PSYC 412 Lecture Notes - Lecture 12: External Validity, Psychiatric Medication, Mental Disorder
14 May 2018
School
Department
Course
Professor
PSYC 412: Developmental Psychopathology
Jan 15th, 2018
Lecture 3:
•
Wrap-up from Lecture #2: Sampling
o Who am I interested in studying? What population am I interested in? Define
your sample
▪ Inclusion criteria: Who can be here? i.e. Im only interested in girls: so
one criteria would be must be a girl
▪ Exclusion criteria: Who do I not want to include in my sample?
o Example of profs study comparing social behavior of children with (a) anxiety
and/or depression, (b) behavioral difficulties, (c) both
▪ Inclusion criteria:
• Must be clinically referred; i.e. seeking services, its gotten bad
enough that theyre getting help
• Presence of anxiety and/or depression and/or behavioral
difficulties
• Between 8 and 14 years of age
• Living with parent or current guardian for at least 3 months
▪ Exclusion criteria
• Kids werent taking psychiatric medication
o You have to carefully consider this: the people in your study affects
conclusions you draw and limits what you can say about your findings
▪ Can change findings, interpretations
• For example, if you only recruit girls in your sample and find
association between puberty and depressive symptoms, you
cannot say in children theres an association because we dont
know about all children, just girls
o How will you get your sample?
o In 2010, chief of statistics Canada publicly resigned because of census policy
changes
▪ All Canadians complete short form census (survey you do every 5-7
years to get basic information about population of Canada, i.e. average
age, family size, etc.)
find more resources at oneclass.com
find more resources at oneclass.com
▪ 20% are required to fill out a longer version of the census survey with
50 questions with more detail like level of education, income, etc.
▪ In 2010 the government wanted to make the long form of the census
optional, until that point it was mandatory
• Mr. Sheikh quits because of this: saying this plan wont work if
you make it optional
• If its optional, the people who choose to complete it will be
different than those who dont: systematically excluding people
who we really care about, and the sample we end up with is no
longer representative of the entire population of Canada
▪ In 2011 the National Household Survey response rate average was
68.6% (it was 94% in mandatory long form)
• Participation rates dramatically dropped once it was made
optional
• This matters because information from census data informs
policies
o Key issue in sampling: we are interested because it represents a population
▪ Sample is group filling it out, supposed to represent population, which
is the group were really interested in
▪ Random selection: everyone in population of interest has equal
chance of being chosen
• Thats how long form of census works: everyone in Canada has
equal chance of being chosen to fill out the long form census,
and this is what makes us confident that results represent
population were interested in
▪ Sample of convenience
• Schools, clinics, etc. who are willing to participate in studies
• You have to be careful when talking about findings and who
you generalize your findings to
o How you get participants shapes who you get
▪ how shapes the who
• Example: camp fairs where summer camps try to recruit families
o Random selection is very expensive and difficult so most studies rely on
samples of convenience
find more resources at oneclass.com
find more resources at oneclass.com
o To whom do you want your results to generalize?
▪ Sample vs. population
▪ External validity: extent I would expect results I got to generalize to
different groups of people/external conditions
• To what extent would I get these results if I did this study with
different people?
o Experimental/Study validity
▪ Internal validity
▪ External validity
▪
Construct validity
•
PART TWO
o Questions answered by developmental psychopathologists
▪ Epidemiology
▪ Correlates of disorder
▪ Who develops problems and what happens to those who have them?
▪ Interventions
▪ Overarching theme: how do we establish causality?
o Epidemiology:
▪ How common is a given problem and who has this problem?
▪ Identify rates of a problem in a population
• New cases over a time period = the incidence of a disorder
• All cases during a time period = the prevalence of a disorder
▪ Sampling and measurement of disorder are
critical
for this type of
study
▪ Epidemiologists typically use random sampling (probability sampling or
population-based sampling)
• Everyone in population of interest has equal chance of being
selected to be in that sample
▪ Sample of convenience
• Sample who is readily available
▪ How common is a given problem?
• Sampling
o Need to be able to generalize to the population of
interest
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
50 questions with more detail like level of education, income, etc. In 2010 the government wanted to make the long form of the census optional, until that point it was mandatory: mr. sheikh quits because of this: saying this plan won(cid:284)t work if you make it optional. In 2011 the national household survey response rate average was. Journalist concluded that living at home in your early 20s key risk factor for violent behavior. It is possible that living at home causes violence (a -> b) A and b, like a substance abuse problem. I. e. school attendance in depressed and non- depressed teens: matched for gender, age, ses. If you recruit a sample of kids living in poverty when they are 5, measure their conduct symptoms, and return. If you remove the intervention and see it wasn(cid:284)t the cause of the behavior change then you shouldn(cid:284)t see any behavior go back up.
