Traditional interviews by senior managers and senior
technical staff have emerged as the best way to choose recruits,
according to a new BCS book, Professional Issues in
IT.
A variation on this theme is panel interviews, but these tend to
favour smooth-talkers, said author Frank Bott, who has worked in
the software industry and as a professor of computing science in
the UK and the US.
"The panel may contain a majority of people who are neither
professionally competent nor operationally involved," he said.
"Independent panel members are thought to help prevent nepotism and
corruption but evidence suggests they are often responsible for bad
appointments."
References are being used less often, said Bott, because
legislation allows people to see what has been written about them
and sue for damages. Employers who feel they have been misled by a
reference can also sue.
Bott also raised questions over psychometric tests. These fall
into three categories:
lAbility tests measure verbal, numerical and general skills.
lAptitude tests measure potential to learn skills needed for the
job and have been used widely for recruiting trainee programmers.
These tests can be "fairly effective," said Bott, but the results
are questionable if candidates have had the chance to practice,
which he said is usually the case. In addition, it is difficult to
design good tests for higher-level skills.
lPersonality tests aim to assess characteristics that
significantly affect how someone might behave in relationships with
other people. "Unfortunately there are several competing theories
of personality and although the tests are widely used their value
is far from clear," said Bott.
Situational assessments are often used by big companies when
recruiting graduates. Shortlisted candidates are put through
various situations together and are assessed. This is expensive,
said Bott. He believes a variation on the theme can be used in
traditional interviews, with the interviewer asking what the
candidate would do in a particular situation.
Task assessment is a variation on situational assessments and
aptitude tests. Candidates are asked to carry out some of the job
tasks, such as writing a program.
"This works very well if the tasks lend themselves to being
assessed in this way," Bott said. "The trouble is that where a job
involves some skills that can be assessed in this way and some that
cannot, the former tend to be over-emphasised. So ability to write
a short program can be easily assessed but ability to write a
2,000-statement program cannot: it would take too long."
To the list of formal methods of recruitment Bott adds nepotism
and cronyism. "Cronyism in particular should not be rejected as
unfair or ineffective," he said. "If one has worked with a person
before and seen they are effective in the role one is looking to
fill, then offering them the job is a low-risk way of filling
it."
Top of Bott's list for recruiting staff is a series of
traditional individual interviews with senior management and senior
technical staff. The drawback is that this approach makes it
difficult to show whether equal opportunities legislation has been
complied with. But, Bott said, "This can be a very reliable method
of selection, particularly if records are kept so you can go back
and see how effective each interviewer's judgement has been."
Professional Issues in IT covers topics including the nature of
a profession, financing a start-up company, intellectual property
rights and data protection. It costs £20, or £15 to BCS
members.
www.bcs.org/books/professionalissues