A Wordle of my PhD thesis


Comments

Post a Comment

Popular posts from this blog

Apple Or Android? What Your Choice Of Operating System Says About You

Image
This post previously appeared on CREST's Blog in December 2016. Your mobile phone provides all kinds of useful data about what you do, and where. But does even the choice of handset say something about you? Heather Shaw and CREST Associate David Ellis tell us more. Digital traces can provide scientists and law-enforcement agencies with a range of information about groups and individuals. In particular, smartphones can log communications (e.g., calls and messages) and behaviour from multiple internal sensors (e.g., movement and location). For example, patterns of smartphone usage can be indicative of a person’s sleep-wake activity. Location information, on the other hand, is already used to provide alibis, and can accurately predict where a person works and lives. In a very short space of time, the smartphone has become a mini version of ourselves. This is probably why people become increasingly anxious if someone else attempts to use their smartphone! We found that A...
Read more

Kernel Density Plots: Has the histogram had its day?

Image
Simple statistical concepts include the mean, median, standard deviation, and percentiles. These are useful for summarising data. Except these summary statistics are only useful under certain circumstances. When basic assumptions are not met, then any conclusions based on simple summary statistics are likely to be inaccurate. Unable to give a hint as to what is wrong, the numbers can often look perfectly reasonable. Lets consider a sample of 64 reaction time observations (in milliseconds): Mean = 387ms Median = 340ms These look ok, until you view the distribution, which is not unimodal.  Despite being a staple in data visualisation, histograms can often be a poor method for determining the shape of a given distribution because they are strongly affected by the number of bins used. For example, visualising the same data with only four bins can make the same observations appear normally distributed. Similarly, a box-plot can also hide data irregulari...
Read more

When Comfort Replaces Curiosity

There will come a point where a survey can only take you so far. This is especially true when it contains moderation or mediation models. These problems are serious yet are routinely ignored and remain widespread. It's one of those classic cases where the statistical machinery is more powerful than the data it is being applied to. The models answer questions that a single survey will never be equipped to ask. None of this means surveys are useless, but it does mean we should be much more honest about what they can and cannot tell us. What makes this more interesting is that researchers often seem more content to try to drag more out of a survey or psychometric instrument, rather than switch up their methods.  I remain struck by how rarely researchers move beyond their comfort zone, either methodologically or by stepping away from their home discipline. When this happens, it is rarely done with any real conviction. Largely because it involves questioning many methodological and phil...
Read more