Showing posts with label inference. Show all posts
Showing posts with label inference. Show all posts

Tuesday, 3 July 2012

The Sociology Paper Title Generator

After noticing a trend towards quite generic naming of papers in sociology I decided to write a quick script to generate new ones. I took some components of some recent working papers and combined them at random. Initially, I thought it would generate some hilariously ridiculous sounding papers. More worryingly, it generates lots of very plausible sounding research topics. Here are the first 11 it generated:
A spatial analysis of cooperation and end of life care : in Eleven European countries 
The social determinants of Labour market outcomes and Class mobility : in Eleven European countries 
Revisiting Class mobility and part-time working : new perspectives for European research 
Measuring Fertility history and UK immigration policy : recent evidence from Germany 
Understanding employers' gendered hiring preferences and wives' earnings : assessing government responsiveness to public opinion 
Gender inequalities in part-time working and Employment change : recent evidence from Germany 
Long term trends in extreme right party membership and corruption : Does money matter? 
The effects of democracy on extreme right party membership and Class mobility : Does money matter? 
The social consequences of child mortality and Labour market outcomes : assessing government responsiveness to public opinion 
The effects of democracy on educational expansion and education :  revisiting the welfare state conundrum 
A spatial analysis of conflict and maternal education : in Eleven European countries

With the exception of Measuring Fertility history and UK immigration policy : recent evidence from Germany, none of these papers sound implausible and I really want to hear more about Long term trends in extreme right party membership and corruption. 


There's a serious side to this. Part of the problem with moving social science forwards is the difficulty in dismissing alternate explanations with limited data. If we really can connect any two variables together into a plausible hypothesis, how can we hope to dismiss all alternate explanations of any finding?