in behavior and decision-making.

Short-term outcomes are most likely include changes in skills, attitudes and knowledge.

To rule out that a relationship between two events has been distorted by other, external factors, it is necessary to control for confounding. Confounding factors may actually be the reason we see particular outcomes, which may have nothing to do with what is being measured.
To rule out confounding, additional information must be gathered and analyzed. This includes any information that can possibly influence outcomes.

Example:
When mounting a campaign against alcohol-impaired driving, it is important to know whether other interventions aimed at road traffic safety are being undertaken at the same time. Similarly, if the campaign coincides with tighter regulations around BAC limits and with increased enforcement and roadside testing by police, it would be difficult to say whether any drop in the rate of drunk-driving crashes was attributable to the campaign or to these other measures.

A causal relationship exists when one event (cause) is necessary for a second event (effect) to occur. The order in which the two occur is also critical.
Example:
★ For intoxication to occur, there must be heavy drinking, which precedes intoxication.
★ Determining cause and effect is an important function of evaluation, but it is also a major challenge.

Causation can be complex:
★ Some causes may be necessary for an effect to be observed, but may not be sufficient; other factors may also be needed.
★ While one cause may result in a particular outcome, other causes may have the same effect.
Being able to correctly attribute causation is critical, particularly when conducting an evaluation and interpreting the findings.

An association exists when one event is more likely to occur because another event has taken place. However, although the two events may be associated. One does not necessarily cause the other. The second event can still occur independently of the first.

Example:
Some research supports an association between certain patterns of drinking and the incidence of violence. However, even though harmful drinking and violent behavior may co-occur, there is no evidence showing that it is drinking that causes violence.

One of the most important issues in interpreting research findings is understanding how outcomes relate to the intervention that is being evaluated. This involves making the distinction between association and causation and the role that can be played by confounding factors in skewing the evidence.

★ Relevance
★ Results/Impact
★ Sustainability
★ Effectiveness
★ Efficiency

Where quantitative data have been collected, statistical analysis can help measure the degree of change that has taken place and allow an assessment to be made about the consistency of data.

This process usually includes the following steps:
★ Organizing the data for analysis (data preparation)
★ Describing the data
★ Interpreting the data (assessing the findings against the adopted evaluation criteria)

The purpose of the data analysis and interpretation phase is to transform the data collected into credible evidence about the development of the intervention and its performance.

A Venn Diagram is one way that sets can be visualized.

If two sets have no members in common then the two sets are called disjoint set.

Two operations to perform sets are:
★ Intersection
★ Union

A set that contains no members is called the empty set and is denoted with the symbol '∅'.

There are two objects of set:
★ Members
★ Elements

A set is a collection of distinct objects that have a specific property.
For example, the set of all digits greater than 5 includes the numbers 6, 7, 8 and 9.

The normal distribution is a pattern for the distribution of data which follows a bell-shaped curve. The normal distribution is called normal because the data points are concentrated in the center near the mean. The data does not usually contain extreme values and the probability of deviations of the data points from the mean are (nearly) identical in either direction.

A measure of the dispersion of the data that is shown in a box plot is called the interquartile range. It is the difference between the upper quartile and the lower quartile.

The quartiles of a set of values are the three points that divide the data set into four equal groups, each representing a fourth of the population being sampled.

The mode of a set of numbers is the number that appears most often.

Data analysis is the process of systematically applying statistical and/or logical techniques to describe and illustrate, condense and recap and evaluate data.

A counting principle that you may appear on a data analysis question is the inclusion-exclusion principle, which states that the number of elements in the union of two finite sets equals the count of the members in the two sets minus the number of elements in the intersection of the two sets. The number of elements in a set is denoted using two parallel lines, ||. Using set notation, the inclusion-exclusion principle for sets D and E is written as the following:

|D∪E| = |D| + |E| - |D∩E|

All normal distributions satisfy the following properties:

★ Approximately 68% of the observations fall within 1 standard deviation of the mean
★ Approximately 95% of the observations fall within 2 standard deviations of the mean
★ The mean, median and mode of the data are nearly all equal