Where research questions come from
researchers may be told to research a specific question - ie from a research superviser or they may just look into something that they are passionate about
What is the first step after you have a research question
look at the existing literature
can tell you
1. If your question has been researched or not
2. level of demand for research for your question (Even if your question has already been researched this might not mean that its not worth looking into as research into a specific thing can indicate a demand for knowledge regarding a specific thing which could show that your research is wanted)
3. other experimenters thought processes on your question
4. how you might be able to approach the question
what are research reports
summarize novel (knew studies)
what are review articles
summarize existing information
can include theoretical reviews
these include proposals for novel ideas and frameworks (theoretical bc this framework/idea has not been tested yet)
metanalysis when results from multiple studies are quantified together and used to draw conclusions
what are scientific articles
include books, journals and thesis
What are good resources to search key words for if a subject has literature on it or not
JSTOR, google schoolar, and psychINFO
What makesit easier to get funding for a question
- if it represents a gap in research or controversial claim
- If the application is considered important
If a subject already has research on it what should you ask yourself
- Is the research convincing/replicatable
- Can other methods be used to test the question
- Would the results be the same under different circumstances
What is step 1
Finding a research question
researchers may be assigned a research question ie by a research supervisor
if they aren’t assigned a question they will likely select one based on personal interest
What is step 2
Reviewing the literature
Why do we review literature
1. Can tell us if our question has already been answered
2. to see the level of interest in our particular question (if there is lots of recent research on our question suggests that there is a lot of present interest in it and therefore that our research might be wanted)
3. Get ideas on how we could approach solving the question
4. To see others thought processes related to our question
What types of literature should we look at
scientific literature, includes journals, books and thesis
research report (like how a news reporter will talk about new events in the world a research report talks about new research)
review articles involve summaries of past writing (helpful when entering the feild makes sense you want to know what has already been established, likely reaffirmed and represents thte current most stable understanding before we move on to new ideas)
theoretical articles - proposals for new frameworks/methods (although it includes the new aspect all of their ideas would have to come reviewing the old and they have not yet tested the new so these ideas primarily exist within the realm of reviewing the old)
meta analysis - where the results of multiple studies are quantified together
thesis although we might not be able to access the thesis itself we might be able to access the literature it sights
putting key words into data? bases like google schoolar, psychINFO, and JSTOR is also a good place to start
What is the 3rd step
Describing research question
Is it feasible
can it be tested in a way that is ethical
can we get enough funding to do the types of tests we need
can we get enough relevant participants
Is there existing research on it and how is our perception of that research
is the research convincing/do we think it was done correctly
could other methods be used
could the conclusion differ in a different situation
Will it be of interest to the public (and therefore likely to receive funding)
Does it test a controversial claim or fill gaps in research
Does it have an application that is considered important
Step 4
Forming a hypothesis
theory:
a general principle about how a phenomenon functions over a range of situations
must be based on observations and be about a broad range of situations
may be tested or untested (but must be based on observations can not be all theoretical)
must define the phenomenon beyond face value
must include variables, functions, structures, process and organize processes
hypothesis - like hypothetical - about an outcome that has not happened yet. Whet we predict should happen in a specific test of the phenomen that we run if the theory about it holds true.
Theory needs to involve data that has been observed accross a wide range of situations
what are 2 ways to form hypothesis
1. have a general question about phenomena in the world and see if any existing theories provide possible explanations about it
2. look at existing theories and see if there is a tiatuion that the theory’s logic could be applied to but has not been yet
best hypothesises often compare two theories
Null hypothesis - null like no, no relationship exists between the variables
alternative hypothesis - there is a relationship between the variables (alternative to null - nullify like end - no more - null like no (more), no realtionship between the variables)
What makes a good hypothesis
should logically follow from a theory
in some cases there may not be a theory it can follow from then usually inductive reasoning is used (inductive reasoning involves seeing a rule exist in a specific situation and assuming that the rule must exist IN more situations (based on the assumption that a specific characteristic of the situation that is INcluded in other situations made this outcome occur). hypothesis that follow from a theory usually use deductive reasoning - start broad and then get specific.
should have 2 potential outcomes (this makes sense as we want it to be falsifiable and if it has only one outcome then it can not be falsifiable)
should be falsifiable
ideally positive (tests for the existence of a relationship - null vs alternative) rather then negative (negative like no- absense of a relationship) as it is easier to test for the existence of a relationship then the absense of a relationship
Step 5
Designing your study
variables
quantitative - numerical variables ie individuals exact iq scores
Categorical - none numerical labels ie iq under 100 and iq over 100 (does not tell us the actual number of the participants iq)
operational definition- how we define our variables
Sampling
can do random sampling-take a very large sample from the population - everyone has an equally likely chance of being picked - assume if you get enough people you will eventually adequately represent the population
Convenience sampling - sample a certain population that you conveniently have access to ie in our psych courses we have to sign up to do a certain amount of research participation this is convenience sampling bc we take from an easy pool (does not represent the whole pop)
might not want a representative sample bc might want to study a very specific feature that not all of the population has - ie altzhiemers
experimental approaches - we interfere with the variables - ie we set the independent variable to a certain level and then examine what the corresponding variable impact is in the depdneing variable (dependent variable like depends on bc we want to see how it’s expression depends on what we do to the independent variable)
non-experimental approach - does not mean its not scientific it just means that we directly do not interfere with the variables
confounding variables
ie we are trying to determine if the level of background noise present impacts performance on an iq test.
We might have in our participants some people who slept less before taking the test, people who didn’t eat before taking the test vs people who did, etc.
One way we could try to eliminate confounding variables is to give each participants a snack before taking the test (as as to at least ensure that none of them are starving before the test). We could also treat potential confounding variables as independent variables (the variables who we have at different levels for different people and compare the impacts) ie if we have large enough ammounts of people who got the same ammount of sleep we could make groups for certain ammounts who are in the same distraction or no distraction group and compare their performances and could only compare people who got the same ammount of sleep across groups. Important that if you decide to treat a confounding variable as an independent variable you plan to do so in advance
Lab vs feild research
Lab research have more control over confounding variables so have more internal validity however the real world usually includes confounding varabels that are not made to be consistent to this degree so it has lower external validity
Feild research - feild suggests outside, external. Have higher external validity bc there is lower interference (ie we just observe phenomena -making the setting more like how it usually is - in the real world usually we don’t have one force controlling multiple variables for a large group of people - might have many sources/some being controlled by chance/nature) lower internal validity - we can be less sure what caused the relationship that we saw INside our experiment
Way to solve the problem of lab vs feild research
ideally do lab research first- as this will tell us what confounding variables to look for and give us data for when these variables are controlled by one force to when they likely aren’t (in the feild)
correlational analysis
internal vs external validity
internal validity - liek how valid the results were within the experiment - how sure we can be that the changes or lack of in the dependent variable we saw were caused by the manipulation of the independent variable and not a confounding variable
external validity - how sure we can be that this relaitonship between the variables exists outside of (external to) the experiment
What is step 6
Statistical analysis
Descriptive statistics
range - highest and lowest value of our data (range does refer to this in outside speak - ie talking about the range of human emotion refering to what exists of it can be demonstrated by showing the highest and lowest extremes and assuming everything falls in the middle)
mode - mode in french means fashion (if something is in fashion it will be one of the most common things you see) - mode for statisitics simmilarily refers to the most common (most frequently repeated number in the data set)
mean - the average, can be highly skewed by outliers- ie if you are taking the mean for finances of people in a room and have lots of people who are poor and one billionaire you will geth that the average financial level of the people in the room is being a millionaire. Be suspicious if mean is used espeically since median is a better measure.
Median - you write all data points from lowest to highest - if there are an odd number of data points the median =s the middle value, if there are an even number of data points the median =s the average of the two middle values. Median will not be skwewed by outliers due to this.
Standard deviation - the average distance that each data point is away from the mean. If the standard deviation is high then the majoirty of the data points are far away from the mean suggesting that the mean was skewed by outliars and therefore is not a good representation of the majoirty of the data points/sample. Can reduce the impact of outliars by eliminating values from the data set that are a certain ammount of standard deviations away from the mean.
