Research Data Analysis

Views:
 
Category: Education
     
 

Presentation Description

No description available.

Comments

By: kiernanst (20 month(s) ago)

Very interseting advice, is it possible for you to send it to me at stephenmurray@yahoo.com?

Presentation Transcript

Data Analysis Workshop : 

8th Grade Project / The Walker School / Brian Surkan Data Analysis Workshop

Example: XVI Century Ship Voyage : 

Example: XVI Century Ship Voyage 50% of the women on the voyage from England to America died!

What’s Missing? : 

What’s Missing? How many women? (n: sample size) How many total persons? (p: sample population) Average statistics for such voyages Exceptional circumstances Woman already ill / pregnant Unusually bad weather Special type of ship

Example: Diabetes : 

Example: Diabetes Four out of five dentists recommend Crest toothpaste.

First: Do No Harm ! : 

First: Do No Harm ! Avoid conclusions beyond support of data E.g., Walker MS students not overweight ≠ Americans not overweight Focus on results, not intentions E.g., Minimum wage Complex Causes E.g., Obesity: too much food or not enough exercise? Interviews Treat as case studies, NOT universal truth

Validating Data : 

Validating Data Respondents (n): Those people who complete survey E.g., Did enough people complete the survey (often < 5%)? Filtering: Should some extreme results be discarded? E.g., Surveys where the person put 1 for every answer Scoring: How to tally/weight survey answers E.g., Group 4/5 & 5/5 (very good and excellent)? False Assumptions: Fair conclusions based on the data E.g., If 70% of Americans plan to vote Obama, it doesn’t mean that they all like him. Perhaps 30% just hate McCain.

Sources of Error : 

Sources of Error All measurements involve error Identifying possible errors helps interpret the data What may cause data from the sample population to NOT represent the target population? Unrepresentative sample (ask only your friends) Self-selected sample – respondents volunteered Ambiguous (unclear) questions Omitted (missing) questions / answers / information Data entry error (mistyped information) Non-standardized data field (e.g., type your age) Biased collection agent (e.g., fox counting the hens)

Fact v. Opinion : 

Fact v. Opinion Opinions/feelings do not alter facts E.g., I may feel rich even if I have no money. Ad Populum Fallacy What is popular is not always right/good/moral Where possible, ask for facts E.g., How old were you when you left Vietnam? Establish Context E.g., In what year did you leave Vietnam? E.g., What were your other options?

Causality v. Correlation : 

Causality v. Correlation Causality: A causes B E.g., He turned the key and the bus engine started VERY hard to prove Circumstantial evidence not enough Correlation: A and B generally happen together E.g., He jumped on the bus and then it drove away No clear causal relationship If you cannot prove A causes B: call them correlated

Common Data Analysis Mistakes : 

Common Data Analysis Mistakes Over-generalization Survey Walker MS and conclude for all 11-15 year olds Generalize for a large group from a few interviews Self-selection Bias Can you trust the people who chose to take the survey Did they ONLY do it for the Oreo/prize? Did they put down serious answers? Are they a representative sample of the target population?

Data Analysis Workshop : 

8th Grade Project / The Walker School / Brian Surkan Data Analysis Workshop