Module 2

Question 1

Define a parameter

Answer 1

A parameter is the numerical measure that we are specifically interested in.

Question 2

Define a sample

Answer 2

A sample is a subgroup of individuals from the population.

Question 3

Define an estimator

Answer 3

The average of the sample is an estimator of the population’s average.

More precisely, an estimator is a formula or a calculation recipe which allows us to obtain an approximation of the unknown parameter
of the population from the values observed in the sample.

Question 4

Define an error in the estimation process.

Answer 4

When a sample is used there is invariably a loss of information.

Question 5

When is a study biased?

Answer 5

A study is biased if it’s set procedure has a systematic tendency to over or underestimate the value of the parameter of interest in population.

Let us remind ourselves that a study is biased if its methodology leads

Question 6

When can a Bias arise?

Answer 6

In a study, a bias can arise if the estimator is badly chosen.

Question 7

What is an example of a bias in a study?

Answer 7

If we try to estimate the average of the population using the maximum observed value in the sample, the estimator tends to aim too high.

Question 8

Defineb an estimator without bias.

Answer 8

When we use an average estimator, the estimator is an intuitive estimator, it aims neither too high nor too low in average, when the sample is selected randomly.

Question 9

How can we avoid biases?

Answer 9

By paying particular attention to the study’s design to avoid biases slipping through as much as possible.

Question 10

What are a few different types of bias?

Answer 10

Among the common sources of biases, there is poorly chosen estimator, selection bias, nonresponse bias and measurement bias.

Question 11

Define estimation errors?

Answer 11

Divergence of the sample variable from the true population variable

Question 12

Can Biases be corrected?

Answer 12

In the vast majority of cases, there is a statistical method that can correct biases retroactively. However it is best practice to make the most effort possible in order to reduce these biases to a minimum, before the study begins.

Question 13

Define Selection Bias

Answer 13

This occurs when the sample is not representative of the population.

All the members of the population should have the same probability of being selected in the sample.

Question 14

What are some issues that contribute to selection bias?

Answer 14

-People who are impossible to select for the sample for many technical and logistical reasons.

-People who are not part of the population target that interfere in the sample

• There is no statistical method which allows us to repair this after the fact.

Question 15

What is the strategy to avoid introducing selection bias in our sample.

Answer 15

• Correctly identify the population.
• Selecting a sampling pool that corresponds, if not entirely, as much as possible to this population.
• Prioritize using a chance selection

Question 16

Define nonresponse bias

Answer 16

When selected individuals, refuse to provide this information to us, this can be corrected in situations where we’re able to well identify the characteristics of those who are less inclined to respon, However it is strongly recommended to prevent nonresponse in advance as much as possible.

Question 17

How to avoid nonresponse bias

Answer 17

• Doing a follow-up with the surveyed people.
• Reflect on the factors which may lead to a non-response and make decisions that will help reduce it.

For example, a very long questionnaire may dissuade many people
from answering it.

Question 18

Define Measuremeny Bias

Answer 18

This bias arises when it isn’t possible to accurately measure our variable of interest.

Question 19

Questions to ask when planning a study

Answer 19

1- What precisely is the population that I wish to observe?
2- How do I select a sample, that is the most representative of the population?
3- Does my sampling method use chance?
In other words, does my selected sample resemble simple random sampling?
4- How do I reduce the risk that individuals from this population
5- How can I most accurately measure my variable of interest from the selected individuals of my sample?

Question 20

Define Statistical Inference

Answer 20

Statistical inference consists of extending a conclusion drawn from a sample to a population.

Question 21

Why is selection bias common in UX studies?

Answer 21

Often, a convenient sample is used, meaning that we pick from a subgroup of participants, who for logistical reasons, are easier to reach.

Question 22

Why is meaurement bias common in UX studies?

Answer 22

Psychometrics scales , sophisticated tools or insufficient ecological validity can all contribute to measurement bias.

Question 23

Define the variable of interest

Answer 23

A changing quantity that is measured

Question 24

Define the population

Answer 24

The entire group of people, objects, or events that a researcher is interested in studying

Question 25

Keep in mind:

Answer 25

• Population-sample duality is fundamental
• Estimation of a parameter is not the same as parameter of the population

Question 26

What are the steps it would require estimating a characteristic of a population.

Answer 26

1. Determine the variable of interest
   
   2. Recruit a sample of the target audience
   3. Estimate / calculate the variables collected
   4. Approximate the estimated variable to the parameter of the population
   5. Conclude whether the estimator represents the parameter of the population

Question 27

An example of selection Bias:

Answer 27

Conveniently recruiting a friend to participate in a user test.

Identify

Identify the

Following a sampling protocol will reduce selection bias:

• Identify the population for the study
• Select a sample that corresponds as much as possible to the intended population
• Randomly select individuals who will form the sample (simple random sampling)

Question 28

An example of Non-Reponse Bias:

Answer 28

A participant feels uncomfortable sharing their relationship status in an interview.

Some possibilities to reduce non-responce bias:
* Send a follow-up reminder to the surveyed people
* Reflect on factors that lead to non-response (e.g., long questionnaire or sensitive questions)

Question 29

An example of Measure bias:

Answer 29

Feeling the need to please the interviewer by trying to give answers that the interviewer wants to hear.

Some possibilities to reduce non-responce bias:
* Quality of the questionnaire
* Method to administer the questionnaire
* Anonymous survey

Question 30

Examples of selection bias in UX:

Answer 30

• Recruit a sub-group population for convenience (e.g., students, friends, or mailing lists)
• Compensating participants for their time (UX test takes a long time)
• Compensation amount (high monetary compensation leads to greater selection bias)

Question 31

Examples of measurement bias in UX:

Answer 31

• Using inaccurate tools to measure the variable of interest (e.g., joy, mental load, or activation)
• Ecological validity is not respected (a study not reflecting the natural context of use)

Question 32

An example of ecological validity:

Answer 32

Playing a mobile game or texting while walking outside instead of being seated in a research lab