detailed lecture notes...missed it? no prob!

Announcements: [email protected] Brain Day Association of Uoft BDUT PSYCH 100 LECTURE 2: Research Methods Not a huge date/name guy for the exam-hint More interested in whether you understand the concepts Need to know wilhelm wundt for instance Characteristics of Good Psychological Research: When studying human psychology we have to be more careful than studying other things ethically etc STUFF WE NEED for good psychological research... 1. THEORETICAL FRAMEWORK: Is a systematic way of organizing and explaining observations We need a Hypothesis- which flows from the theory or from an important question Theories are never done, we are looking for the exceptions that might not be found until years down the road We might prefer the simpler theory if possible but exceptions are important 2. STANDARDIZED PROCEDURE: Means-procedure=same for all subjects except where variation introduced to death hypothesis The differences must not be about how we ran the study The results are a result of our intentional manipulation Because of that we go to great lengths to ensure everything is as standard as possible We want to counter balance Avoid confounds WE dont rely on any one study as being confirmative, the be all and end all We dont want to rely on any one study as something we would put full faith in We need studies to be replicated Preferably by other researches or in other parts of the world If we dont, we have to be more critical or skeptical Standardize procedure in ways that others can do it in as close to the same way 3. GENERALIZABILITY: The sample is representative of population Procedure is sensible and relevant to to world outside the lab Cant survey EVERYBODY so we take a representative sample Procedure that is sensible and relevant to circumstances outside the lab IF you use a biased sample you have to be concerned whether you can generalize your sample to a larger group We also want a procedure that has ecological validity-measuring something that is real www.notesolution.com How much sense does it make? Is it realistic Sensible, relevant outside of lab There are things below the level of conscious awareness Subliminal perception might have higher rates in a lab than in a store 4. OBJECTIVE MEASUREMENTS: Reliable (produce consistent results) Measures that are valid (that assess the dimensions they purport to assess Need test/questionaries that measure what we think they are measuring Are we measuring something well defined? We want to make sure they are assessing what we believe they are assessing PSYCHOMETRICS: --form of statistic--ensuring measurements reliable, valid etc. lets say we are trying to measure anxiety-how will we measure it? PSYCHOLOGICAL RESEARCH: Naturalistic observation: observe ppl. In their natural setting without altering their setting Man hiding in bush with camera filming lions As removed as possible Have as little influence as possible on the groups Great way to come up with ideas for other studies BUT we cant infer causality Can be with anyone with anywhere Ie: children on the playground forming peer groups, you watch them Surveys Can be manipulated and used in a great way Election surveys- not ALL polling is perfect-there is a margin of error Have to be wary of surveys-you can create a survey that biases the answers you get--suggests the answers in the question Who is the source? Who paid for the survey? What are their stakes? We must be skeptical Case study: A single person Person with a rare disease; if they get better do we know that are treatment worked or could they have gotten naturally better You have evidence that people have gotten better But through empirical evidence found to be random Anecdotal evidence can be very powerful--it helped you or someone you know and it makes it seem real--those types of conclusion are very dangerous The affect is placebo-believing something works is enough to make it work for some people Unless we have multiple data to look at and compare we cant be sure www.notesolution.com Correlational A measure of association between two variables Self esteem and grade point average are moderately positively correlated Its not a perfect relationship but it can be measured -1 or +1 are perfect correlations 0 = no correlation Height and weight very strongly correlated Strength of a correlation: Perfect=1.0 Strong 0.6 to 0.8 Moderately strong 0.3 0.5 No correlation 0 Most human behaviour we discuss in moderately strong, more than that is unusual Cannot tell us about causation We just have to be careful about the data we get and only apply it the way we should What does causal mean here? Correlational research: establishes whether there is a relationship between two or more variable Cannot infer causality 1. Directionality problem ie: which variable affects which 2. Potential for a third variable (confound) if it interferes with the relationship and we arent aware of it From a correlational study we dont know if have a a high GPA causes high self esteem OR vise versa In this case there is a third more important variable ie: high IQ-more likely to have high self esteem and GPA--IQ affects both other these factors than either has with each other When we see correlations as if they are causal thats a huge problem Media is very bad at presenting correlation studies as if they are causal Good to do an experimental study following a correlation study Examples: Smoking is correlated with lung cancer For a long time you couldnt say that cigarettess CAUSE cancer Couldnt do human experiments because of ethics Cant do things we think will cause harm For decades ppl. Said you cant prove it After enough correlational studies and experiments with animals now health canada says smoking causes cancer We say now this is a correlation that is indeed causal 2. High grades are correlated with successful careers www.notesolution.com No directionality problem Also influences by other factors Third variable confound that makes this relationship of much less significance ie: IQ 3. Eating frosted flakes is correlated with lower cancer rates Age is the confound -ppl. Who eat frosted flakes are mostly kids who have the lowest rates of cancer 4. In Taiwan birth control is correlated with number of household appliances Education--if you have a higher education level you are more likely to use birth control If you make more money you are more likely to buy more appliances There are ways to infer and test coming out of a correlation study Correlations are important but they are different that experimental Media presents studies that are correlational as if they are experimental Experimental This is so could because we can infer causation Considered the most powerful tool for determining causal relationship RANDOM ASSIGNMENT: ensures that every participant has and equal chance of being assigned to any of the conditions Number our of a hat, flip coin etc This minimizes the chance that a pre-existing difference between groups is the casuse of the experimental effect If people self select if could effect results You need to randomly assign to ensure any preexisting differences will be random OUR CLASS EXPERIMENT (what he TOLD us it was about: Theory: the more education in specializes ares leads to better performance in those areas Hypothesis: Independent variable: years of study and area of major Dependent variable: number of words generated on a word generation task and verbal speech ^^^^ FORGET THE ABOVE The hypothesis and all that was a total lie. Debrief: what is was actually about What is the purpose of the study: To see the effects of anxiety of performance Experimental hypothesis: Anxiety will decrease performance Those in the anxiety condition will generate less words than those in the control condition www.notesolution.com