In an article published in Science in 1976 Allison and Cichetti looked at “ecological and constitutional correlates” of sleeping patterns in mammals. This is an old article that used rather dated analytical methods applied to a limited data set. The questions that can be answered from any data set will be limited to those involving the variables selected for inclusion in any analysis. So, although the paper was published in Science, the original write up is not necessarily an example of good data analysis.

The authors’ concluded that slow wave sleep may be related to size whilst paradoxical sleep duration may be related to predator danger. However the number of hours of both types of sleep are themselves highly correlated.

d<-read.csv("/home/msc_scripts/data/sleep.csv")
dt(d)

The datathat are provided includes the following variables.

  1. BodyWt=body weight in kg.
  2. BrainWt= brain weight in g,
  3. Sleep=total sleep (hrs/day) (sum of slow wave and paradoxical sleep),
  4. MaxAge=maximum life span (years), Gestation=gestation time (days)

I have added an additional columnn that was not included in the original paper. This is diet. The species have been classified into five rather broad classes.

  1. CARN=Carnivore,
  2. GRAN=granivore,
  3. INSECT=Insectivore,
  4. OMNI=Omnivore,
  5. HERB=Herbivore

Yuur task is to investigate these data, form some research questions that may be related to ideas in the original article. There may be some relationship between diet and the number of hours spent sleeping. You should also look at covariates and other patterns in the data.

In addition to using the exercise to practice applying some of the methods taught on the course you should critically evaluate the strength of the evidence provided by this data set to address any sort of hypothesis regarding causal links between the variables.

How useful are small data sets such as this one, based as it is on aggregated, species level, measures taken from a rather arbitrary sample?

Does your own analysis help you to suggest improvements in the design of studies of this nature in order to address more complex questions in a non trivial manner?

Write up your finding in the form of a very brief research article.

Sleep in Mammals: Ecological and Constitutional Correlates Truett Allison; Domenic V. Cicchetti Science, New Series, Vol. 194, No. 4266. (Nov. 12, 1976), pp. 732-734.