0:06

So, what have we done so far?

We've taken our raw survey about 30 survey items and

identify nine underlying constructs or factors, and I've put some names on them,

and this is all based on the questions that load onto those particular factors.

So, we had a factor relating to perceptions of financial freedom.

We had a factor that related to seems like having a positive outlook.

Societal indifference kind of questions about, that seem to tap into a theme

of patriotism, theme of being self-confident, of being image conscious.

A theme of it being adventurous.

A theme of the importance of family and children, and

a theme of environmental indifference.

Well, that's great.

So now, I have nine items or

nine factors that summarized all of my original survey items.

So far, subsequent analysis.

Take your original survey items and put those aside.

We're going to just be working with these nine factors for

the remainder of our analysis.

Now, we looked at the factor loadings already.

The other piece that we want to take a look at in our output is the factor

scores.

And again, we get a very lengthy output to get to the rotated result.

We have our factor loading matrix.

So looking at those items again and

here quiet evenings at home, children being the focus of the marriage.

That showed up as a theme.

Now, is that going to be a significant predictor of people's interest in

buying a sports car?

It might be,

but that's what we're going to allow our subsequent analysis to tell us.

1:48

And if we continue to scroll down, this is what we're interested in.

The factor scores after the rotation.

Notice each observation, we now have the factor scores.

Now these numbers not going to be too meaningful just interpreting them on their

own, but think of them as relative to each other.

Some individuals are going to score high on a factor,

some individuals are going to score low on a factor.

What we're going to be doing, let's say, let's take this set of items.

Let's take D1 through D9, our constructs that we've identified.

Put that into a regression analysis,

which of these factors is a significant predictor of purchase intentions?

So, that's one analysis that we can do.

You can also form segments using this information.

We'll tak a look at that in the next module of this course.

So, let's run our linear regression.

Our Y variable, that's going to be purchase intention.

Our Xs, well, that's going to be the set of factors that we've now identified.

2:57

So rather than running this on the raw survey results,

we're going to run it on our new factor scores.

So what we've done already is take the 30 survey items,

chop it down to a smaller number that contains a good chunk of the original

information and let's use those as our predictors.

So we know which survey items move together, that's what's allowed us to

understand those preferences and those beliefs that people have.

3:36

So show you where we can do that within Excel stat.

I'm just going to use linear regression,

then use the software package of your choice, whether it's R, Real Statistics,

MATLAB, Excel Stack, SPSS, same techniques are going to be there for all of these.

We're going to go back to the raw data.

Purchase intentions is my first column.

Where are my predictors?

Well, it's on my results tab from my factor analysis.

It's going to be the factor scores.

4:09

And we can click through the different options, make sure that missing data,

we're just not going to accept it.

We don't have any in this case.

We're going to get our descriptive statistics.

We're good to go.

So we're going to take a look at the linear regression results from our

analysis that we conducted and

use the tool that you're most comfortable with as far as conducting that analysis,

but here's what we're ultimately interested in.

So, we've now gone from using the raw survey data's been processed down with

factor analysis.

The resulting factor scores becoming our input,

our Y variable now being purchase intentions and

we can see which are the significant drivers of purchase intentions.

So, anything that we're highlighting and Excel stats helped us out a bit.

Those are the significant drivers, we can see which one has the biggest impact.

5:07

Now I've put the names back on to these, so that we can interpret it, but

that positive outlook is associated with higher purchase intentions.

That patriotism dimension, self-confidence being image conscious,

all associated with higher purchase intention.

Being family bound, the importance of children,

quiet nights at home not a significant driver.

The financial dimension also not a significant driver.

The big one and now,

the reason we can say this is because the factor scores are all on the same scale.

So we're comparing apples to apples when we look across these coefficients,

but that positive outlook and

the adventurous factor are the big drivers of purchase intentions.

And so, if what we're thinking about is how do I build a marketing campaign?

What dimensions do I want to appeal to?

What kind of imagery do I want to use?

Well, what I want to focus on is playing up that positive aspect,

that positive outlook and I want to focus on that adventurous lifestyle.

I don't want to be talking about society.

I don't want to be putting children seat in the back of the car.

I don't want to be talking about the financial elements here.

Those aren't affecting the purchase decisions.

It's about a positive outlook.

It's about the patriotism, self-confidence and

being image-oriented, being adventurous.

Those are the things that seem to really drive that purchase intention question.

And so, those are the insights that can help us as far as developing the marketing

campaign for the product.

So now that we've kind of run through this,

let me show you one more set of results where factor analysis can be used.

This was an analysis done of shopping behavior.

So survey, conducted to see which sets of stores that people frequently shopped at.

And so, I've thrown this into the factor analysis and

highlighted shopping at which stores tended to load on to which factors.

So the first factor, we see some retail stores.

So Abercrombie and Fitch, Aeropostale, American Eagle and Hollister.

The second factor, Barnes and Noble, Borders, Target and Starbucks.

So in a sense, what would these stores tend to have in come when we

think about them or the people who are shopping at these stores?

Nordstrom, Neiman Marcus tend to go strongly together.

Best Buy, Circuit City, Comp USA, all electronics,

consumer electronics oriented.

K-Mart, Sam's, Sears, Walmart, more of a value dimension,

K&B Toys limited to Toys R' Us, perhaps younger dimension.

7:58

Electronic boutique games and GameStop of video game dimension.

Dick's Sporting Goods, Nike, Sports Authority,

all tending to go together obviously a sports dimension there.

And so you can see, when we put the original data into factor analysis,

it doesn't know what each response or what each survey item corresponds to,

but what we're getting factors that make sense.

So factor analysis, a very versatile dimension reduction tool.

8:29

So helps us understand which survey items tend to move together,

what are some common, allows us to look at what might be those common themes.

If we're in the process of designing the survey, it's a good opportunity to

identify the ability to eliminate items that might potentially be redundant.

And as we've already looked at, it can be used as an input or subsequent analysis.

So, we've used it as an input into regression analysis.

We'll look at in the next module, how do we take these results and

can they aid us in forming market segments?