College aantekeningen
Exstensive lecture notes - Introduction to research in marketing Fall
Extensive lecture notes from all lectures of introduction to research in marketing.
[Meer zien]Voorbeeld 4 van de 44 pagina's
Voorbeeld 4 van de 44 pagina's
In winkelwagengesponsord bericht van onze partner
2 beoordelingen
Door: Mathias • 2 jaar geleden
Door: natashadhillon • 3 jaar geleden
Voorbeeld van de inhoud
Intro. to research in marketing (Fall)Lecture 1 introduction, data exploration & visualization
> All lectures will be recorded and put on canvas
> This course provides students a toolbox to approach marketing-related problems from a rigorous, analytical,
data-based perspective
> For this course you need statistical software package R
> Digital exam (not sure yet if its on-campus or at home)
> The book is recommended, but not required
Every week
1. Lectures (on-campus and live-stream)
2. Video tutorials (online)
3. Lab sessions (on-campus)
Evaluation
> 7 short assignments (one each week) and a digital exam
> To pass the course:
• Assignments and exam both higher than 5
• Final grade (assignments (20%) + exam (80%)) higher than 5.5
• You can’t resit the assignments à if you fail them the exam counts as 100%
• No resit can be taken if you do not participate in the assignments
Assignments à Deadline is exactly one week after the assignment
is posted
Total error framework
This framework shows you what the data is composed off. When
collecting data, you want to get as close to the true values as
possible. However, there often is some sampling error,
measurement error and statistical error that can mess up your
data. Therefore, what you observe might now be the actual value.
Example
For the control group, on average, you observe a 5. On average for
the treatment group, you observe a 6.5, but there is some error. The
true value might be the same in both groups, but the error that
slipped into the treatment group, which skewed it up. This will bias
the results.
It is important that you realize that these errors can have an impact
on your data. And you also need to understand how to reduce these
errors in your data.
Review: sampling
Typically, you can’t research the entire population you’re interested in –
you have to take a sample from the population. So, what you observe are
characteristics of the sample and not the population. However, with the
characteristics of the sample you aim to estimate the characteristics of the
population. Doing so might introduce errors.
Sampling error
> Coverage error
> Sample error
> Non-response error
,Problem
With the adoption of the smartphone everyone can screen their calls. Most people will not pick up the phone if they
see it’s an unknown number. Very few people will still pick up the phone and the ones who do are probably very
different from the ones who don’t. This introduces a very large non-response error in the telephone survey.
In practice
Basically, with almost every survey you will have to use sampling methods, since there will be very few occasions
where you have data for the whole population. Therefore, there will always be some error and bias. One way to
minimize this is by using post-stratification weights.
Let’s say your population is 50% female, but your sample is 80% female.
You’re interested in measuring some quality like: “how likely are you to buy
brand Z (on a 1-5 scale, 5 = most interested, 1 = least)?”
> A simple average will underestimate males, who in this case like
brand Z more than females
Review: measurement scales
Data is usually distinguished between two broad types of variables:
Non-metric: these outcomes can be categorical (labels) or directional – can measure only the direction of the
response (e.g., yes/no).
• Nominal (categorical): number serves only as a label or tag for identifying or classifying objects in mutually
exclusive and collectively exhaustive categories (example: SNR, gender) à mutually exclusive: not at the
same time, collective exhaustive: at least one.
• Ordinal: numbers are assigned to objects to indicate the relative positions of some characteristic of objects,
but not the magnitude of difference between them (example: preference for brands or any other ranking
(1) Brand B, (2) Brand A, (3) Brand B) à this doesn’t tell you anything about the difference between the
brands or about the intensity).
Metric variables: in contrast, when scales are continuous, they not only measure direction or classification, but
intensity as well (e.g., strongly agree or somewhat agree).
• Interval: numbers are assigned to objects to indicate the relative position of some characteristics of objects
with differences between objects being comparable: zero point is arbitrary (Examples: Likert scale,
satisfaction scale, perceptual constructs, temperature (Fahrenheit/ Celcius à it can go below zero).
• Ratio: the most precise scale; absolute zero point. Has all the advantages of other scales (Example: weight,
height, age, income, temperature (Kelvin à there is no negative with Kelvin, zero kelvin is the lowest
theoretical temperature possible))
,The difference between the two is that with interval the zero point is arbitrary and with ratio the zero point is
absolute.
Why is it important to know what scale something is measured on?
The right statistical technique depends on what scale is used. It’s especially important to know whether the scale is
metric or non-metric.
> It makes no sense to calculate the mean of a nominal or ordinal scaled variable: average SNR of the class is …??
> Statistical software packages (like R) expect variables to be of a particular type (results may depend on it)
Summated scales
Measuring some variables is easier than others. In marketing we’re often interested in measuring attitudes, feelings,
or beliefs that are more abstract than, say, age or income.
> More than one question is needed to capture all facets (reducing measurement error).
> Example: satisfaction with purchase experience à How satisfied are you with your recent purchase:
1. I am satisfied with my overall experience with …
2. As a whole, I am not satisfied with … (reversed question “not” à are people paying attention while
answering the questions?)
3. How satisfied are you overall with the quality of …?
Measurements
… “what we observe is not nature itself, but nature exposed to our
method of questioning”
> Validity: Does it measure what it’s supposed to?
> Reliability: is it stable? E.g., Do you measure the same thing in
different points of time?
We want our measure to be valid and reliable.
In practice
> Validity: do these coefficients make sense? (i.e., do the effect sizes
and signs give plausible model results?
• Example: 10% increase in prices leads to 200% increase in
sales (???) à not logical the results are probable not valid
> Reliability: how much do these results change if:
• We add additional control variables to the model
• We take away some observations (e.g., outliers)
• We estimate the same model on a new dataset
Review
Two possible outcomes of hypothesis testing:
> Fail to reject the null (null is true)
> Reject the null (alternative is true)
Two types of errors when hypothesis testing
> Type 1: when you reject the null, while in reality the null hypothesis is
true
> Type 2: when you fail to reject the null, while in reality there actually is
a difference.
, Review: statistical testing
> P-value = probability of observed data or statistic (or more extreme à higher or lower value of the data)
given that the null hypothesis is true
> If it’s “low” then the data are unlikely according to the null, and you can reject the null (low chance of type 1
error)
> Typically, we set the threshold (α) at 0.05 (that is: reject the null if p-value < α)
Hypothetical example
Online store wants to test new website design:
> Null hypothesis: proportions are the same
> Two-tailed test: Z = 2.54, p-value = 0.01
> 1% probability of finding this difference in proportions (or a more extreme
difference) if in fact there’s no difference à with this p-value we could feel
pretty confident about rejecting the null
In practice
> p-value is often misused and overused
• Not the probability of hypothesis being tested is true
• Not the probability that observed deviation was produced by chance alone
• Not a measure of the size or importance (it doesn’t indicate whether it is an important difference, it only
indicates whether it’s a statistical significant difference).
• Not a good way to base business decisions
• By itself, not a good measure of evidence
> Consider p-values along with interpretation (do the results make sense), power, measurement, study design
(sampling), numerical and graphical summaries of the data (soon)
Exploratory data analysis: data preparation
> Always, always, always explore your data before running any model
> In the real world, data come with all sorts of errors, missing values, etc.
> Just a few examples of problems in MANY datasets:
• Recode missing observation (e.g., 9999 = “missing)
• Reverse code negative worded questions (these have to be corrected)
• Check that variables have correct range, are not invalid
• Check mutually consistent (age = 18, birthday 4/30/1901)
• …
Visualization
Purpose
Graphs help you explore the data, they help you get to know and make sense of the data and also to communicate
the results to others.
• Exploration of the data
• Understanding and making sense of the data
• Communicating the results
Choosing the right chart type (this depends on what you want to show)
1. Showing the composition of distribution of one variable
2. Comparing data points or variables across multiple subunits
Dit zijn jouw voordelen als je samenvattingen koopt bij Stuvia:
Bewezen kwaliteit door reviews
Studenten hebben al meer dan 850.000 samenvattingen beoordeeld. Zo weet jij zeker dat je de beste keuze maakt!
In een paar klikken geregeld
Geen gedoe — betaal gewoon eenmalig met iDeal, creditcard of je Stuvia-tegoed en je bent klaar. Geen abonnement nodig.
Direct to-the-point
Studenten maken samenvattingen voor studenten. Dat betekent: actuele inhoud waar jij écht wat aan hebt. Geen overbodige details!
Veelgestelde vragen
Wat krijg ik als ik dit document koop?
Je krijgt een PDF, die direct beschikbaar is na je aankoop. Het gekochte document is altijd, overal en oneindig toegankelijk via je profiel.
Tevredenheidsgarantie: hoe werkt dat?
Onze tevredenheidsgarantie zorgt ervoor dat je altijd een studiedocument vindt dat goed bij je past. Je vult een formulier in en onze klantenservice regelt de rest.
Van wie koop ik deze samenvatting?
Stuvia is een marktplaats, je koop dit document dus niet van ons, maar van verkoper sannevangurp. Stuvia faciliteert de betaling aan de verkoper.
Zit ik meteen vast aan een abonnement?
Nee, je koopt alleen deze samenvatting voor €5,48. Je zit daarna nergens aan vast.
Is Stuvia te vertrouwen?
4,6 sterren op Google & Trustpilot (+1000 reviews)
Afgelopen 30 dagen zijn er 69052 samenvattingen verkocht
Opgericht in 2010, al 15 jaar dé plek om samenvattingen te kopen