I'm just making this up to understand the rules...

Claim: People wearing roller blades can get to places in lesser time than those who don't. (Assuming on foot)

Now, I'd like to express this using symbols (that's what I call them, unless there's another name for this).

I'm guessing: If there's no problem with the symbol to be used, I'd like to use t as the time taken to walk. So my $H_0$ would be that even with roller blades on, they would need the same amount of time.

Based on this, is it correct to express the following?

$H_0$ : $\mu$ = t

Or is there a better way to express this?

Please feel free to correct me - I really want to learn this!



For simplicity, let us fix definite start and end points to the hypothesis, so in words the hypothesis might be

"People traveling from point A to point B by roller blade arrive sooner than those who walk."

Probably the simplest way to express this hypothesis is to add the qualifier "... on average".

Then the hypothesis becomes a definite statement about the conditional expectation of the travel time, given the mode of transportation.

To proceed you would then introduce a symbol for the travel time, say $T$, and a symbol for the mode of travel. If we are assuming that all travelers either use roller blades or they walk, then we can represent the travel mode by a dummy variable such as $$R=\begin{cases}1 & \text{if roller blade} \\ 0 & \text{if walk} \end{cases}$$

Then the hypothesis becomes

$$\mathbb{E}[\,T\mid R=1\,] < \mathbb{E}[\,T\mid R=0\,]$$

where $\mathbb{E}[\,x\mid a\,]$ should be read as "the expected value of $x$, given $a$".

Usually the symbol $H_0$ that you use is reserved for the null hypothesis, which is typically the negation of the hypothesis of interest. So for example in this case, $H_0$ might be "roller blading is no faster than walking".

One last point: The hypothesis as expressed in the equation above is not actually empirically testable, because the theoretical expectations cannot be literally computed. So usually a next step is to use the theoretical hypothesis to generate predictions about what would be observed in a data-set (sample) of actually observed $(T,R)$ pairs.

  • $\begingroup$ That is one of the most beautiful explanations I've seen in a long time on the entire stackexchange. Thanks! I do have 2 questions on the what you've mentioned. First, how do I express the statement roller blading is no faster than walking as a symbols. Second, I thought that T is reserved for other things like the T-Score. So are we really free to use that symbol? Thanks a again! $\endgroup$ – itsols Sep 29 '16 at 1:19
  • 1
    $\begingroup$ For the first: In the hypothesis (gray box), change $<$ to $geq$ (this is literally the "negation of the hypothesis of interest"). For the second: I am not really a statistician, so I tend to use whatever symbols are convenient! (I mentioned $H_0$ because that one is pretty standard in this context; AFAIK for the various "t" statistics the lower-case $t$ is used ... but that is also used for "time" series of course!) $\endgroup$ – GeoMatt22 Sep 29 '16 at 1:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.