20
$\begingroup$

I have two variables which don't show much correlation when plotted against each other as is, but a very clear linear relationship when I plot the logs of each variable agains the other.

So I would end up with a model of the type:

$$\log(Y) = a \log(X) + b$$ , which is great mathematically but doesn't seem to have the explanatory value of a regular linear model.

How can I interpret such a model?

|cite|improve this question
$\endgroup$
  • 5
    $\begingroup$ I've nothing substantial to add to the existing answers, but a logarithm in the outcome and the predictor is an elasticity. Searches for that term should find some good resources for interpreting that relationship, which is not very intuitive. $\endgroup$ – Upper_Case-Stop Harming Monica Mar 26 at 18:30
  • $\begingroup$ The interpretation of a log-log model, where the dependent variable is log(y) and the independent variable is log(x), is: $\%Δ=β_1\%Δx$. $\endgroup$ – Bob Mar 27 at 15:52
  • 3
    $\begingroup$ The complementary log-log link is an ideal GLM specification when the outcome is binary (risk model) and the exposure is cumulative, such as number of sexual partners vs. HIV infection. jstor.org/stable/2532454 $\endgroup$ – AdamO Mar 27 at 15:58
  • 2
    $\begingroup$ @Alexis you can see the sticky points if you overlay the curves. Try curve(exp(-exp(x)), from=-5, to=5) vs curve(plogis(x), from=-5, to=5). The concavity accelerates. If the risk of event from a single encounter was $p$, then the risk after the second event should be $1-(1-p)^2$ and so on, that's a probabilistic shape logit won't capture. High high exposures would skew logistic regression results more dramatically (falsely according to the prior probability rule). Some simulation would show you this. $\endgroup$ – AdamO Mar 27 at 20:52
  • 1
    $\begingroup$ @AdamO There's probably a pedagogical paper to be written incorporating such a simulation which motivates how to chose a particular dichotomous outcome link out of the three, including situations where it does and does not make a difference. $\endgroup$ – Alexis Mar 27 at 21:05
27
$\begingroup$

You just need to take exponential of both sides of the equation and you will get a potential relation, that may make sense for some data.

$$\log(Y) = a\log(X) + b$$

$$\exp(\log(Y)) = \exp(a \log(X) + b)$$

$$Y = e^b\cdot X^a$$

And since $e^b$ is just a parameter that can take any positive value, this model is equivalent to:

$$Y=c \cdot X^a$$

It should be noted that model expression should include the error term, and these change of variables has interesting effects on it:

$$\log(Y) = a \log(X) + b + \epsilon$$

$$Y = e^b\cdot X^a\cdot \exp(\epsilon)$$

That is, your model with a additive errors abiding to the conditions for OLS (normally distributed errors with constant variance) is equivalent to a potential model with multiplicative errors whose logaritm follows a normal distribution with constant variance.

|cite|improve this answer
$\endgroup$
  • 3
    $\begingroup$ OP may be interested to know that this distribution has a name, the log-normal: en.wikipedia.org/wiki/Log-normal_distribution $\endgroup$ – gardenhead Mar 26 at 19:40
  • 2
    $\begingroup$ What about the effect of Jensen's inequality? Generally for convex g, $E[g(X)]≥g(E[X])$ $\endgroup$ – Stats Mar 26 at 21:17
14
$\begingroup$

You can take your model $\log(Y)=a\log(X)+b$ and calculate the total differential, you will end up with something like : $$\frac{1}YdY=a\frac{1}XdX$$ which yields to $$\frac{dY}{dX}\frac{X}{Y}=a$$

Hence one simple interpretation of the coefficient $a$ will be the percent change in $Y$ for a percent change in $X$. This implies furthermore that the variable $Y$ growths at a constant fraction ($a$) of the growth rate of $X$.

|cite|improve this answer
$\endgroup$
  • $\begingroup$ So if the log-log plot is linear, that would imply a constant growth rate? $\endgroup$ – Dimitriy V. Masterov Mar 26 at 17:22
  • $\begingroup$ Not actually, the growth rate of $Y$ will be constant if and only if $a=0$. $\endgroup$ – RScrlli Mar 26 at 18:18
  • $\begingroup$ Not over time, the growth rate with respect to the growth in x. $\endgroup$ – Dimitriy V. Masterov Mar 26 at 18:21
  • $\begingroup$ reordering doesn't help, i'd remove it $\endgroup$ – Aksakal Mar 26 at 18:21
  • 1
    $\begingroup$ @DimitriyV.Masterov Ok, then since the $\log(Y)$ is linear in $\log(X)$ it means that the variable $Y$ grows at a constant fraction of the growth rate of $X$. Is there something wrong with my answer according to you? $\endgroup$ – RScrlli Mar 26 at 18:36
7
$\begingroup$

Intuitively $\log$ gives us the order of magnitude of a variable, so we can view the relationship as the orders of magnitudes of the two variables are linearly related. For example, increasing the predictor by one order of magnitude may be associated with an increase of three orders of magnitude of the response.

When plotting using a log-log plot we hope to see a linear relationship. Using an example from this question, we can check the linear model assumptions:

log-log

|cite|improve this answer
$\endgroup$
  • 3
    $\begingroup$ +1 for an intuitive answer to an unintuitive concept. However, the image you have included clearly violates constant error variance across the predictor. $\endgroup$ – Frans Rodenburg Mar 27 at 7:24
  • 1
    $\begingroup$ The answer is right, but authorship attribution is wrong. The image shouldn't be attributed to Google Images but, at least, to the web page where it is to be found, that can be find out just by clicking in Google images. $\endgroup$ – Pere Mar 27 at 8:42
  • $\begingroup$ @Pere I cannot find the original source of the image unfortunately (at least using reverse image search) $\endgroup$ – qwr Mar 27 at 9:33
  • $\begingroup$ It seems to come originally from diagramss.us though that site is down and most of its pages are not in the Web Archive apart from its homepage $\endgroup$ – Henry Mar 28 at 9:03
4
$\begingroup$

Reconciling the answer by @Rscrill with actual discrete data, consider

$$\log(Y_t) = a\log(X_t) + b,\;\;\; \log(Y_{t-1}) = a\log(X_{t-1}) + b$$

$$\implies \log(Y_t) - \log(Y_{t-1}) = a\left[\log(X_t)-\log(X_{t-1})\right]$$

But

$$\log(Y_t) - \log(Y_{t-1}) = \log\left(\frac{Y_t}{Y_{t-1}}\right) \equiv \log\left(\frac{Y_{t-1}+\Delta Y_t}{Y_{t-1}}\right) = \log\left(1+\frac{\Delta Y_t}{Y_{t-1}}\right)$$

$\frac{\Delta Y_t}{Y_{t-1}}$ is the percentage change of $Y$ between periods $t-1$ and $t$, or the growth rate of $Y_t$, say $g_{Y_{t}}$. When it is smaller than $0.1$, we have that an acceptable approximation is

$$\log\left(1+\frac{\Delta Y_t}{Y_{t-1}}\right) \approx \frac{\Delta Y_t}{Y_{t-1}}=g_{Y_{t}}$$

Therefore we get

$$g_{Y_{t}}\approx ag_{X_{t}}$$

which validates in empirical studies the theoretical treatment of @Rscrill.

|cite|improve this answer
$\endgroup$
  • 1
    $\begingroup$ This is probably what a mathematician would call intuitive :) $\endgroup$ – Richard Hardy Mar 28 at 10:59
2
$\begingroup$

A linear relationship between the logs is equivalent to a power law dependence: $$Y \sim X^\alpha$$ In physics such behavior means that the system is scale free or scale invariant. As an example, if $X$ is distance or time this means that the dependence on $X$ cannot be characterized by a characteristic length or time scale (as opposed to exponential decays). As a result, such a system exhibits a long-range dependence of the $Y$ on $X$.

|cite|improve this answer
$\endgroup$

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