12.12.2018 Views

South African Psychiatry - November 2018

South African Psychiatry - November 2018

South African Psychiatry - November 2018

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

CULINARY CORNER<br />

More to the point, from a theoretical point of view, in<br />

1925 Freud published an elaboration of his original<br />

theory of penis envy. In his view, the moment at which<br />

a girl discovers her lack of a penis is a moment of<br />

immense psychic trauma. From that moment on, the<br />

girl will want a penis and as she matures, will hope<br />

for second best: her father’s penis. As incest is taboo,<br />

this will inspire feelings of shame in her and she will<br />

ultimately sublimate her desire for her father’s penis<br />

into a desire for a child.<br />

As Coffey (2017) notes, Freud’s theory of penis envy<br />

may be less based on observable phenomena<br />

in girls and young women in general, than on his<br />

daughter’s specific issues.<br />

MAIN MEAL<br />

DATA THUGS<br />

applications. They may process survey results through<br />

a computer application known as SPRITE (Sample<br />

Parameter Reconstruction via Interactive Technique).<br />

This can be used to determine whether survey results<br />

on paper appear to have been fabricated. They<br />

also have the aptly named GRIM, which stands for<br />

Granularity-Related Inconsistency of Means. This<br />

program tests to see whether certain figures reported<br />

in an article are mathematically possible. If they’re<br />

not, then the authors either miscalculated or they<br />

made it up (Bartlett, <strong>2018</strong>).<br />

In his essay, “I Want to Burn Things to the Ground”<br />

Bartlett (<strong>2018</strong>) notes that the problem is exacerbated<br />

by the unwillingness of researchers to share their<br />

data freely (often to protect the anonymity of<br />

respondents).<br />

UNFORTUNATELY, CONCEALING DATA<br />

MAKES IT EASIER TO FUDGE WHAT IS<br />

FOUND. RESEARCHERS OFTEN DO<br />

NOT STATE WHAT THEY ARE TRYING<br />

TO DISCOVER BEFORE THEY ATTEMPT<br />

TO DISCOVER IT. THE FREEDOM TO<br />

ALTER HYPOTHESES THEN FACILITATES<br />

P-HACKING - WHICH REFERS TO WHEN<br />

A RESEARCHER GOES SEARCHING FOR<br />

PATTERNS IN STATISTICAL NOISE.<br />

REPORTING P-VALUES<br />

It gets worse - or better. In the early 2010s a computer<br />

program known as Statcheck was developed by<br />

psychologists at the University of Amsterdam. This<br />

program is used to detect instances in reported<br />

research where the p-value does not correspond to<br />

the value of the reported statistic, whether this be x²,<br />

F, r, t or z.<br />

Social psychology studies have been going through<br />

something of a crisis related to high levels of poor<br />

replication. To add to this growing sense of insecurity,<br />

a group of research investigators consisting of<br />

freelance scientists and fringe academics have<br />

sprung up. They make it their business to challenge<br />

the statistical findings in published research.<br />

These individuals belong to a body called the Society<br />

for the Improvement of Psychological Science and<br />

are often portrayed as “data thugs”. While they are<br />

to be welcomed on the one hand for contributing<br />

to discussions regarding the quality of data, their<br />

dissection of studies are also stifling new research<br />

projects as potential researchers experience<br />

growing fear that their findings may not withstand<br />

their withering scrutiny (Bartlett, <strong>2018</strong>).<br />

“Data thugs” are able to contribute to their researchassaulting<br />

activities by means of computer<br />

USING STATCHECK, CANADIAN<br />

RESEARCHERS, GREEN ET AL (<strong>2018</strong>),<br />

SURVEYED ARTICLES IN SEVERAL<br />

PRESTIGIOUS EUROPEAN AND AMERICAN<br />

PSYCHOLOGY JOURNALS. THEY FOUND<br />

THAT THERE WAS A HIGH LEVEL OF<br />

INCOHERENCE BETWEEN THE P-VALUE<br />

AND THE RELEVANT TEST STATISTICS IN<br />

THE RESEARCH INVESTIGATED. NEARLY<br />

HALF OF ALL THE PAPERS CONTAINED<br />

AT LEAST ONE SUCH ERROR, AS DID 10%<br />

OF ALL NULL HYPOTHESIS SIGNIFICANCE<br />

TESTS.<br />

The researchers then turned their attention to<br />

Canadian journals and discovered similar rates of<br />

p-value errors. The authors recommend that in the<br />

future, journals include explicit checks of statistics in<br />

their editorial process.<br />

60 * SOUTH AFRICAN PSYCHIATRY ISSUE 17 <strong>2018</strong>

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!