A Mere Test

R Markdown Using the knitr package

This is just a test for using knitr in Blogger. Part is just copy-pasted from the knitr package.

Let's try a histogram of u^2

hist(rnorm(100)^0.2)

A Minimal Example for Markdown

This is just copied from the knitr example. This is a minimal example of using knitr to produce an HTML page from Markdown.

R code chunks

# The sum of two uniform variables
S = runif(1000, 1, 2) + runif(1000, 0, 1)
hist(S, breaks = 25)

Math

Let's try some LaTeX math as usual: \( P(j|X) = \frac{e^{u_j}}{1+e^{u_j}} \).

Misc

You can indent code chunks so they can nest within other environments such as lists.

1. the area of a circle with radius x

   pi * x^2
   

   ## Error: object 'x' not found
   

2. OK, that is great

Markdown is super easy to write. Go to knitr homepage for details.

Global Branding

I have been doing some reading on how to manage global brands. As is often the case with branding, there is just too much stuff written and published. My reductionist take can be summarized like this

Global standardization matters as it allows us to gain scale and therefore reduce marketing costs. There are synergies in communication; and consistency across global media can help build a stronger brand. In some cases, the "global" aspect of the brand concept can add positive associations. However, there are many marketing decisions that cannot be easily centralized and need to be kept at the local level. Typically, these decisions are related to distribution and pricing.

Pricing needs to be optimized locally because price sensitivity and disposable income vary widely across the globe. Distribution relies too much on existing institutions, which tend to be traditional full of local regulations and cultural norms. A similar argument can be made for sales force management, promotional activities and some product features.

In broad terms, we are distinguishing between the more strategic decisions from the tactical decisions. This is not clear cut, but branding, brand building communication and choosing a price point should be a responsibility of the global brand manager. Tactical activities (the so called "marketing mix") can be customized to the local needs and conditions.

Jim Stengel's Brand Ideals

Jim Stengel, a Procter & Gamble marketing executive, has proposed the concept of the brand ideal. What is a "brand ideal"? According to his book "Grow", it is a "shared goal of improving people's lives." Brands that operate according to their ideals, are more likely to grow profitably and, Stengel claims, provide higher return to the shareholders.
The book is based on a study, narcissistically called the Stengel Study that identified brands or business (the term is used interchangeably, which is acceptable in this context) which have shown significant growth in the past 10 years and have developed a strong bond with consumers. The Stengel Study finally converged on 50 brands that have created "meaningful relationships with people" and "outpaced" the competition.
If we write down in which way each of these top 50 brands improve people lives and then take a look at the list, we'll see a pattern emerging. All of these ideals can be summarized into 5 fundamental "human values" that improve people's lives. Let me just state them (I am typing verbatim from page 38 in my edition)

1. Eliciting Joy: Activating experiences of happiness, wonder and limitless possibilities
2. Enabling Connection: Enhancing the ability of people to connect with one another and the whole world in a meaningful way.
3. Inspiring Exploration: Helping people explore new horizons and new experiences.
4. Evoking Pride: Giving people increased confidence, strength, security and vitality
5. Impacting Society: Affecting society broadly, including by challenging the status quo and redefining categories.

The book then continues with a description of each of these five ideals and how to align an organization and a brand to choose among them and associate it to your brand.
By page 50, I stopped counting how many things I don't like about the book. It is full of vanity, the methodology is just wrong from what I can see, the proposal is neither new nor well thought. So I will tell you what I do like.
I do think the world is rewarding brands who can focus on improving people's lives. I think there are more ideals than Stengel's "fundamentals", such as "safety", "simplifying life", etc. In fact, the book makes it sound as if these ideals are so basic that they cover all ideals and they are distinct. I think the list is incomplete and the ideals as they are described are ambiguous and overlapping. But it cannot be denied that as the world becomes more complex and the underlying technology of products is more inscrutable, consumers will rely on ideals to choose and like brands. So the list may need to be rethought, but coming up with a list based on contemporary cases should be applauded, even if the applause is feeble. Bottomline: Ideals matter.

Solitary Confinement, in the First Person

This is a horrifying and heart breaking account of someone sentenced to solitary confinement for more than 25 years.

"Life in the box is about an austere sameness that makes it difficult to tell one day from a thousand others. Nothing much and nothing new ever happen to tell you if it’ a Monday or a Friday, March or September, 1987 or 2012. The world turns, technology advances, and things in the streets change and keep changing all the time. Not so in a solitary confinement unit, however."

There is also this

"Had I known in 1987 that I would spend the next quarter-century in solitary confinement, I would have certainly killed myself. If I took a month to die and spent every minute of it in severe pain, it seems to me that on a balance that fate would still be far easier to endure than the last twenty-five years have been. If I try to imagine what kind of death, even a slow one, would be worse than twenty-five years in the box—and I have tried to imagine it—I can come up with nothing. Set me afire, pummel and bludgeon me, cut me to bits, stab me, shoot me, do what you will in the worst of ways, but none of it could come close to making me feel things as cumulatively horrifying as what I’ve experienced through my years in solitary. Dying couldn’t take but a short time if you or the State were to kill me; in SHU I have died a thousand internal deaths. The sum of my quarter-century’s worth of suffering has been that bad."

We need to rethink the way society punishes crime. We have to agree that there are some levels of cruelty that should not be reached. There is a reason we don't torture prisoners in an enlightened society. Same applies here.

One Sentence from the First Paragraph of John Stuart Mill's Autobiography

"But I have thought that in an age in which education, and its improvement, are the subject of more, if not of profounder study than at any former period of English history, it may be useful that there should be some record of an education which was unusual and remarkable, and which, whatever else it may have done, has proved how much more than is commonly supposed may be taught, and well taught, in those early years which , in the common modes of what is called instruction, are little better than wasted."

Of course, you can read the whole thing, for free.

Doubling the Population of the USA

The population density of the US is 34 inhabitants/km^2. If you double that, you get to the population density of the state of Georgia, hardly an overpopulated place for world standards. Triple the population and you get to the population density of Spain (quite underpopulated for European standards). If the US had 6 times more inhabitants (1.9 billion), it would be as densely populated as Italy. I know these things don’t scale up linearly and all these regions have different natural resources, etc. But it is interesting to get some perspective to show that nothing essential would change by doubling the population.

Source: Wikipedia, of course

How to write LaTeX in Blogger

I just copied this great script from Mathjax. To be fair, I found it here. Now I type equations beautifully (which is not the same as typing beautiful equations).
The simple OLS coefficient estimate:
$$\hat{\beta} = (X'X)^{-1}(X'y)$$
Logit probabilities for a respondent \(i \) choosing alternative \(j\):
$$P_{ij}=\frac{e^{U_{ij}}}{\sum_k e^{U_{ik}}} $$
and so on ...

How to Say the Age of a Baby or Child ("Rule of 12")

How should we say how old a child is? A 2-day old infant is is 2 days old. But a four weeks old baby is not a 30 days old. Nor is a 14 month toddler a 57 week old.  Like everything we measure the age of a child is best stated in relation to the magnitude. Nobody measures his or her own weight in tons or mg.

I will propose that we summarise the standard practice according to some descriptive rules.

1.  0-14 days : express age in days.
2. 14 days - 13 weeks (3 months) : express age in weeks
3. 3 months - 12 months: express age in months and half months (e.g "8 and a half months") 
4. 12 months - 24 months : express age in months after the year (e.g. "a year and three months")
5. 2 years - 4 years: in years with half years (e.g. "2 years and a half")

These rules seem ok, and they seem to follow standard practice. But I was looking for a simple, parsimonious rule. After looking at different options, I propose: Rule 1: Do not use any number higher than 12. Notice that this rule would not match the cutoffs of the originally proposed rules, but it would be a good approximation. What about the half months and half years? We can add Rule 2: Use half days, weeks, months, years, as granular units. This means that 6 months would hit Rule 1 as 12 half months. How would the descriptive rules above change?

1. 0-6 days: age in half days.
2. 6-12 days: age in full days.
3. 12 days - 6 weeks: age in half weeks (e.g. one can say "3 weeks and a half")
4. 7 weeks - 12 weeks: age in full weeks.
5. 13 weeks - 6 months: age in half months (e.g. "4 and a half months")
6. 6 months - 12 months: age in full months. 
7. 12 months - 6 years: age in half years. 
8. older than 6: age in full years.

I am quite satisfied with my parsimonious Rules 1 and 2. I still think it is ok to add the months to children between 1 and 2 years old, but that would make the rules too complex.