|Equivalence of instruments
Alternate forms or split-half
Instrument is given in different forms such as written
and oral or in-person and over the phone. This also
includes instruments in in different languages.
The two tests are given to participants
at the same time and the scores should be similar. This
is usually done when researchers are testing to see
if a tried and true instrument can be used in a different
order (like when you took the GRE) or a shorter version
(such as the 24 item acculturation instrument that is
now down to four items!)
The different forms of instrument administration need
to be tested to assure equivalency.
|Decreasing Measurement error
Transitory personal factors
Different types of data collection methods
Inconsistencies in the data collection instruments:
For example: On a Likert scale: are the possible responses
equal in both the spacing and conceptually.
On a visual analog scale: the markings should be equal
distance from each other and each visual analog line
should be the same length
There should be consistency
as to the data collection if it seems relevant. Examples
include researcher characteristics, the physical setting,
the weather or the time of day.
Temporary conditions such as pain, anxiety or mood
need to be assessed
The participant may respond in a certain way due to
social desirability, boredom, or decreased attention
span. For example when administering a questionnaire
to teens they could respond to the age of the data collector
or it could take longer than 20 minutes to take and
they can start circling any answer just to get done.
Another technique is to put few questions on a page
and have more pages. At the expense of trees, it keep
participants engaged longer because it "feels"
like they are making faster progress.
Another example would be always having the "good"
answer in the same spot on a questionnaire. For example,
on the questionnaire. I used for pregnant teens, mixed
in with questions about the good things about school
(hang out with friends, get a better education, be able
to provide for my baby) there would be a bad thing about
school (such as having lots of homework). This technique
discourages just circling all the "goods"
without really reading the questions. It also let me
know if it was more likely they just circled all the
Data are collected in different ways such as: Self-report,
an oral questionnaire, a written questionnaire, on the
phone, on the computer, read by the data collector,
read by the participant, open ended questions, and fixed
response questions. Each one will have reliability threats
and strengths depending on the characteristics of the
Measure the distance between possible answers to assure
they are equal. Make sure the phrase meanings are also
consistent. For example on a Likert scale: It would
read: Very unlikely Kind of unlikely Unlikely Likely
Kind of likely and Very likely.