Study Guides (238,439)
Canada (115,137)
SOAN 2120 (89)


16 Pages
Unlock Document

University of Guelph
Sociology and Anthropology
SOAN 2120
David Walters

Lecture 12 Testing (strongest impact when the pre test and post tests are close) history and maturation (Strongest impact when the pre test and post tests are further apart) history has its strongest effect when there is a pre test some people proceed if the pre tests are the same. But that is not necessarily mean you are safe, because people need to be the same on all possible variables that influence the dep var selection bias is an internal validity threat that is likely when you dont have random assignment Think of internal validity as : is the experiment internally valid. Or is there something about (internal) to the design that impacts the outcome. External validity: is the experimental valid in the real world. Similar to generalizability as I discussed it before midterm number one Main effect: when one variable has an effect on the dependent variable. The interaction effect is when the effect depends on another level (category) of the other variable. factorial designs: two or more independent variables in combination. main effects: present in single treatment designs interaction effects: produce an effect beyond that of each variable alone Ethnography comes from cultural anthropology. Ethno means people or folk, and graphy means describing something. Ethnography means describing a culture and understanding another way of life from the native point of view. Ethnography assumes that people make inferences - that is go beyond what is explicitly seen or said to what is meant or implied. What two types of designs does a Solomon Four Group Design combine?- combination of classical and a prexperimental design (two group post testing) 3*3 factorial design is. That's a bonus John Cena does not like cross-tabs, borderline cases or none of the above John Cena likes correlation and Anova Factorial designs use more than one independent variable. Maybe you'd look at the effect of gender, sex, and age on income Hypothesis Testing as a Decision Making Problem - There are four possible outcomes when we conduct a hypothesis test 1. If H0 is not correct and we reject it, we have made the correct decision 2. If the null H0 is correct and we accept it, we have made the correct decision 3. If H0 is correct and we reject it, we have made a Type I Error 4. If H0 is not correct and we accept it, we have made a Type II Error - Understand the relationship between alpha levels (ex. .05 and .01) and the propensity to make a Type I or Type II Error - Easier to reject the null at 95 than 99 because you don’t have to be as confident o Also easier to incorrectly reject it o He said think about this until review class kk homie Three Types of Analyses - Two variables classified as quantitative versus categorical o If both variables are quantitative, there is a correlation o Two categorical variables, there is a cross-tab o One quantitative, one categorical – analysis of variance - Categorical (qualitative variables) o Gender (1 male 2 female 3 other) region (4 west 3 ontario 2 quebec 1 east)  (1 strongly agree 2 neither agree or disagree 3 strongly disagree – NOT ORDINAL – Not quantitative because there are NOT 4 CATEGORIES – IT IS THEREFORE CATEGORICAL)  Ordinal level are treated as quantitative because they have at least 4 categories - Midterm Question: which technique would be used to examine the relationship between the following variables? o Gender and the variable for whether or not you think the PM is doing a good job?  One quantitative one categorical – cross tab Starts at Type Null is true and you turn it down and you reject it is a type one error If the alternative is null and you fail to reject the null= type 2 error We do not know if we made an error. One way there is a small chance we are making a mistake. Multiple Choice Questions If a researcher selects an alpha level of .05 instead of .01 She needs to be 95 percent confident as opposed to 99.9 percent a) She would potentially be less likely to make a type 1 error b) The chance of making a type 1 error would be unaffected c) She would potentially be more likely to make a type 1 error d) The chance of making a type 11 error would be unaffected Midterm 2: In addition to acdc, there will be abba Variable classification - Analyze the relationship between two variables a. Categorical – gender, marital status b. Quantitative – education - Other acceptable quantitative variables - Conceptually continuous (ordinal) likert scale – as long as it has 4 categories Lecture 14 Tuesday November 1 st Field Research - Terminology – refer to the same thing - Anthropology – most anthropologists use participant observation o Based on the concept of naturalism – observe people in their home/natural environment - No statistics or mathematics, just face-to-face social interactions with people in their home environment - Generates qualitative, descriptive data - Participant observation is used when other methods are not practical - Much less structured – social environment and interactions guide the study - Focus groups – do not confuse with participant observation – much smaller groups and you tap into their ideas. Still qualitative but not the same objective - Focus groups are 8-12 individuals where you survey them in a 90 minute period. You do this to tap into their opinions. They are in the private sector where they get your opinions about a topic - Field research is not focusing in one the private sector but rather acedmia to publish in a peer reviewed journal Ethnography - Describing a culture and understanding another way of life from your own perspective (interpretive school) - Displays of behaviour do not give meaning – instead someone tries to understand what is meant by the behaviour - The reality of the members (meant or implied) – what is implied by the behaviour is the most important Ethnomethodology encompasses the theory, the philosophy, and the method o Findings are the result of the method used as from the social life or environment - Social interaction is a process of reality construction - “The study of common-sense” – specialized and highly detailed analysis of micro- situations (transcripts, conversations, videotape). Analyze behaviour, mannerism, language - The results we get are contributed to the methods we get - Social interaction is a process of reality constriction - They are interested in common sense- specialized and highly detailed analysis of micro situations The Field Study - There are several steps – listed on pg 224 of book - Begin with topic/group you want to study, not a specific hypothesis o Pick a group you are passionate and excited about o Researchers can be involved or uninvolved in their study. You can also do it alone or with a team - Small scale social research - Pick a group, a subculture, or a social setting o Ex. A police department, a sports team, or a bar - Choose a group you can gain access to o Some groups have barriers – firefighters, police, construction o Some groups do not want to be studied – homeless, criminals, deviants Access to the Group - Speak to gatekeeper – owner of business, leader of team, etc o Negotiate your role – how involved will you be? o Must establish trust and confidence (rapport) o Rapport is critical: can often be challenging o Promises – gatekeeper may want to be first person to review, recommended to not give them creative control over the project Informants - People you will interview o Must be familiar with group, must be involved in group, must be available - Establish a key informant who you will have closest contact with - You are an outsider – may be difficult to maintain relationships and establish rapport - They are aware that they are always watching Know Your Role - Knowing your role is critical o You must know how to dress, behave, and act in the host environment o Avoid offending the group o Make the group feel as comfortable as possible - Deviant groups will often try to test you - Try to become an insider – allows more access to the group - Implications – may become emotionally attached to the group, objectivity becomes compromised Data - What do the data look like? - Observe and document EVERYTHING – physical characteristics, surroundings, smells - Not numerical values - The field notes are everything - Video, and audio, diagrams - If the group allows, you can record or take audio - Transcribing is the most time consuming part - Accuracies depend don what is observed, remembered, told by members – they may lie or deceive you o see page 153 table o see 144-147 - Interviewing - Follow-up interviews – unstructured, non-directed o Valuable for picking up on themes o Listening is important o To compare to quantitative interviews pg 242- interviewing vs quantitative interviews: quantitative interviews have a time limit and are heavily structured. Qualitative interviews do not do this. In quantitative there are pre set answers Issues with conducting field researchers - Researchers go into the field with subjective biases - Bias changes the structure of relationships - Researchers are typical people - The objective is to be value free and neutral – is this possible? - “tunnel vision” – if we have a theory prior to research, we look for answers to it rather than observing all information. We are going one way and one way only- we want to check our biases at the door - Consequence: our research becomes researcher dependent (our research becomes dependent on us rather than the group we are studying) - Subjective interpretations are based on a limited amount of information Final Points: 1) It is possible to draw conclusions from objective facts 2) People view the world differently as a result of social differences, values, personal history or even genetic predispositions Colour blindness example - People view the world differently as a result of their social differences - It is possible to draw different conclusions from objective facts - Different interpretations of behaviour can lead to different conclusions Summary - Objective of research is to provide a thorough account of social behaviour of a group of people or social environment - Success can be based on a debriefing – share your findings and conclusions with the group - Different interpretations of behaviour can lead to different conclusions - The question you ask, the way you participate, what you choose to others can influence the direction of your study. Thus, the participant can sometimes have as much influence on the findings as the subjects- this can be seen as a limitation to participant observation- methodologists embrace this as it is not seen as a limitation Quantitative methods are more objective than qualitative methods -Qualitative methods are subjective Missing Data – on midterm - Data can be missing for a variety of reasons o Data may not be applicable to them o They may not know the data to the answer o They may not want to respond to a sensitive question - Missing data can be problematic if a large proportion of respondents are missing - Key assumption – if data is missing, it is missing at random - In this class if there is missing data we delete it o List-wise deletion – if someone is missing information for any of the variables we wipe them out from the analysis Week 8 Lecture 15 Experimental Designs - Effect of an independent variable upon a dependent variable o Differences in outcomes between two groups - Equivalent scores o Average scores on the dependent variable o Scores on all other variables which are likely to impact the dependent variable - Differences: manipulation: observed at the end are attributable to the manipulation How do we establish the equivalence of two or more groups? o Researchers typically use a random number generator o People do not get to choose their own condition - Objective – equal probability of assignment - Even with random assignment, the two or more groups will not be exactly equal in all aspects o There may be slight differences - If the assignment is random, we can use the test of statistical significance to assess if any differences are JUST due to random processes - As the sample size increases, individual differences decrease - Without random assignment, it is possible the differences could be due to pre-existing differences between the groups o Known as assignment bias - We could compare pre-test scores (dependent variable measures) before the experiment begins - People must be equal in terms of independent variable, but also must be equal in every other variable that could influence dependent variable o This happens automatically with random assignment Even with RA the two groups will not be exactly equal in all respects If the assignment is random, we can use test of statistical significance (manipulation vs random variation) -As a sample size increases, individual differences decline (the law of large numbers) Experimental Designs -Without RA: findings may be attributed to pre existing differences btw two groups -This is called selection bias -We could compare pre test scores (dependent variable measures) before the experiment begins -Significantly differentEquivalent-> proceed as if random assignment had take place -If you randomly assign people to large sample size, they will be statistically significant with one another- if you don’t this is a problem Experimental Design Notation - Found on pg 181 R O1 X O2 R O3 O4 - X represents experimental variable or effect (the independent variable) - O represents a dependent variable observation- you might give the groups a survey (the o’s are numerical) - R represents random assignment – they did not get to pick weather or not they were going to be in the experimental group - One group gets a treatements and then they are to fill out another survey again - This experiment has a pre-test and a post-test o Ex. Group 1 is observed, given treatment, then observed again o Group 2 is observed, no treatment, observed again Experimental Designs (Many Groups) - Simulated driving video game to assess the impact of various substances on driving impairment - Group 1 – given substance a – 6 crashes - Group 2 – given substance b – 15 crashes - Group 3 – given substance c – 8 crashes - Group 4 – given water – 2 crashes - Experimental design would be as follows R X1 O1 R X2 O2 R X3 O3 R O4 *** In this example there is no pre-test. Does this invalidate the findings? *** Internal Validity - Whether the experimental condition (treatment) actually had an effect within the experiment itself 1. History o When come event other than the experimental manipulation affect
More Less

Related notes for SOAN 2120

Log In


Don't have an account?

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.