3 the “five step” process Developmental TheoryAsk a (developmental) research question.Develop a hypothesis.Construct a methodology to test your hypothesis.Draw a conclusion.Share your findings.
4 !! NO !! research questions Does god exist? Is this a good research question?!! NO !!Good research questions must involve something that can be empirically defined and measured.
5 i<clicker Which of these is the best research question? A) Alcoholic parents tend to neglect their children more than non-alcoholic parents.B) What factors influence high school dating?C) Does the frequency and graphic quality of violence in video games influence levels of anti-social thinking in adolescent males?D) How much longer until class is over?E) B & C are equally good.
6 definition and measurement Operational definitionthe description of the variable of interest in measurable terms.So, how might we operationally define?AggressionHappinessMeasurementDevice used to detect the events/phenomena to which the operational definition refers.With this in mind, take a moment to generate a research question of your own.
7 types of measurement Subjective measures Introspective reportsSurvey/QuestionnaireObjective measuresStandardized testingNaturalistic observationPhysiological measuresfMRI, galvanic skin responseLevels of hormones, neurotransmittersAre physiological measures subjective or objective measures?A) subjectiveB) objectiveC) both
8 developing hypotheses Developing a hypothesis:Consider your research question.What do you think you’ll find?Why?That is, what is your (theoretical/conceptual/empirical) justification for you hypothesis?
9 testing hypotheses How are you going to test your hypothesis? Non-experimental DesignObservation of variables of interestExperimental DesignManipulation of variables of interest
10 non-experimental designs Systematic ObservationNaturalisticStructuredSelf-report SurveyClinicalQualitativeCase study – in-depth study of individualEthnography – in-depth study of cultureWhat are the (dis)advantages of a naturalistic vs. structured observation?Why choose a survey over an observation?Ask Yourself! What design is best for study non-normative development?
11 tracking development Cross sectional Benefits – cheap way to capture change over timeProblems – cohort effect and other group differencesLongitudinalBenefits – confidence that change being captured is genuine changeProblems – reduction of sample size and learning effectCross-sequential
12 cross-sequentialTime 1 Time 22nd4th6th4th6th8th
13 cross-sequential Time 1 Time 2 2nd 4th 6th 4th 6th 8th Ask Yourself! How do cross-sequential designs reveal any cohort effects?2nd4th6th4th6th8th
14 population and sampleYou are asking a question about behavior in a given populationIt is difficult (if not impossible) to ever study an entire population – so what do we study instead?teenagers
15 population and sampleYou are asking a question about behavior in a given populationIt is difficult (if not impossible) to ever study an entire population – so what do we study instead?a sample.teenagers
16 population and sampleHow do we make sure that we can accurately generalize from a sample to a population?We choose a representative sample.controlled samplingrandom sampling
17 i<clicker Which makes for a good sample? A) when it is chosen to match the demographics of the population as closely as possibleB) when it is randomly selected from the populationC) both.
18 relationships between variables What is a correlation?Relationship between two variablesA is related to BPositive relationship: A+/B+, A-/B-Negative relationship: A+/B-, A-/B+
19 relationships between variables As a person gets angrier, they also get more violent.A) PositiveB) NegativePositive. As anger increases, violence increases.As a person gets older, they start to remember fewer vocabulary words.A) PositiveB) NegativeNegative. As age increases, vocabulary memory decreases.As calorie consumption drops, people have less energy.A) PositiveB) NegativePositive. As calorie consumption decreases, energy levels decrease (movement is happening in the same direction).
20 relationships between variables Ultimately, we are typically interested in whether or not one variable causes another.T/F: All variables that are causally related are correlated.T/F: All variables that are correlated are causally related.
21 Ask Yourself!A researcher compares older adults with chronic heart disease to those with no major health problems and finds that the first group scores lower on mental tests. Can the researcher conclude that heart disease causes a decline in intellectual functioning in late adulthood?A) yesB) no
22 correlation vs. causation Two variables are correlatedX YThree possible relationshipsX causes YY causes XZ causes both X and Ywith correlation, we cannot know which it is.22
25 experimental designTo establish causation, we must conduct an experiment.Experimentation requires manipulation.A BA is the independent variable -- manipulatede.g., amount of television violence watchedB is the dependent variable -- measurede.g., amount of aggressive behavior exhibited25
36 normal distributionMany things tend to be normally distributed in a given population.So, we should expect most people to fall somewhere close to the middle, with the extreme cases being less frequent.IQ is normally distributed.mean
37 Income is one thing that is not normally distributed. A) True B) False Can you think of others?
38 design considerations ValidityBeing able to draw accurate inferences (conclusions) about what you are studying from your measurementsInvalid in definitionExamples?Invalid in detection (measurement)
39 i<clickerIf I asked college students and mature adults to rate on a scale how much fun they have ever day and I found college students rated their level of fun much higher, could I then draw the conclusion that college students are happier than mature adults?A) yesB) noWhy?Issue of invalid definition.If I was interested in studying freshman college students’ general anxiety levels and so I polled all of my Intro students using a standard anxiety scale at the beginning of class right before they took an exam, could I draw conclusions about general levels of anxiety from those measurements?A) yesB) noWhy?Issue of invalid detection.
40 other considerations Internal validity External validity Study was designed so that you were able to draw accurate inferences about causal relation between independent and dependent variables.External validityStudy was designed so that your independent and dependent variables are defined in natural/realistic way.You can have internal validity but not have external validity – why?
41 design considerations ReliabilityThe tendency for measurement to produce the same results when used in the same way (or under the same conditions).Type 1 error (false positive)You want a measurement that is stable enough that it won’t detect changes in your variable when changes haven’t actually occurred.PowerThe tendency for measurement to produce different results when used in different ways (or under different conditions).Type 2 error (false negative)You want a measurement that is sensitive enough to detect changes in your variable when changes actually occur…
42 i<clickerYou want a measurement that is sensitive enough to detect changes in your variable when changes actually occur…This is an issue of:A) powerB) reliabilityYou also want a measurement that is stable enough that it won’t detect changes in your variable when changes haven’t actually occurred.This is an issue of:A) powerB) reliability
43 other considerations Biases in observation Double-blind experiments Participant biasesDemand characteristicsObserver biasesConfirmation biasDouble-blind experiments
44 other considerations Ethical practices Informed consentDebriefingSpecial considerations for childrenAre there things we shouldn’t study?
45 final steps Drawing conclusions Sharing your findings What kinds of conclusions can you draw?Can you generalize to a population?How broad of a population?LimitationsSharing your findingsConference presentationsPublications